implemented last remaining codegen for word-byte division and remainders.

Add support for built-in functions to Vim syntax file (and other fixes)

.gitignore

@@ -1,20 +1,20 @@
-**/.idea/
-/build/
-/dist/
-/output/
+.idea/workspace.xml
+.idea/discord.xml
+build/
+dist/
+output/
 .*cache/
 *.directory
 *.prg
-*.asm
 *.bin
 *.labels.txt
 *.vm.txt
 *.vice-mon-list
 docs/build
 out/
-**/*.interp
-**/*.tokens
-
+parser/**/*.interp
+parser/**/*.tokens
+parser/**/*.java
 *.py[cod]
 *.egg
 *.egg-info
@@ -25,10 +25,9 @@ __pycache__/
 parser.out
 parsetab.py
 .pytest_cache/
-compiler/src/prog8_kotlin.jar
-compiler/src/compiled_java
 .attach_pid*
 
 .gradle
-build/
 /prog8compiler.jar
+sd*.img
+

.idea/.gitignore

@@ -0,0 +1,2 @@
+# Default ignored files
+/shelf/

.idea/dictionaries/irmen.xml

@@ -0,0 +1,3 @@
+<component name="ProjectDictionaryState">
+  <dictionary name="irmen" />
+</component>

.idea/inspectionProfiles/Project_Default.xml

@@ -0,0 +1,17 @@
+<component name="InspectionProjectProfileManager">
+  <profile version="1.0">
+    <option name="myName" value="Project Default" />
+    <inspection_tool class="DuplicatedCode" enabled="true" level="WEAK WARNING" enabled_by_default="true">
+      <Languages>
+        <language minSize="70" name="Kotlin" />
+        <language isEnabled="false" name="Groovy" />
+      </Languages>
+    </inspection_tool>
+    <inspection_tool class="PyInterpreterInspection" enabled="false" level="WARNING" enabled_by_default="false" />
+    <inspection_tool class="SpellCheckingInspection" enabled="true" level="TYPO" enabled_by_default="true">
+      <option name="processCode" value="false" />
+      <option name="processLiterals" value="true" />
+      <option name="processComments" value="false" />
+    </inspection_tool>
+  </profile>
+</component>

.idea/kotlinc.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Kotlin2JvmCompilerArguments">
+    <option name="jvmTarget" value="11" />
+  </component>
+</project>

.idea/libraries/KotlinJavaRuntime.xml

@@ -0,0 +1,19 @@
+<component name="libraryTable">
+  <library name="KotlinJavaRuntime">
+    <CLASSES>
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-reflect.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-test.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib-jdk7.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib-jdk8.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES>
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib-sources.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-reflect-sources.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-test-sources.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib-jdk7-sources.jar!/" />
+      <root url="jar://$KOTLIN_BUNDLED$/lib/kotlin-stdlib-jdk8-sources.jar!/" />
+    </SOURCES>
+  </library>
+</component>

.idea/libraries/antlr_4_9_complete.xml

@@ -0,0 +1,9 @@
+<component name="libraryTable">
+  <library name="antlr-4.9-complete">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-4.9-complete.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/antlr_runtime_4_9.xml

@@ -0,0 +1,9 @@
+<component name="libraryTable">
+  <library name="antlr-runtime-4.9">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-runtime-4.9.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/dbus_java_3_2_4.xml

@@ -0,0 +1,9 @@
+<component name="libraryTable">
+  <library name="dbus-java-3.2.4">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/dbusCompilerService/lib/dbus-java-3.2.4.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/javax_json_api_1_1_4.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="javax.json-api-1.1.4">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/javax.json-api-1.1.4.jar!/" />
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/javax.json-1.1.4.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/kotlinx_cli_jvm.xml

@@ -0,0 +1,9 @@
+<component name="libraryTable">
+  <library name="kotlinx-cli-jvm">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/compiler/lib/kotlinx-cli-jvm-0.3.1.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/slf4j_api_1_7_30.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="slf4j-api-1.7.30">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/slf4j-api-1.7.30.jar!/" />
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/slf4j-simple-1.7.30.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/takes_http.xml

@@ -0,0 +1,12 @@
+<component name="libraryTable">
+  <library name="takes-http">
+    <CLASSES>
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/cactoos-0.42.jar!/" />
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/commons-lang3-3.7.jar!/" />
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/commons-text-1.4.jar!/" />
+      <root url="jar://$PROJECT_DIR$/httpCompilerService/lib/takes-1.19.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/unittest_libs.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="unittest-libs">
+    <CLASSES>
+      <root url="file://$PROJECT_DIR$/compiler/lib" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+    <jarDirectory url="file://$PROJECT_DIR$/compiler/lib" recursive="false" />
+  </library>
+</component>

.idea/markdown-navigator-enh.xml

@@ -0,0 +1,29 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="MarkdownEnhProjectSettings">
+    <AnnotatorSettings targetHasSpaces="true" linkCaseMismatch="true" wikiCaseMismatch="true" wikiLinkHasDashes="true" notUnderWikiHome="true" targetNotWikiPageExt="true" notUnderSourceWikiHome="true" targetNameHasAnchor="true" targetPathHasAnchor="true" wikiLinkHasSlash="true" wikiLinkHasSubdir="true" wikiLinkHasOnlyAnchor="true" linkTargetsWikiHasExt="true" linkTargetsWikiHasBadExt="true" notUnderSameRepo="true" targetNotUnderVcs="false" linkNeedsExt="true" linkHasBadExt="true" linkTargetNeedsExt="true" linkTargetHasBadExt="true" wikiLinkNotInWiki="true" imageTargetNotInRaw="true" repoRelativeAcrossVcsRoots="true" multipleWikiTargetsMatch="true" unresolvedLinkReference="true" linkIsIgnored="true" anchorIsIgnored="true" anchorIsUnresolved="true" anchorLineReferenceIsUnresolved="true" anchorLineReferenceFormat="true" anchorHasDuplicates="true" abbreviationDuplicates="true" abbreviationNotUsed="true" attributeIdDuplicateDefinition="true" attributeIdNotUsed="true" footnoteDuplicateDefinition="true" footnoteUnresolved="true" footnoteDuplicates="true" footnoteNotUsed="true" macroDuplicateDefinition="true" macroUnresolved="true" macroDuplicates="true" macroNotUsed="true" referenceDuplicateDefinition="true" referenceUnresolved="true" referenceDuplicates="true" referenceNotUsed="true" referenceUnresolvedNumericId="true" enumRefDuplicateDefinition="true" enumRefUnresolved="true" enumRefDuplicates="true" enumRefNotUsed="true" enumRefLinkUnresolved="true" enumRefLinkDuplicates="true" simTocUpdateNeeded="true" simTocTitleSpaceNeeded="true" />
+    <HtmlExportSettings updateOnSave="false" parentDir="" targetDir="" cssDir="css" scriptDir="js" plainHtml="false" imageDir="" copyLinkedImages="false" imagePathType="0" targetPathType="2" targetExt="" useTargetExt="false" noCssNoScripts="false" useElementStyleAttribute="false" linkToExportedHtml="true" exportOnSettingsChange="true" regenerateOnProjectOpen="false" linkFormatType="HTTP_ABSOLUTE" />
+    <LinkMapSettings>
+      <textMaps />
+    </LinkMapSettings>
+  </component>
+  <component name="MarkdownNavigatorHistory">
+    <PasteImageHistory checkeredTransparentBackground="false" filename="image" directory="" onPasteImageTargetRef="3" onPasteLinkText="0" onPasteImageElement="1" onPasteLinkElement="1" onPasteReferenceElement="2" cornerRadius="20" borderColor="0" transparentColor="16777215" borderWidth="1" trimTop="0" trimBottom="0" trimLeft="0" trimRight="0" transparent="false" roundCorners="false" showPreview="true" bordered="false" scaled="false" cropped="false" hideInapplicableOperations="false" preserveLinkFormat="false" scale="50" scalingInterpolation="1" transparentTolerance="0" saveAsDefaultOnOK="false" linkFormat="0" addHighlights="false" showHighlightCoordinates="true" showHighlights="false" mouseSelectionAddsHighlight="false" outerFilled="false" outerFillColor="0" outerFillTransparent="true" outerFillAlpha="30">
+      <highlightList />
+      <directories />
+      <filenames />
+    </PasteImageHistory>
+    <CopyImageHistory checkeredTransparentBackground="false" filename="image" directory="" onPasteImageTargetRef="3" onPasteLinkText="0" onPasteImageElement="1" onPasteLinkElement="1" onPasteReferenceElement="2" cornerRadius="20" borderColor="0" transparentColor="16777215" borderWidth="1" trimTop="0" trimBottom="0" trimLeft="0" trimRight="0" transparent="false" roundCorners="false" showPreview="true" bordered="false" scaled="false" cropped="false" hideInapplicableOperations="false" preserveLinkFormat="false" scale="50" scalingInterpolation="1" transparentTolerance="0" saveAsDefaultOnOK="false" linkFormat="0" addHighlights="false" showHighlightCoordinates="true" showHighlights="false" mouseSelectionAddsHighlight="false" outerFilled="false" outerFillColor="0" outerFillTransparent="true" outerFillAlpha="30">
+      <highlightList />
+      <directories />
+      <filenames />
+    </CopyImageHistory>
+    <PasteLinkHistory onPasteImageTargetRef="3" onPasteTargetRef="1" onPasteLinkText="0" onPasteImageElement="1" onPasteLinkElement="1" onPasteWikiElement="2" onPasteReferenceElement="2" hideInapplicableOperations="false" preserveLinkFormat="false" useHeadingForLinkText="false" linkFormat="0" saveAsDefaultOnOK="false" />
+    <TableToJsonHistory>
+      <entries />
+    </TableToJsonHistory>
+    <TableSortHistory>
+      <entries />
+    </TableSortHistory>
+  </component>
+</project>

.idea/markdown-navigator.xml

@@ -0,0 +1,57 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="FlexmarkProjectSettings">
+    <FlexmarkHtmlSettings flexmarkSpecExampleRendering="0" flexmarkSpecExampleRenderHtml="false">
+      <flexmarkSectionLanguages>
+        <option name="1" value="Markdown" />
+        <option name="2" value="HTML" />
+        <option name="3" value="flexmark-ast:1" />
+      </flexmarkSectionLanguages>
+    </FlexmarkHtmlSettings>
+  </component>
+  <component name="MarkdownProjectSettings">
+    <PreviewSettings splitEditorLayout="SPLIT" splitEditorPreview="PREVIEW" useGrayscaleRendering="false" zoomFactor="1.0" maxImageWidth="0" synchronizePreviewPosition="true" highlightPreviewType="LINE" highlightFadeOut="5" highlightOnTyping="true" synchronizeSourcePosition="true" verticallyAlignSourceAndPreviewSyncPosition="true" showSearchHighlightsInPreview="true" showSelectionInPreview="true" lastLayoutSetsDefault="false">
+      <PanelProvider>
+        <provider providerId="com.vladsch.md.nav.editor.javafx.html.panel" providerName="JavaFX WebView" />
+      </PanelProvider>
+    </PreviewSettings>
+    <ParserSettings gitHubSyntaxChange="false" correctedInvalidSettings="false" emojiShortcuts="1" emojiImages="0">
+      <PegdownExtensions>
+        <option name="ANCHORLINKS" value="true" />
+        <option name="ATXHEADERSPACE" value="true" />
+        <option name="FENCED_CODE_BLOCKS" value="true" />
+        <option name="INTELLIJ_DUMMY_IDENTIFIER" value="true" />
+        <option name="RELAXEDHRULES" value="true" />
+        <option name="STRIKETHROUGH" value="true" />
+        <option name="TABLES" value="true" />
+        <option name="TASKLISTITEMS" value="true" />
+      </PegdownExtensions>
+      <ParserOptions>
+        <option name="COMMONMARK_LISTS" value="true" />
+        <option name="EMOJI_SHORTCUTS" value="true" />
+        <option name="GFM_TABLE_RENDERING" value="true" />
+        <option name="PRODUCTION_SPEC_PARSER" value="true" />
+        <option name="SIM_TOC_BLANK_LINE_SPACER" value="true" />
+      </ParserOptions>
+    </ParserSettings>
+    <HtmlSettings headerTopEnabled="false" headerBottomEnabled="false" bodyTopEnabled="false" bodyBottomEnabled="false" addPageHeader="false" imageUriSerials="false" addDocTypeHtml="true" noParaTags="false" plantUmlConversion="0">
+      <GeneratorProvider>
+        <provider providerId="com.vladsch.md.nav.editor.javafx.html.generator" providerName="JavaFx HTML Generator" />
+      </GeneratorProvider>
+      <headerTop />
+      <headerBottom />
+      <bodyTop />
+      <bodyBottom />
+    </HtmlSettings>
+    <CssSettings previewScheme="UI_SCHEME" cssUri="" isCssUriEnabled="false" isCssUriSerial="true" isCssTextEnabled="false" isDynamicPageWidth="true">
+      <StylesheetProvider>
+        <provider providerId="com.vladsch.md.nav.editor.javafx.html.css" providerName="Default JavaFx Stylesheet" />
+      </StylesheetProvider>
+      <ScriptProviders>
+        <provider providerId="com.vladsch.md.nav.editor.hljs.html.script" providerName="HighlightJS Script" />
+      </ScriptProviders>
+      <cssText />
+      <cssUriHistory />
+    </CssSettings>
+  </component>
+</project>

.idea/misc.xml

@@ -0,0 +1,22 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ANTLRGenerationPreferences">
+    <option name="perGrammarGenerationSettings">
+      <list>
+        <PerGrammarGenerationSettings>
+          <option name="fileName" value="$PROJECT_DIR$/parser/antlr/prog8.g4" />
+          <option name="autoGen" value="true" />
+          <option name="outputDir" value="$PROJECT_DIR$/parser/src/prog8/parser" />
+          <option name="libDir" value="" />
+          <option name="encoding" value="" />
+          <option name="pkg" value="" />
+          <option name="language" value="" />
+          <option name="generateListener" value="false" />
+        </PerGrammarGenerationSettings>
+      </list>
+    </option>
+  </component>
+  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" default="false" project-jdk-name="Kotlin SDK" project-jdk-type="KotlinSDK">
+    <output url="file://$PROJECT_DIR$/out" />
+  </component>
+</project>

.idea/modules.xml

@@ -0,0 +1,14 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ProjectModuleManager">
+    <modules>
+      <module fileurl="file://$PROJECT_DIR$/compiler/compiler.iml" filepath="$PROJECT_DIR$/compiler/compiler.iml" />
+      <module fileurl="file://$PROJECT_DIR$/compilerAst/compilerAst.iml" filepath="$PROJECT_DIR$/compilerAst/compilerAst.iml" />
+      <module fileurl="file://$PROJECT_DIR$/dbusCompilerService/dbusCompilerService.iml" filepath="$PROJECT_DIR$/dbusCompilerService/dbusCompilerService.iml" />
+      <module fileurl="file://$PROJECT_DIR$/docs/docs.iml" filepath="$PROJECT_DIR$/docs/docs.iml" />
+      <module fileurl="file://$PROJECT_DIR$/examples/examples.iml" filepath="$PROJECT_DIR$/examples/examples.iml" />
+      <module fileurl="file://$PROJECT_DIR$/httpCompilerService/httpCompilerService.iml" filepath="$PROJECT_DIR$/httpCompilerService/httpCompilerService.iml" />
+      <module fileurl="file://$PROJECT_DIR$/parser/parser.iml" filepath="$PROJECT_DIR$/parser/parser.iml" />
+    </modules>
+  </component>
+</project>

.idea/uiDesigner.xml

@@ -0,0 +1,124 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="Palette2">
+    <group name="Swing">
+      <item class="com.intellij.uiDesigner.HSpacer" tooltip-text="Horizontal Spacer" icon="/com/intellij/uiDesigner/icons/hspacer.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="1" hsize-policy="6" anchor="0" fill="1" />
+      </item>
+      <item class="com.intellij.uiDesigner.VSpacer" tooltip-text="Vertical Spacer" icon="/com/intellij/uiDesigner/icons/vspacer.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="1" anchor="0" fill="2" />
+      </item>
+      <item class="javax.swing.JPanel" icon="/com/intellij/uiDesigner/icons/panel.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="3" hsize-policy="3" anchor="0" fill="3" />
+      </item>
+      <item class="javax.swing.JScrollPane" icon="/com/intellij/uiDesigner/icons/scrollPane.png" removable="false" auto-create-binding="false" can-attach-label="true">
+        <default-constraints vsize-policy="7" hsize-policy="7" anchor="0" fill="3" />
+      </item>
+      <item class="javax.swing.JButton" icon="/com/intellij/uiDesigner/icons/button.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="3" anchor="0" fill="1" />
+        <initial-values>
+          <property name="text" value="Button" />
+        </initial-values>
+      </item>
+      <item class="javax.swing.JRadioButton" icon="/com/intellij/uiDesigner/icons/radioButton.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="3" anchor="8" fill="0" />
+        <initial-values>
+          <property name="text" value="RadioButton" />
+        </initial-values>
+      </item>
+      <item class="javax.swing.JCheckBox" icon="/com/intellij/uiDesigner/icons/checkBox.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="3" anchor="8" fill="0" />
+        <initial-values>
+          <property name="text" value="CheckBox" />
+        </initial-values>
+      </item>
+      <item class="javax.swing.JLabel" icon="/com/intellij/uiDesigner/icons/label.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="0" anchor="8" fill="0" />
+        <initial-values>
+          <property name="text" value="Label" />
+        </initial-values>
+      </item>
+      <item class="javax.swing.JTextField" icon="/com/intellij/uiDesigner/icons/textField.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="8" fill="1">
+          <preferred-size width="150" height="-1" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JPasswordField" icon="/com/intellij/uiDesigner/icons/passwordField.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="8" fill="1">
+          <preferred-size width="150" height="-1" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JFormattedTextField" icon="/com/intellij/uiDesigner/icons/formattedTextField.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="8" fill="1">
+          <preferred-size width="150" height="-1" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JTextArea" icon="/com/intellij/uiDesigner/icons/textArea.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JTextPane" icon="/com/intellij/uiDesigner/icons/textPane.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JEditorPane" icon="/com/intellij/uiDesigner/icons/editorPane.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JComboBox" icon="/com/intellij/uiDesigner/icons/comboBox.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="0" hsize-policy="2" anchor="8" fill="1" />
+      </item>
+      <item class="javax.swing.JTable" icon="/com/intellij/uiDesigner/icons/table.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JList" icon="/com/intellij/uiDesigner/icons/list.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="2" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JTree" icon="/com/intellij/uiDesigner/icons/tree.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3">
+          <preferred-size width="150" height="50" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JTabbedPane" icon="/com/intellij/uiDesigner/icons/tabbedPane.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="3" hsize-policy="3" anchor="0" fill="3">
+          <preferred-size width="200" height="200" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JSplitPane" icon="/com/intellij/uiDesigner/icons/splitPane.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="3" hsize-policy="3" anchor="0" fill="3">
+          <preferred-size width="200" height="200" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JSpinner" icon="/com/intellij/uiDesigner/icons/spinner.png" removable="false" auto-create-binding="true" can-attach-label="true">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="8" fill="1" />
+      </item>
+      <item class="javax.swing.JSlider" icon="/com/intellij/uiDesigner/icons/slider.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="8" fill="1" />
+      </item>
+      <item class="javax.swing.JSeparator" icon="/com/intellij/uiDesigner/icons/separator.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="6" anchor="0" fill="3" />
+      </item>
+      <item class="javax.swing.JProgressBar" icon="/com/intellij/uiDesigner/icons/progressbar.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="0" fill="1" />
+      </item>
+      <item class="javax.swing.JToolBar" icon="/com/intellij/uiDesigner/icons/toolbar.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="6" anchor="0" fill="1">
+          <preferred-size width="-1" height="20" />
+        </default-constraints>
+      </item>
+      <item class="javax.swing.JToolBar$Separator" icon="/com/intellij/uiDesigner/icons/toolbarSeparator.png" removable="false" auto-create-binding="false" can-attach-label="false">
+        <default-constraints vsize-policy="0" hsize-policy="0" anchor="0" fill="1" />
+      </item>
+      <item class="javax.swing.JScrollBar" icon="/com/intellij/uiDesigner/icons/scrollbar.png" removable="false" auto-create-binding="true" can-attach-label="false">
+        <default-constraints vsize-policy="6" hsize-policy="0" anchor="0" fill="2" />
+      </item>
+    </group>
+  </component>
+</project>

.idea/vcs.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="VcsDirectoryMappings">
+    <mapping directory="$PROJECT_DIR$" vcs="Git" />
+  </component>
+</project>

.travis.yml

@@ -0,0 +1,11 @@
+language: java
+sudo: false
+# jdk: openjdk8
+# dist: xenial
+
+before_install:
+  - chmod +x ./gradlew
+
+script:
+  - ./gradlew test
+

README.md

@@ -1,105 +1,136 @@
-Prog8 - Structured Programming Language for 8-bit 6502/6510 microprocessors
-===========================================================================
+[![saythanks](https://img.shields.io/badge/say-thanks-ff69b4.svg)](https://saythanks.io/to/irmen)
+[![Build Status](https://travis-ci.org/irmen/prog8.svg?branch=master)](https://travis-ci.org/irmen/prog8)
+[![Documentation](https://readthedocs.org/projects/prog8/badge/?version=latest)](https://prog8.readthedocs.io/)
+
+Prog8 - Structured Programming Language for 8-bit 6502/65c02 microprocessors
+============================================================================
 
 *Written by Irmen de Jong (irmen@razorvine.net)*
 
 *Software license: GNU GPL 3.0, see file LICENSE*
 
 
-This is a structured programming language for the 8-bit 6502/6510 microprocessor from the late 1970's and 1980's
+This is a structured programming language for the 8-bit 6502/6510/65c02 microprocessor from the late 1970's and 1980's
 as used in many home computers from that era. It is a medium to low level programming language,
-which aims to provide many conveniences over raw assembly code (even when using a macro assembler):
+which aims to provide many conveniences over raw assembly code (even when using a macro assembler).
 
-- reduction of source code length
-- easier program understanding (because it's higher level, and more terse)
-- option to automatically run the compiled program in the Vice emulator  
+Documentation
+-------------
+Full documentation (syntax reference, how to use the language and the compiler, etc.) can be found at:
+https://prog8.readthedocs.io/
+
+
+What does Prog8 provide?
+------------------------
+
+- reduction of source code length over raw assembly
 - modularity, symbol scoping, subroutines
-- subroutines have enforced input- and output parameter definitions
-- various data types other than just bytes (16-bit words, floats, strings, 16-bit register pairs)
-- automatic variable allocations, automatic string variables and string sharing
-- constant folding in expressions (compile-time evaluation)
-- automatic type conversions
-- floating point operations
-- abstracting away low level aspects such as ZeroPage handling, program startup, explicit memory addresses
+- various data types other than just bytes (16-bit words, floats, strings)
+- automatic variable allocations, automatic string and array variables and string sharing
+- subroutines with input parameters and result values
+- high-level program optimizations
+- small program boilerplate/compilersupport overhead
+- sane variable initialization, programs can be restarted again just fine after exiting to basic
+- conditional branches
+- floating point operations  (requires the C64 Basic ROM routines for this)
+- 'when' statement to provide a concise jump table alternative to if/elseif chains
+- structs to group together sets of variables and manipulate them at once
+- many built-in functions such as ``sin``, ``cos``, ``rnd``, ``abs``, ``min``, ``max``, ``sqrt``, ``msb``, ``rol``, ``ror``, ``swap``, ``sort`` and ``reverse``
+- various powerful built-in libraries to do I/O, number conversions, graphics and more  
+- convenience abstractions for low level aspects such as ZeroPage handling, program startup, explicit memory addresses
+- fast execution speed due to compilation to native assembly code
+- variables are allocated statically
+- inline assembly allows you to have full control when every cycle or byte matters
+- supports the sixteen 'virtual' 16-bit registers R0 .. R15 from the Commander X16, and provides them also on the C64.
+- encode strings and characters into petscii or screencodes as desired (C64/Cx16)
+
+*Rapid edit-compile-run-debug cycle:*
+
+- use a modern PC to do the work on, use nice editors and enjoy quick compilation times
+- can automatically run the program in the Vice emulator after succesful compilation
 - breakpoints, that let the Vice emulator drop into the monitor if execution hits them
 - source code labels automatically loaded in Vice emulator so it can show them in disassembly
-- conditional gotos
-- various code optimizations (code structure, logical and numerical expressions, ...) 
+
+*Two supported compiler targets* (contributions to improve these or to add support for other machines are welcome!):
+
+- "c64": Commodore-64  (6510 CPU = almost a 6502)
+- "cx16": [CommanderX16](https://www.commanderx16.com)  (65c02 CPU)
+- If you only use standard kernal and prog8 library routines, it is possible to compile the *exact same program* for both machines (just change the compiler target flag)!
 
 
-It is mainly targeted at the Commodore-64 machine at this time.
 
-Documentation is online at https://prog8.readthedocs.io/
-
-
-Required tools:
----------------
+Additional required tools
+-------------------------
 
 [64tass](https://sourceforge.net/projects/tass64/) - cross assembler. Install this on your shell path.
 A recent .exe version of this tool for Windows can be obtained from my [clone](https://github.com/irmen/64tass/releases) of this project.
 For other platforms it is very easy to compile it yourself (make ; make install).
 
-A **Java runtime (jre or jdk), version 8 or newer**  is required to run the packaged compiler.
-If you want to build it from source, you'll need a Kotlin 1.3 SDK as well (or for instance,
+A **Java runtime (jre or jdk), version 11 or newer**  is required to run a prepackaged version of the compiler.
+If you want to build it from source, you'll need a Java SDK + Kotlin 1.3.x SDK (or for instance,
 IntelliJ IDEA with the Kotlin plugin).
 
-It's handy to have a C-64 emulator or a real C-64 to run the programs on. The compiler assumes the presence
-of the [Vice emulator](http://vice-emu.sourceforge.net/)
+It's handy to have an emulator (or a real machine perhaps!) to run the programs on. The compiler assumes the presence
+of the [Vice emulator](http://vice-emu.sourceforge.net/)  for the C64 target,
+and the [x16emu emulator](https://github.com/commanderx16/x16-emulator) for the CommanderX16 target.
 
 
 Example code
 ------------
 
-When this code is compiled::
+This code calculates prime numbers using the Sieve of Eratosthenes algorithm::
 
-    %import c64lib
-    %import c64utils
-    %import c64flt
-
-    ~ main {
+    %import textio
+    %zeropage basicsafe
+    
+    main {
+    
+        ubyte[256] sieve
+        ubyte candidate_prime = 2       ; is increased in the loop
+    
         sub start() {
-            ; set text color and activate lowercase charset
-            c64.COLOR = 13
-            c64.VMCSB |= 2
-
-            ; use optimized routine to write text
-            c64scr.print("Hello!\n")
-
-            ; use iteration to write text
-            str question = "How are you?\n"
-            for ubyte char in question
-                c64.CHROUT(char)
-
-            ; use indexed loop to write characters
-            str bye = "Goodbye!\n"
-            for ubyte c in 0 to len(bye)
-                c64.CHROUT(bye[c])
-
-
-            float clock_seconds = ((mkword(c64.TIME_LO, c64.TIME_MID) as float)
-                                    + (c64.TIME_HI as float)*65536.0)
-                                     / 60
-            float hours = floor(clock_seconds / 3600)
-            clock_seconds -= hours*3600
-            float minutes = floor(clock_seconds / 60)
-            clock_seconds = floor(clock_seconds - minutes * 60.0)
-
-            c64scr.print("system time in ti$ is ")
-            c64flt.print_f(hours)
-            c64.CHROUT(':')
-            c64flt.print_f(minutes)
-            c64.CHROUT(':')
-            c64flt.print_f(clock_seconds)
-            c64.CHROUT('\n')
+            sys.memset(sieve, 256, false)   ; clear the sieve
+            txt.print("prime numbers up to 255:\n\n")
+            ubyte amount=0
+            repeat {
+                ubyte prime = find_next_prime()
+                if prime==0
+                    break
+                txt.print_ub(prime)
+                txt.print(", ")
+                amount++
+            }
+            txt.nl()
+            txt.print("number of primes (expected 54): ")
+            txt.print_ub(amount)
+            txt.nl()
+        }
+    
+        sub find_next_prime() -> ubyte {
+            while sieve[candidate_prime] {
+                candidate_prime++
+                if candidate_prime==0
+                    return 0        ; we wrapped; no more primes
+            }
+    
+            ; found next one, mark the multiples and return it.
+            sieve[candidate_prime] = true
+            uword multiple = candidate_prime
+    
+            while multiple < len(sieve) {
+                sieve[lsb(multiple)] = true
+                multiple += candidate_prime
+            }
+            return candidate_prime
         }
     }
 
 
 
-you get a program that outputs this when loaded on a C-64:
 
-![c64 screen](docs/source/_static/hello_screen.png)
+when compiled an ran on a C-64 you'll get:
 
+![c64 screen](docs/source/_static/primes_example.png)
 
 One of the included examples (wizzine.p8) animates a bunch of sprite balloons and looks like this:
 
@@ -109,3 +140,12 @@ Another example (cube3d-sprites.p8) draws the vertices of a rotating 3d cube:
 
 ![cube3d screen](docs/source/_static/cube3d.png)
 
+If you want to play a video game, a fully working Tetris clone is included in the examples:
+
+![tehtriz_screen](docs/source/_static/tehtriz.png)
+
+There are a couple of examples specially made for the CommanderX16 compiler target.
+For instance here's a well known space ship animated in 3D with hidden line removal,
+in the CommanderX16 emulator:
+
+![cobra3d](docs/source/_static/cobra3d.png)

build_the_compiler.sh

@@ -1,31 +0,0 @@
-#!/usr/bin/env bash
-
-echo "Compiling the parser..."
-java -jar ./parser/antlr/lib/antlr-4.7.2-complete.jar -o ./parser/src/prog8/parser -Xexact-output-dir -no-listener -no-visitor ./parser/antlr/prog8.g4
-
-
-PARSER_CLASSES=./out/production/parser
-COMPILER_JAR=prog8compiler.jar
-ANTLR_RUNTIME=./parser/antlr/lib/antlr-runtime-4.7.2.jar
-
-mkdir -p ${PARSER_CLASSES}
-javac  -d ${PARSER_CLASSES} -cp ${ANTLR_RUNTIME} ./parser/src/prog8/parser/prog8Lexer.java  ./parser/src/prog8/parser/prog8Parser.java
-
-echo "Compiling the compiler itself..."
-kotlinc -verbose -include-runtime -d ${COMPILER_JAR} -cp ${ANTLR_RUNTIME}:${PARSER_CLASSES} ./compiler/src/prog8
-
-echo "Finalizing the compiler jar file..."
-# add the antlr parser classes
-jar ufe ${COMPILER_JAR} prog8.CompilerMainKt -C ${PARSER_CLASSES} prog8
-
-# add the resources
-jar uf ${COMPILER_JAR} -C ./compiler/res .
-
-# add the antlr runtime classes
-rm -rf antlr_runtime_extraction
-mkdir antlr_runtime_extraction
-(cd antlr_runtime_extraction; jar xf ../${ANTLR_RUNTIME})
-jar uf ${COMPILER_JAR} -C antlr_runtime_extraction org
-rm -rf antlr_runtime_extraction
-
-echo "Done!"

compiler/build.gradle

@@ -1,25 +1,51 @@
 plugins {
-    id "org.jetbrains.kotlin.jvm" version "1.3.20"
+    id 'java'
     id 'application'
+    id "org.jetbrains.kotlin.jvm" version "1.4.30"
+    id 'org.jetbrains.dokka' version "0.9.18"
+    id 'com.github.johnrengelman.shadow' version '6.1.0'
 }
 
+targetCompatibility = 11
+sourceCompatibility = 11
+
 repositories {
+    mavenLocal()
     mavenCentral()
-    jcenter()
+    maven { url "https://kotlin.bintray.com/kotlinx" }
 }
 
-def kotlinVersion = '1.3.20'
+def prog8version = rootProject.file('compiler/res/version.txt').text.trim()
 
 dependencies {
-    implementation project(':parser')
-    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8:$kotlinVersion"
-    runtime "org.jetbrains.kotlin:kotlin-reflect:$kotlinVersion"
+    implementation project(':compilerAst')
+    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
+    // implementation "org.jetbrains.kotlin:kotlin-reflect"
+    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.1'
+    // implementation 'net.razorvine:ksim65:1.8'
+    // implementation "com.github.hypfvieh:dbus-java:3.2.4"
 
-	testImplementation "org.jetbrains.kotlin:kotlin-test-junit5:$kotlinVersion"
+    testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
     testImplementation 'org.junit.jupiter:junit-jupiter-api:5.3.2'
     testImplementation 'org.hamcrest:hamcrest-junit:2.0.0.0'
+    testRuntimeOnly 'org.junit.jupiter:junit-jupiter-engine:5.3.2'
 }
 
+compileKotlin {
+    kotlinOptions {
+        jvmTarget = "11"
+        useIR = true
+        // verbose = true
+        // freeCompilerArgs += "-XXLanguage:+NewInference"
+    }
+}
+
+compileTestKotlin {
+    kotlinOptions {
+        jvmTarget = "11"
+        useIR = true
+    }
+}
 
 sourceSets {
     main {
@@ -37,32 +63,45 @@ sourceSets {
     }
 }
 
-
+startScripts.enabled = true
 
 application {
-    mainClassName = 'prog8.CompilerMainKt'
+    mainClass = 'prog8.CompilerMainKt'
+    mainClassName = 'prog8.CompilerMainKt'  // deprecated
     applicationName = 'p8compile'
 }
 
-task p8vmScript(type: CreateStartScripts) {
-    mainClassName = "prog8.StackVmMainKt"
-    applicationName = "p8vm"
-    outputDir = new File(project.buildDir, 'scripts')
-    classpath = jar.outputs.files + project.configurations.runtime
+artifacts {
+    archives shadowJar
 }
 
-applicationDistribution.into("bin") {
-            from(p8vmScript)
-            fileMode = 0755
-}
 
-task fatJar(type: Jar) {
-    manifest {
-        attributes 'Main-Class': 'prog8.CompilerMainKt'
-    }
+shadowJar {
     archiveBaseName = 'prog8compiler'
-    destinationDir = rootProject.projectDir
-    from { project.configurations.runtime.collect { it.isDirectory() ? it : zipTree(it) } }
-    with jar
+    archiveVersion = prog8version
+    // minimize()
+}
+
+
+test {
+    // Enable JUnit 5 (Gradle 4.6+).
+    useJUnitPlatform()
+
+    // Always run tests, even when nothing changed.
+    dependsOn 'cleanTest'
+
+    // Show test results.
+    testLogging {
+        events "skipped", "failed"
+    }
+}
+
+
+dokka {
+    outputFormat = 'html'
+    outputDirectory = "$buildDir/kdoc"
+}
+
+task wrapper(type: Wrapper) {
+    gradleVersion = '6.7'
 }
-// build.finalizedBy(fatJar)

compiler/compiler.iml

@@ -1,17 +1,24 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <module type="JAVA_MODULE" version="4">
+  <component name="FacetManager">
+    <facet type="Python" name="Python">
+      <configuration sdkName="Python 3.9" />
+    </facet>
+  </component>
   <component name="NewModuleRootManager" inherit-compiler-output="true">
     <exclude-output />
     <content url="file://$MODULE_DIR$">
-      <sourceFolder url="file://$MODULE_DIR$/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/res" type="java-resource" />
+      <sourceFolder url="file://$MODULE_DIR$/src" isTestSource="false" />
       <sourceFolder url="file://$MODULE_DIR$/test" isTestSource="true" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
     </content>
-    <orderEntry type="jdk" jdkName="1.8" jdkType="JavaSDK" />
+    <orderEntry type="jdk" jdkName="11" jdkType="JavaSDK" />
     <orderEntry type="sourceFolder" forTests="false" />
     <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
-    <orderEntry type="library" name="antlr-runtime-4.7.2" level="project" />
-    <orderEntry type="library" name="testlibs" level="project" />
-    <orderEntry type="module" module-name="parser" />
+    <orderEntry type="library" name="unittest-libs" level="project" />
+    <orderEntry type="library" name="kotlinx-cli-jvm" level="project" />
+    <orderEntry type="module" module-name="compilerAst" />
+    <orderEntry type="library" name="Python 3.9 interpreter library" level="application" />
   </component>
 </module>

compiler/res/.editorconfig

@@ -0,0 +1,20 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+indent_size = 4
+indent_style = space
+insert_final_newline = true
+max_line_length = 120
+tab_width = 8
+trim_trailing_whitespace = true
+ij_smart_tabs = true
+
+[*.p8]
+indent_size = 4
+indent_style = space
+
+[*.asm]
+indent_size = 8
+indent_style = tab

compiler/res/prog8lib/c64/floats.asm

@@ -0,0 +1,680 @@
+; --- low level floating point assembly routines for the C64
+
+FL_ONE_const	.byte  129     			; 1.0
+FL_ZERO_const	.byte  0,0,0,0,0		; 0.0
+FL_LOG2_const	.byte  $80, $31, $72, $17, $f8	; log(2)
+
+
+floats_store_reg	.byte  0		; temp storage
+
+
+ub2float	.proc
+		; -- convert ubyte in SCRATCH_ZPB1 to float at address A/Y
+		;    clobbers A, Y
+		stx  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  P8ZP_SCRATCH_B1
+		lda  #0
+		jsr  GIVAYF
+_fac_to_mem	ldx  P8ZP_SCRATCH_W2
+		ldy  P8ZP_SCRATCH_W2+1
+		jsr  MOVMF
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+b2float		.proc
+		; -- convert byte in SCRATCH_ZPB1 to float at address A/Y
+		;    clobbers A, Y
+		stx  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		lda  P8ZP_SCRATCH_B1
+		jsr  FREADSA
+		jmp  ub2float._fac_to_mem
+		.pend
+
+uw2float	.proc
+		; -- convert uword in SCRATCH_ZPWORD1 to float at address A/Y
+		stx  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		lda  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  GIVUAYFAY
+		jmp  ub2float._fac_to_mem
+		.pend
+
+w2float		.proc
+		; -- convert word in SCRATCH_ZPWORD1 to float at address A/Y
+		stx  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  P8ZP_SCRATCH_W1
+		lda  P8ZP_SCRATCH_W1+1
+		jsr  GIVAYF
+		jmp  ub2float._fac_to_mem
+		.pend
+
+
+cast_from_uw	.proc
+		; -- uword in A/Y into float var at (P8ZP_SCRATCH_W2)
+		stx  P8ZP_SCRATCH_REG
+		jsr  GIVUAYFAY
+		jmp  ub2float._fac_to_mem
+		.pend
+
+
+cast_from_w	.proc
+		; -- word in A/Y into float var at (P8ZP_SCRATCH_W2)
+		stx  P8ZP_SCRATCH_REG
+		jsr  GIVAYFAY
+		jmp  ub2float._fac_to_mem
+		.pend
+
+
+cast_from_ub	.proc
+		; -- ubyte in Y into float var at (P8ZP_SCRATCH_W2)
+		stx  P8ZP_SCRATCH_REG
+		jsr  FREADUY
+		jmp  ub2float._fac_to_mem
+		.pend
+
+
+cast_from_b	.proc
+		; -- byte in A into float var at (P8ZP_SCRATCH_W2)
+		stx  P8ZP_SCRATCH_REG
+		jsr  FREADSA
+		jmp  ub2float._fac_to_mem
+		.pend
+
+cast_as_uw_into_ya	.proc               ; also used for float 2 ub
+		; -- cast float at A/Y to uword into Y/A
+		jsr  MOVFM
+		jmp  cast_FAC1_as_uw_into_ya
+		.pend
+
+cast_as_w_into_ay	.proc               ; also used for float 2 b
+		; -- cast float at A/Y to word into A/Y
+		jsr  MOVFM
+		jmp  cast_FAC1_as_w_into_ay
+		.pend
+
+cast_FAC1_as_uw_into_ya	.proc               ; also used for float 2 ub
+		; -- cast fac1 to uword into Y/A
+		stx  P8ZP_SCRATCH_REG
+		jsr  GETADR     ; into Y/A
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+cast_FAC1_as_w_into_ay	.proc               ; also used for float 2 b
+		; -- cast fac1 to word into A/Y
+		stx  P8ZP_SCRATCH_REG
+		jsr  AYINT
+		ldy  $64
+		lda  $65
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+
+stack_b2float	.proc
+		; -- b2float operating on the stack
+		inx
+		lda  P8ESTACK_LO,x
+		stx  P8ZP_SCRATCH_REG
+		jsr  FREADSA
+		jmp  push_fac1._internal
+		.pend
+
+stack_w2float	.proc
+		; -- w2float operating on the stack
+		inx
+		ldy  P8ESTACK_LO,x
+		lda  P8ESTACK_HI,x
+		stx  P8ZP_SCRATCH_REG
+		jsr  GIVAYF
+		jmp  push_fac1._internal
+		.pend
+
+stack_ub2float	.proc
+		; -- ub2float operating on the stack
+		inx
+		lda  P8ESTACK_LO,x
+		stx  P8ZP_SCRATCH_REG
+		tay
+		lda  #0
+		jsr  GIVAYF
+		jmp  push_fac1._internal
+		.pend
+
+stack_uw2float	.proc
+		; -- uw2float operating on the stack
+		inx
+		lda  P8ESTACK_LO,x
+		ldy  P8ESTACK_HI,x
+		stx  P8ZP_SCRATCH_REG
+		jsr  GIVUAYFAY
+		jmp  push_fac1._internal
+		.pend
+
+stack_float2w	.proc               ; also used for float2b
+		jsr  pop_float_fac1
+		stx  P8ZP_SCRATCH_REG
+		jsr  AYINT
+		ldx  P8ZP_SCRATCH_REG
+		lda  $64
+		sta  P8ESTACK_HI,x
+		lda  $65
+		sta  P8ESTACK_LO,x
+		dex
+		rts
+		.pend
+
+stack_float2uw	.proc               ; also used for float2ub
+		jsr  pop_float_fac1
+		stx  P8ZP_SCRATCH_REG
+		jsr  GETADR
+		ldx  P8ZP_SCRATCH_REG
+		sta  P8ESTACK_HI,x
+		tya
+		sta  P8ESTACK_LO,x
+		dex
+		rts
+		.pend
+
+push_float	.proc
+		; ---- push mflpt5 in A/Y onto stack
+		; (taking 3 stack positions = 6 bytes of which 1 is padding)
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  P8ESTACK_LO,x
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  P8ESTACK_HI,x
+		dex
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  P8ESTACK_LO,x
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  P8ESTACK_HI,x
+		dex
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  P8ESTACK_LO,x
+		dex
+		rts
+		.pend
+
+pop_float	.proc
+		; ---- pops mflpt5 from stack to memory A/Y
+		; (frees 3 stack positions = 6 bytes of which 1 is padding)
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #4
+		inx
+		lda  P8ESTACK_LO,x
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		inx
+		lda  P8ESTACK_HI,x
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		lda  P8ESTACK_LO,x
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		inx
+		lda  P8ESTACK_HI,x
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		lda  P8ESTACK_LO,x
+		sta  (P8ZP_SCRATCH_W1),y
+		rts
+		.pend
+
+pop_float_fac1	.proc
+		; -- pops float from stack into FAC1
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  pop_float
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jmp  MOVFM
+		.pend
+
+copy_float	.proc
+		; -- copies the 5 bytes of the mflt value pointed to by SCRATCH_ZPWORD1,
+		;    into the 5 bytes pointed to by A/Y.  Clobbers A,Y.
+		sta  _target+1
+		sty  _target+2
+		ldy  #4
+_loop		lda  (P8ZP_SCRATCH_W1),y
+_target		sta  $ffff,y			; modified
+		dey
+		bpl  _loop
+		rts
+		.pend
+
+inc_var_f	.proc
+		; -- add 1 to float pointed to by A/Y
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		stx  P8ZP_SCRATCH_REG
+		jsr  MOVFM
+		lda  #<FL_ONE_const
+		ldy  #>FL_ONE_const
+		jsr  FADD
+		ldx  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  MOVMF
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+dec_var_f	.proc
+		; -- subtract 1 from float pointed to by A/Y
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		stx  P8ZP_SCRATCH_REG
+		lda  #<FL_ONE_const
+		ldy  #>FL_ONE_const
+		jsr  MOVFM
+		lda  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  FSUB
+		ldx  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  MOVMF
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+
+pop_2_floats_f2_in_fac1	.proc
+		; -- pop 2 floats from stack, load the second one in FAC1 as well
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		jsr  pop_float
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  pop_float
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		jmp  MOVFM
+		.pend
+
+
+fmath_float1	.byte 0,0,0,0,0	; storage for a mflpt5 value
+fmath_float2	.byte 0,0,0,0,0	; storage for a mflpt5 value
+
+
+push_fac1	.proc
+		; -- push the float in FAC1 onto the stack
+		stx  P8ZP_SCRATCH_REG
+_internal	ldx  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  MOVMF
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		ldx  P8ZP_SCRATCH_REG
+		jmp  push_float
+		.pend
+
+
+pow_f		.proc
+		; -- push f1 ** f2 on stack
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		jsr  pop_float
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  pop_float
+		stx  P8ZP_SCRATCH_REG
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  CONUPK		; fac2 = float1
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		jsr  FPWR
+		jmp  push_fac1._internal
+		.pend
+
+div_f		.proc
+		; -- push f1/f2 on stack
+		jsr  pop_2_floats_f2_in_fac1
+		stx  P8ZP_SCRATCH_REG
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  FDIV
+		jmp  push_fac1._internal
+		.pend
+
+add_f		.proc
+		; -- push f1+f2 on stack
+		jsr  pop_2_floats_f2_in_fac1
+		stx  P8ZP_SCRATCH_REG
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  FADD
+		jmp  push_fac1._internal
+		.pend
+
+sub_f		.proc
+		; -- push f1-f2 on stack
+		jsr  pop_2_floats_f2_in_fac1
+		stx  P8ZP_SCRATCH_REG
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  FSUB
+		jmp  push_fac1._internal
+		.pend
+
+mul_f		.proc
+		; -- push f1*f2 on stack
+		jsr  pop_2_floats_f2_in_fac1
+		stx  P8ZP_SCRATCH_REG
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  FMULT
+		jmp  push_fac1._internal
+		.pend
+
+neg_f		.proc
+		; -- toggle the sign bit on the stack
+		lda  P8ESTACK_HI+3,x
+		eor  #$80
+		sta  P8ESTACK_HI+3,x
+		rts
+		.pend
+
+var_fac1_less_f	.proc
+		; -- is the float in FAC1 < the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #255
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_lesseq_f	.proc
+		; -- is the float in FAC1 <= the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #0
+		beq  +
+		cmp  #255
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_greater_f	.proc
+		; -- is the float in FAC1 > the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #1
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_greatereq_f	.proc
+		; -- is the float in FAC1 >= the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #0
+		beq  +
+		cmp  #1
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_notequal_f	.proc
+		; -- are the floats numbers in FAC1 and the variable AY *not* identical?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		and  #1
+		rts
+		.pend
+
+vars_equal_f	.proc
+		; -- are the mflpt5 numbers in P8ZP_SCRATCH_W1 and AY identical?
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _false
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _false
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _false
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _false
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _false
+		lda  #1
+		rts
+_false		lda  #0
+		rts
+		.pend
+
+equal_f		.proc
+		; -- are the two mflpt5 numbers on the stack identical?
+		inx
+		inx
+		inx
+		inx
+		lda  P8ESTACK_LO-3,x
+		cmp  P8ESTACK_LO,x
+		bne  _equals_false
+		lda  P8ESTACK_LO-2,x
+		cmp  P8ESTACK_LO+1,x
+		bne  _equals_false
+		lda  P8ESTACK_LO-1,x
+		cmp  P8ESTACK_LO+2,x
+		bne  _equals_false
+		lda  P8ESTACK_HI-2,x
+		cmp  P8ESTACK_HI+1,x
+		bne  _equals_false
+		lda  P8ESTACK_HI-1,x
+		cmp  P8ESTACK_HI+2,x
+		bne  _equals_false
+_equals_true	lda  #1
+_equals_store	inx
+		sta  P8ESTACK_LO+1,x
+		rts
+_equals_false	lda  #0
+		beq  _equals_store
+		.pend
+
+notequal_f	.proc
+		; -- are the two mflpt5 numbers on the stack different?
+		jsr  equal_f
+		eor  #1		; invert the result
+		sta  P8ESTACK_LO+1,x
+		rts
+		.pend
+
+vars_less_f	.proc
+		; -- is float in AY < float in P8ZP_SCRATCH_W2 ?
+		jsr  MOVFM
+		lda  P8ZP_SCRATCH_W2
+		ldy  P8ZP_SCRATCH_W2+1
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #255
+		bne  +
+		lda  #1
+		rts
++		lda  #0
+		rts
+		.pend
+
+vars_lesseq_f	.proc
+		; -- is float in AY <= float in P8ZP_SCRATCH_W2 ?
+		jsr  MOVFM
+		lda  P8ZP_SCRATCH_W2
+		ldy  P8ZP_SCRATCH_W2+1
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #255
+		bne  +
+-		lda  #1
+		rts
++		cmp  #0
+		beq  -
+		lda  #0
+		rts
+		.pend
+
+less_f		.proc
+		; -- is f1 < f2?
+		jsr  compare_floats
+		cmp  #255
+		beq  compare_floats._return_true
+		bne  compare_floats._return_false
+		.pend
+
+
+lesseq_f	.proc
+		; -- is f1 <= f2?
+		jsr  compare_floats
+		cmp  #255
+		beq  compare_floats._return_true
+		cmp  #0
+		beq  compare_floats._return_true
+		bne  compare_floats._return_false
+		.pend
+
+greater_f	.proc
+		; -- is f1 > f2?
+		jsr  compare_floats
+		cmp  #1
+		beq  compare_floats._return_true
+		bne  compare_floats._return_false
+		.pend
+
+greatereq_f	.proc
+		; -- is f1 >= f2?
+		jsr  compare_floats
+		cmp  #1
+		beq  compare_floats._return_true
+		cmp  #0
+		beq  compare_floats._return_true
+		bne  compare_floats._return_false
+		.pend
+
+compare_floats	.proc
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		jsr  pop_float
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  pop_float
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  MOVFM		; fac1 = flt1
+		lda  #<fmath_float2
+		ldy  #>fmath_float2
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP		; A = flt1 compared with flt2 (0=equal, 1=flt1>flt2, 255=flt1<flt2)
+		ldx  P8ZP_SCRATCH_REG
+		rts
+_return_false	lda  #0
+_return_result  sta  P8ESTACK_LO,x
+		dex
+		rts
+_return_true	lda  #1
+		bne  _return_result
+		.pend
+
+set_array_float_from_fac1	.proc
+		; -- set the float in FAC1 in the array (index in A, array in P8ZP_SCRATCH_W1)
+		sta  P8ZP_SCRATCH_B1
+		asl  a
+		asl  a
+		clc
+		adc  P8ZP_SCRATCH_B1
+		ldy  P8ZP_SCRATCH_W1+1
+		clc
+		adc  P8ZP_SCRATCH_W1
+		bcc  +
+		iny
++		stx  floats_store_reg
+		tax
+		jsr  MOVMF
+		ldx  floats_store_reg
+		rts
+		.pend
+
+
+set_0_array_float	.proc
+		; -- set a float in an array to zero (index in A, array in P8ZP_SCRATCH_W1)
+		sta  P8ZP_SCRATCH_B1
+		asl  a
+		asl  a
+		clc
+		adc  P8ZP_SCRATCH_B1
+		tay
+		lda  #0
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		sta  (P8ZP_SCRATCH_W1),y
+		rts
+		.pend
+
+
+set_array_float		.proc
+		; -- set a float in an array to a value (index in A, float in P8ZP_SCRATCH_W1, array in P8ZP_SCRATCH_W2)
+		sta  P8ZP_SCRATCH_B1
+		asl  a
+		asl  a
+		clc
+		adc  P8ZP_SCRATCH_B1
+		adc  P8ZP_SCRATCH_W2
+		ldy  P8ZP_SCRATCH_W2+1
+		bcc  +
+		iny
++		jmp  copy_float
+			; -- copies the 5 bytes of the mflt value pointed to by SCRATCH_ZPWORD1,
+			;    into the 5 bytes pointed to by A/Y.  Clobbers A,Y.
+		.pend
+
+

compiler/res/prog8lib/c64/floats.p8

@@ -0,0 +1,201 @@
+; Prog8 definitions for floating point handling on the Commodore-64
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target c64
+%option enable_floats
+
+floats {
+	; ---- this block contains C-64 floating point related functions ----
+
+        const float  PI     = 3.141592653589793
+        const float  TWOPI  = 6.283185307179586
+
+
+; ---- C64 basic and kernal ROM float constants and functions ----
+
+		; note: the fac1 and fac2 are working registers and take 6 bytes each,
+		; floats in memory  (and rom) are stored in 5-byte MFLPT packed format.
+
+
+; note: fac1/2 might get clobbered even if not mentioned in the function's name.
+; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
+
+romsub $bba2 = MOVFM(uword mflpt @ AY) clobbers(A,Y)        ; load mflpt value from memory  in A/Y into fac1
+romsub $bba6 = FREADMEM() clobbers(A,Y)                     ; load mflpt value from memory  in $22/$23 into fac1
+romsub $ba8c = CONUPK(uword mflpt @ AY) clobbers(A,Y)       ; load mflpt value from memory  in A/Y into fac2
+romsub $ba90 = FAREADMEM() clobbers(A,Y)                    ; load mflpt value from memory  in $22/$23 into fac2
+romsub $bbfc = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+romsub $bc0c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded)
+romsub $bc0f = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+romsub $bbd4 = MOVMF(uword mflpt @ XY) clobbers(A,Y)        ; store fac1 to memory  X/Y as 5-byte mflpt
+
+; fac1-> signed word in Y/A (might throw ILLEGAL QUANTITY)
+; (tip: use floats.FTOSWRDAY to get A/Y output; lo/hi switched to normal little endian order)
+romsub $b1aa = FTOSWORDYA() clobbers(X) -> ubyte @ Y, ubyte @ A       ; note: calls AYINT.
+
+; fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
+; (tip: use floats.GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
+romsub $b7f7 = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
+
+romsub $bc9b = QINT() clobbers(A,X,Y)           ; fac1 -> 4-byte signed integer in 98-101 ($62-$65), with the MSB FIRST.
+romsub $b1bf = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101 ($64-$65) MSB FIRST. (might throw ILLEGAL QUANTITY)
+
+; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1
+; (tip: use floats.GIVAYFAY to use A/Y input; lo/hi switched to normal order)
+; there is also floats.GIVUAYFAY - unsigned word in A/Y (lo/hi) to fac1
+; there is also floats.FREADS32  that reads from 98-101 ($62-$65) MSB FIRST
+; there is also floats.FREADUS32  that reads from 98-101 ($62-$65) MSB FIRST
+; there is also floats.FREADS24AXY  that reads signed int24 into fac1 from A/X/Y (lo/mid/hi bytes)
+romsub $b391 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
+
+romsub $b3a2 = FREADUY(ubyte value @ Y) clobbers(A,X,Y)     ; 8 bit unsigned Y -> float in fac1
+romsub $bc3c = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
+romsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length
+romsub $aabc = FPRINTLN() clobbers(A,X,Y)                   ; print string of fac1, on one line (= with newline) destroys fac1.  (consider FOUT + STROUT as well)
+romsub $bddd = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY ($0100)
+
+romsub $b849 = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for rounding- call this before INT
+romsub $bae2 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
+romsub $bafe = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive!
+romsub $bc5b = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
+
+romsub $b86a = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
+romsub $b867 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
+romsub $b853 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
+romsub $b850 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
+romsub $ba2b = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
+romsub $ba28 = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
+romsub $bb12 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
+romsub $bb0f = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
+romsub $bf7b = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+romsub $bf78 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** mflpt from A/Y
+romsub $bd7e = FINLOG(byte value @A) clobbers (A, X, Y)     ; fac1 += signed byte in A
+
+romsub $aed4 = NOTOP() clobbers(A,X,Y)                      ; fac1 = NOT(fac1)
+romsub $bccc = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
+romsub $b9ea = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
+romsub $bc39 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
+romsub $bc2b = SIGN() -> ubyte @ A                          ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+romsub $bc58 = ABS()                                        ; fac1 = ABS(fac1)
+romsub $bf71 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+romsub $bf74 = SQRA() clobbers(A,X,Y)                       ; fac1 = SQRT(fac2)
+romsub $bfed = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+romsub $bfb4 = NEGOP() clobbers(A)                          ; switch the sign of fac1
+romsub $e097 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
+romsub $e264 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+romsub $e26b = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+romsub $e2b4 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+romsub $e30e = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+
+
+
+asmsub  FREADS32() clobbers(A,X,Y)  {
+	; ---- fac1 = signed int32 from $62-$65 big endian (MSB FIRST)
+	%asm {{
+		lda  $62
+		eor  #$ff
+		asl  a
+		lda  #0
+		ldx  #$a0
+		jmp  $bc4f		; internal BASIC routine
+	}}
+}
+
+asmsub  FREADUS32  () clobbers(A,X,Y)  {
+	; ---- fac1 = uint32 from $62-$65 big endian (MSB FIRST)
+	%asm {{
+		sec
+		lda  #0
+		ldx  #$a0
+		jmp  $bc4f		; internal BASIC routine
+	}}
+}
+
+asmsub  FREADS24AXY  (ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) clobbers(A,X,Y)  {
+	; ---- fac1 = signed int24 (A/X/Y contain lo/mid/hi bytes)
+	;      note: there is no FREADU24AXY (unsigned), use FREADUS32 instead.
+	%asm {{
+		sty  $62
+		stx  $63
+		sta  $64
+		lda  $62
+		eor  #$FF
+		asl  a
+		lda  #0
+		sta  $65
+		ldx  #$98
+		jmp  $bc4f		; internal BASIC routine
+	}}
+}
+
+asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
+	%asm {{
+		sty  $62
+		sta  $63
+		ldx  #$90
+		sec
+		jmp  $bc49		; internal BASIC routine
+	}}
+}
+
+asmsub  GIVAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- signed 16 bit word in A/Y (lo/hi) to float in fac1
+	%asm {{
+		sta  P8ZP_SCRATCH_REG
+		tya
+		ldy  P8ZP_SCRATCH_REG
+		jmp  GIVAYF		; this uses the inverse order, Y/A
+	}}
+}
+
+asmsub  FTOSWRDAY  () clobbers(X) -> uword @ AY  {
+	; ---- fac1 to signed word in A/Y
+	%asm {{
+		jsr  FTOSWORDYA	; note the inverse Y/A order
+		sta  P8ZP_SCRATCH_REG
+		tya
+		ldy  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
+	; ---- fac1 to unsigned word in A/Y
+	%asm {{
+		jsr  GETADR		; this uses the inverse order, Y/A
+		sta  P8ZP_SCRATCH_B1
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		rts
+	}}
+}
+
+sub  print_f  (float value) {
+	; ---- prints the floating point value (without a newline).
+	%asm {{
+		stx  floats_store_reg
+		lda  #<value
+		ldy  #>value
+		jsr  MOVFM		; load float into fac1
+		jsr  FOUT		; fac1 to string in A/Y
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		ldx  floats_store_reg
+		rts
+	}}
+}
+
+%asminclude "library:c64/floats.asm", ""
+%asminclude "library:c64/floats_funcs.asm", ""
+
+}

compiler/res/prog8lib/c64/floats_funcs.asm

@@ -0,0 +1,437 @@
+; --- floating point builtin functions
+
+
+abs_f_stack	.proc
+		; -- push abs(AY) on stack
+		jsr  floats.MOVFM
+		jsr  floats.ABS
+		jmp  push_fac1
+		.pend
+
+abs_f_fac1	.proc
+		; -- FAC1 = abs(AY)
+		jsr  floats.MOVFM
+		jmp  floats.ABS
+		.pend
+
+func_atan_stack	.proc
+		jsr  func_atan_fac1
+		jmp  push_fac1
+		.pend
+
+func_atan_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  ATN
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_ceil_stack	.proc
+		jsr  func_ceil_fac1
+		jmp  push_fac1
+		.pend
+
+func_ceil_fac1	.proc
+		; -- ceil: tr = int(f); if tr==f -> return  else return tr+1
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		ldx  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  MOVMF
+		jsr  INT
+		lda  #<fmath_float1
+		ldy  #>fmath_float1
+		jsr  FCOMP
+		cmp  #0
+		beq  +
+		lda  #<FL_ONE_const
+		ldy  #>FL_ONE_const
+		jsr  FADD
++		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_floor_stack	.proc
+		jsr  func_floor_fac1
+		jmp  push_fac1
+		.pend
+
+func_floor_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  INT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_round_stack	.proc
+		jsr  func_round_fac1
+		jmp  push_fac1
+		.pend
+
+func_round_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  FADDH
+		jsr  INT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_sin_stack	.proc
+		jsr  func_sin_fac1
+		jmp  push_fac1
+		.pend
+
+func_sin_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  SIN
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_cos_stack	.proc
+		jsr  func_cos_fac1
+		jmp  push_fac1
+		.pend
+
+func_cos_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  COS
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_tan_stack	.proc
+		jsr  func_tan_fac1
+		jmp  push_fac1
+		.pend
+
+func_tan_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  TAN
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_rad_stack	.proc
+		jsr  func_rad_fac1
+		jmp  push_fac1
+		.pend
+
+func_rad_fac1	.proc
+		; -- convert degrees to radians (d * pi / 180)
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		lda  #<_pi_div_180
+		ldy  #>_pi_div_180
+		jsr  FMULT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+_pi_div_180	.byte 123, 14, 250, 53, 18		; pi / 180
+		.pend
+
+func_deg_stack	.proc
+		jsr  func_deg_fac1
+		jmp  push_fac1
+		.pend
+
+func_deg_fac1	.proc
+		; -- convert radians to degrees (d * (1/ pi * 180))
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		lda  #<_one_over_pi_div_180
+		ldy  #>_one_over_pi_div_180
+		jsr  FMULT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+_one_over_pi_div_180	.byte 134, 101, 46, 224, 211		; 1 / (pi * 180)
+		.pend
+
+func_ln_stack	.proc
+		jsr  func_ln_fac1
+		jmp  push_fac1
+		.pend
+
+func_ln_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  LOG
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_log2_stack	.proc
+		jsr  func_log2_fac1
+		jmp  push_fac1
+		.pend
+
+func_log2_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  LOG
+		jsr  MOVEF
+		lda  #<FL_LOG2_const
+		ldy  #>FL_LOG2_const
+		jsr  MOVFM
+		jsr  FDIVT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_sign_f_stack	.proc
+		jsr  func_sign_f_into_A
+		sta  P8ESTACK_LO,x
+		dex
+		rts
+		.pend
+
+func_sign_f_into_A	.proc
+		jsr  MOVFM
+		jmp  SIGN
+		.pend
+
+func_sqrt_stack	.proc
+		jsr  func_sqrt_fac1
+		jmp  push_fac1
+		.pend
+
+func_sqrt_fac1	.proc
+		jsr  MOVFM
+		stx  P8ZP_SCRATCH_REG
+		jsr  SQR
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_rndf_stack	.proc
+		jsr  func_rndf_fac1
+		jmp  push_fac1
+		.pend
+
+func_rndf_fac1	.proc
+		stx  P8ZP_SCRATCH_REG
+		lda  #1
+		jsr  FREADSA
+		jsr  RND		; rng into fac1
+		ldx  P8ZP_SCRATCH_REG
+		rts
+		.pend
+
+func_swap_f	.proc
+		; -- swap floats pointed to by SCRATCH_ZPWORD1, SCRATCH_ZPWORD2
+		ldy  #4
+-               lda  (P8ZP_SCRATCH_W1),y
+		pha
+		lda  (P8ZP_SCRATCH_W2),y
+		sta  (P8ZP_SCRATCH_W1),y
+		pla
+		sta  (P8ZP_SCRATCH_W2),y
+		dey
+		bpl  -
+		rts
+		.pend
+
+func_reverse_f	.proc
+		; --- reverse an array of floats (array in P8ZP_SCRATCH_W1, num elements in A)
+_left_index = P8ZP_SCRATCH_W2
+_right_index = P8ZP_SCRATCH_W2+1
+_loop_count = P8ZP_SCRATCH_REG
+		pha
+		jsr  a_times_5
+		sec
+		sbc  #5
+		sta  _right_index
+		lda  #0
+		sta  _left_index
+		pla
+		lsr  a
+		sta  _loop_count
+_loop		; push the left indexed float on the stack
+		ldy  _left_index
+		lda  (P8ZP_SCRATCH_W1),y
+		pha
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		pha
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		pha
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		pha
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		pha
+		; copy right index float to left index float
+		ldy  _right_index
+		lda  (P8ZP_SCRATCH_W1),y
+		ldy  _left_index
+		sta  (P8ZP_SCRATCH_W1),y
+		inc  _left_index
+		inc  _right_index
+		ldy  _right_index
+		lda  (P8ZP_SCRATCH_W1),y
+		ldy  _left_index
+		sta  (P8ZP_SCRATCH_W1),y
+		inc  _left_index
+		inc  _right_index
+		ldy  _right_index
+		lda  (P8ZP_SCRATCH_W1),y
+		ldy  _left_index
+		sta  (P8ZP_SCRATCH_W1),y
+		inc  _left_index
+		inc  _right_index
+		ldy  _right_index
+		lda  (P8ZP_SCRATCH_W1),y
+		ldy  _left_index
+		sta  (P8ZP_SCRATCH_W1),y
+		inc  _left_index
+		inc  _right_index
+		ldy  _right_index
+		lda  (P8ZP_SCRATCH_W1),y
+		ldy  _left_index
+		sta  (P8ZP_SCRATCH_W1),y
+		; pop the float off the stack into the right index float
+		ldy  _right_index
+		pla
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		pla
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		pla
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		pla
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		pla
+		sta  (P8ZP_SCRATCH_W1),y
+		inc  _left_index
+		lda  _right_index
+		sec
+		sbc  #9
+		sta  _right_index
+		dec  _loop_count
+		bne  _loop
+		rts
+
+		.pend
+
+
+
+a_times_5	.proc
+		sta  P8ZP_SCRATCH_B1
+		asl  a
+		asl  a
+		clc
+		adc  P8ZP_SCRATCH_B1
+		rts
+		.pend
+
+func_any_f_into_A	.proc
+		jsr  a_times_5
+		jmp  prog8_lib.func_any_b_into_A
+		.pend
+
+func_all_f_into_A	.proc
+		jsr  a_times_5
+		jmp  prog8_lib.func_all_b_into_A
+		.pend
+
+func_any_f_stack	.proc
+		jsr  a_times_5
+		jmp  prog8_lib.func_any_b_stack
+		.pend
+
+func_all_f_stack	.proc
+		jsr  a_times_5
+		jmp  prog8_lib.func_all_b_stack
+		.pend
+
+func_max_f_stack	.proc
+		jsr  func_max_f_fac1
+		jmp  push_fac1
+		.pend
+
+func_max_f_fac1	.proc
+		; -- max(array) -> fac1,  array in P8ZP_SCRATCH_W1, num elts in A
+_loop_count = P8ZP_SCRATCH_REG
+		stx  floats_store_reg
+		sta  _loop_count
+		lda  #255
+		sta  _minmax_cmp+1		; modifying
+		lda  #<_largest_neg_float
+		ldy  #>_largest_neg_float
+_minmax_entry	jsr  MOVFM
+-		lda  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  FCOMP
+_minmax_cmp	cmp  #255			; modified
+		bne  +
+		lda  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  MOVFM
++		lda  P8ZP_SCRATCH_W1
+		clc
+		adc  #5
+		sta  P8ZP_SCRATCH_W1
+		bcc  +
+		inc  P8ZP_SCRATCH_W1+1
++		dec  _loop_count
+		bne  -
+		ldx  floats_store_reg
+		rts
+_largest_neg_float	.byte 255,255,255,255,255		; largest negative float -1.7014118345e+38
+		.pend
+
+func_min_f_stack	.proc
+		jsr  func_min_f_fac1
+		jmp  push_fac1
+		.pend
+
+func_min_f_fac1	.proc
+		; -- min(array) -> fac1,  array in P8ZP_SCRATCH_W1, num elts in A
+		sta  func_max_f_fac1._loop_count
+		lda  #1
+		sta  func_max_f_fac1._minmax_cmp+1
+		lda  #<_largest_pos_float
+		ldy  #>_largest_pos_float
+		jmp  func_max_f_fac1._minmax_entry
+_largest_pos_float	.byte  255,127,255,255,255		; largest positive float
+		rts
+		.pend
+
+
+func_sum_f_stack	.proc
+		jsr  func_sum_f_fac1
+		jmp  push_fac1
+		.pend
+
+func_sum_f_fac1	.proc
+		; -- sum(array) -> fac1,  array in P8ZP_SCRATCH_W1, num elts in A
+_loop_count = P8ZP_SCRATCH_REG
+		stx  floats_store_reg
+		sta  _loop_count
+		lda  #<FL_ZERO_const
+		ldy  #>FL_ZERO_const
+		jsr  MOVFM
+-		lda  P8ZP_SCRATCH_W1
+		ldy  P8ZP_SCRATCH_W1+1
+		jsr  FADD
+		lda  P8ZP_SCRATCH_W1
+		clc
+		adc  #5
+		sta  P8ZP_SCRATCH_W1
+		bcc  +
+		inc  P8ZP_SCRATCH_W1+1
++		dec  _loop_count
+		bne  -
+		ldx  floats_store_reg
+		rts
+		.pend

compiler/res/prog8lib/c64/graphics.p8

@@ -0,0 +1,370 @@
+%target c64
+%import textio
+
+; bitmap pixel graphics module for the C64
+; only black/white monochrome 320x200 for now
+; assumes bitmap screen memory is $2000-$3fff
+
+graphics {
+    const uword BITMAP_ADDRESS = $2000
+    const uword WIDTH = 320
+    const ubyte HEIGHT = 200
+
+    sub enable_bitmap_mode() {
+        ; enable bitmap screen, erase it and set colors to black/white.
+        c64.SCROLY = %00111011
+        c64.SCROLX = %00001000
+        c64.VMCSB = (c64.VMCSB & %11110000) | %00001000   ; $2000-$3fff
+        clear_screen(1, 0)
+    }
+
+    sub disable_bitmap_mode() {
+        ; enables text mode, erase the text screen, color white
+        c64.SCROLY = %00011011
+        c64.SCROLX = %00001000
+        c64.VMCSB = (c64.VMCSB & %11110000) | %00000100   ; $1000-$2fff
+        txt.fill_screen(' ', 1)
+    }
+
+    sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
+        sys.memset(BITMAP_ADDRESS, 320*200/8, 0)
+        txt.fill_screen(pixelcolor << 4 | bgcolor, 0)
+    }
+
+    sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        if y1>y2 {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            swap(x1, x2)
+            swap(y1, y2)
+        }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as ubyte +1)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1)
+            return
+        }
+
+        word @zp d = 0
+        ubyte positive_ix = true
+        if dx < 0 {
+            dx = -dx
+            positive_ix = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+        internal_plotx = x1
+
+        if dx >= dy {
+            if positive_ix {
+                repeat {
+                    internal_plot(y1)
+                    if internal_plotx==x2
+                        return
+                    internal_plotx++
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    internal_plot(y1)
+                    if internal_plotx==x2
+                        return
+                    internal_plotx--
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if positive_ix {
+                repeat {
+                    internal_plot(y1)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        internal_plotx++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    internal_plot(y1)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        internal_plotx--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+    }
+
+    sub rect(uword x, ubyte y, uword width, ubyte height) {
+        if width==0 or height==0
+            return
+        horizontal_line(x, y, width)
+        if height==1
+            return
+        horizontal_line(x, y+height-1, width)
+        vertical_line(x, y+1, height-2)
+        if width==1
+            return
+        vertical_line(x+width-1, y+1, height-2)
+    }
+
+    sub fillrect(uword x, ubyte y, uword width, ubyte height) {
+        if width==0
+            return
+        repeat height {
+            horizontal_line(x, y, width)
+            y++
+        }
+    }
+
+    sub horizontal_line(uword x, ubyte y, uword length) {
+        if length<8 {
+            internal_plotx=x
+            repeat lsb(length) {
+                internal_plot(y)
+                internal_plotx++
+            }
+            return
+        }
+
+        ubyte separate_pixels = lsb(x) & 7
+        uword addr = get_y_lookup(y) + (x&$fff8)
+
+        if separate_pixels {
+            %asm {{
+                lda  addr
+                sta  P8ZP_SCRATCH_W1
+                lda  addr+1
+                sta  P8ZP_SCRATCH_W1+1
+                ldy  separate_pixels
+                lda  _filled_right,y
+                eor  #255
+                ldy  #0
+                ora  (P8ZP_SCRATCH_W1),y
+                sta  (P8ZP_SCRATCH_W1),y
+            }}
+            addr += 8
+            length += separate_pixels
+            length -= 8
+        }
+
+        if length {
+            %asm {{
+                lda  length
+                and  #7
+                sta  separate_pixels
+                stx  P8ZP_SCRATCH_REG
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lda  addr
+                sta  _modified+1
+                lda  addr+1
+                sta  _modified+2
+                lda  length
+                ora  length+1
+                beq  _zero
+                ldy  length
+                ldx  #$ff
+_modified       stx  $ffff      ; modified
+                lda  _modified+1
+                clc
+                adc  #8
+                sta  _modified+1
+                bcc  +
+                inc  _modified+2
++               dey
+                bne  _modified
+_zero           ldx  P8ZP_SCRATCH_REG
+
+                ldy  separate_pixels
+                beq  _zero2
+                lda  _modified+1
+                sta  P8ZP_SCRATCH_W1
+                lda  _modified+2
+                sta  P8ZP_SCRATCH_W1+1
+                lda  _filled_right,y
+                ldy  #0
+                ora  (P8ZP_SCRATCH_W1),y
+                sta  (P8ZP_SCRATCH_W1),y
+                jmp  _zero2
+_filled_right   .byte  0, %10000000, %11000000, %11100000, %11110000, %11111000, %11111100, %11111110
+_zero2
+            }}
+        }
+    }
+
+    sub vertical_line(uword x, ubyte y, ubyte height) {
+        internal_plotx = x
+        repeat height {
+            internal_plot(y)
+            y++
+        }
+    }
+
+    sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; Midpoint algorithm
+        if radius==0
+            return
+        ubyte @zp ploty
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            internal_plotx = xcenter + radius
+            ploty = ycenter + yy
+            internal_plot(ploty)
+            internal_plotx = xcenter - radius
+            internal_plot(ploty)
+            internal_plotx = xcenter + radius
+            ploty = ycenter - yy
+            internal_plot(ploty)
+            internal_plotx = xcenter - radius
+            internal_plot(ploty)
+            internal_plotx = xcenter + yy
+            ploty = ycenter + radius
+            internal_plot(ploty)
+            internal_plotx = xcenter - yy
+            internal_plot(ploty)
+            internal_plotx = xcenter + yy
+            ploty = ycenter - radius
+            internal_plot(ploty)
+            internal_plotx = xcenter - yy
+            internal_plot(ploty)
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
+            }
+        }
+    }
+
+    sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
+            }
+        }
+    }
+
+
+; here is the non-asm code for the plot routine below:
+;    sub plot_nonasm(uword px, ubyte py) {
+;        ubyte[] ormask = [128, 64, 32, 16, 8, 4, 2, 1]
+;        uword addr = BITMAP_ADDRESS + 320*(py>>3) + (py & 7) + (px & %0000000111111000)
+;        @(addr) |= ormask[lsb(px) & 7]
+;    }
+
+    inline asmsub  plot(uword plotx @XY, ubyte ploty @A) clobbers (A, X, Y) {
+        %asm {{
+            stx  graphics.internal_plotx
+            sty  graphics.internal_plotx+1
+            jsr  graphics.internal_plot
+        }}
+    }
+
+    ; for efficiency of internal algorithms here is the internal plot routine
+    ; that takes the plotx coordinate in a separate variable instead of the XY register pair:
+
+    uword internal_plotx     ; 0..319        ; separate 'parameter' for internal_plot()
+
+    asmsub  internal_plot(ubyte ploty @A) clobbers (A, X, Y) {      ; internal_plotx is 16 bits 0 to 319... doesn't fit in a register
+        %asm {{
+        tay
+        lda  internal_plotx+1
+        sta  P8ZP_SCRATCH_W2+1
+        lsr  a            ; 0
+        sta  P8ZP_SCRATCH_W2
+        lda  internal_plotx
+        pha
+        and  #7
+        tax
+
+        lda  _y_lookup_lo,y
+        clc
+        adc  P8ZP_SCRATCH_W2
+        sta  P8ZP_SCRATCH_W2
+        lda  _y_lookup_hi,y
+        adc  P8ZP_SCRATCH_W2+1
+        sta  P8ZP_SCRATCH_W2+1
+
+        pla     ; internal_plotx
+        and  #%11111000
+        tay
+        lda  (P8ZP_SCRATCH_W2),y
+        ora  _ormask,x
+        sta  (P8ZP_SCRATCH_W2),y
+        rts
+
+_ormask     .byte 128, 64, 32, 16, 8, 4, 2, 1
+
+; note: this can be even faster if we also have a 256 byte x-lookup table, but hey.
+; see http://codebase64.org/doku.php?id=base:various_techniques_to_calculate_adresses_fast_common_screen_formats_for_pixel_graphics
+; the y lookup tables encodes this formula:  BITMAP_ADDRESS + 320*(py>>3) + (py & 7)    (y from 0..199)
+; We use the 64tass syntax for range expressions to calculate this table on assembly time.
+
+_plot_y_values := $2000 + 320*(range(200)>>3) + (range(200) & 7)
+
+_y_lookup_lo    .byte  <_plot_y_values
+_y_lookup_hi    .byte  >_plot_y_values
+
+        }}
+    }
+
+    asmsub get_y_lookup(ubyte y @Y) -> uword @AY {
+        %asm {{
+            lda  internal_plot._y_lookup_lo,y
+            pha
+            lda  internal_plot._y_lookup_hi,y
+            tay
+            pla
+            rts
+        }}
+    }
+
+}
+
+

compiler/res/prog8lib/c64/syslib.p8

@@ -0,0 +1,682 @@
+; Prog8 definitions for the Commodore-64
+; Including memory registers, I/O registers, Basic and Kernal subroutines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target c64
+
+c64 {
+        &ubyte  TIME_HI         = $a0       ; software jiffy clock, hi byte
+        &ubyte  TIME_MID        = $a1       ;  .. mid byte
+        &ubyte  TIME_LO         = $a2       ;    .. lo byte. Updated by IRQ every 1/60 sec
+        &ubyte  STATUS          = $90       ; kernal status variable for I/O
+        &ubyte  STKEY           = $91       ; various keyboard statuses (updated by IRQ)
+        &ubyte  SFDX            = $cb       ; current key pressed (matrix value) (updated by IRQ)
+
+        &ubyte  COLOR           = $0286     ; cursor color
+        &ubyte  HIBASE          = $0288     ; screen base address / 256 (hi-byte of screen memory address)
+        &uword  CINV            = $0314     ; IRQ vector
+        &uword  NMI_VEC         = $FFFA     ; 6502 nmi vector, determined by the kernal if banked in
+        &uword  RESET_VEC       = $FFFC     ; 6502 reset vector, determined by the kernal if banked in
+        &uword  IRQ_VEC         = $FFFE     ; 6502 interrupt vector, determined by the kernal if banked in
+
+        ; the default addresses for the character screen chars and colors
+        const  uword  Screen    = $0400     ; to have this as an array[40*25] the compiler would have to support array size > 255
+        const  uword  Colors    = $d800     ; to have this as an array[40*25] the compiler would have to support array size > 255
+
+        ; the default locations of the 8 sprite pointers (store address of sprite / 64)
+        &ubyte  SPRPTR0         = 2040
+        &ubyte  SPRPTR1         = 2041
+        &ubyte  SPRPTR2         = 2042
+        &ubyte  SPRPTR3         = 2043
+        &ubyte  SPRPTR4         = 2044
+        &ubyte  SPRPTR5         = 2045
+        &ubyte  SPRPTR6         = 2046
+        &ubyte  SPRPTR7         = 2047
+        &ubyte[8]  SPRPTR       = 2040      ; the 8 sprite pointers as an array.
+
+
+; ---- VIC-II 6567/6569/856x registers ----
+
+        &ubyte  SP0X            = $d000
+        &ubyte  SP0Y            = $d001
+        &ubyte  SP1X            = $d002
+        &ubyte  SP1Y            = $d003
+        &ubyte  SP2X            = $d004
+        &ubyte  SP2Y            = $d005
+        &ubyte  SP3X            = $d006
+        &ubyte  SP3Y            = $d007
+        &ubyte  SP4X            = $d008
+        &ubyte  SP4Y            = $d009
+        &ubyte  SP5X            = $d00a
+        &ubyte  SP5Y            = $d00b
+        &ubyte  SP6X            = $d00c
+        &ubyte  SP6Y            = $d00d
+        &ubyte  SP7X            = $d00e
+        &ubyte  SP7Y            = $d00f
+        &ubyte[16]  SPXY        = $d000        ; the 8 sprite X and Y registers as an array.
+        &uword[8]  SPXYW        = $d000        ; the 8 sprite X and Y registers as a combined xy word array.
+
+        &ubyte  MSIGX           = $d010
+        &ubyte  SCROLY          = $d011
+        &ubyte  RASTER          = $d012
+        &ubyte  LPENX           = $d013
+        &ubyte  LPENY           = $d014
+        &ubyte  SPENA           = $d015
+        &ubyte  SCROLX          = $d016
+        &ubyte  YXPAND          = $d017
+        &ubyte  VMCSB           = $d018
+        &ubyte  VICIRQ          = $d019
+        &ubyte  IREQMASK        = $d01a
+        &ubyte  SPBGPR          = $d01b
+        &ubyte  SPMC            = $d01c
+        &ubyte  XXPAND          = $d01d
+        &ubyte  SPSPCL          = $d01e
+        &ubyte  SPBGCL          = $d01f
+
+        &ubyte  EXTCOL          = $d020        ; border color
+        &ubyte  BGCOL0          = $d021        ; screen color
+        &ubyte  BGCOL1          = $d022
+        &ubyte  BGCOL2          = $d023
+        &ubyte  BGCOL4          = $d024
+        &ubyte  SPMC0           = $d025
+        &ubyte  SPMC1           = $d026
+        &ubyte  SP0COL          = $d027
+        &ubyte  SP1COL          = $d028
+        &ubyte  SP2COL          = $d029
+        &ubyte  SP3COL          = $d02a
+        &ubyte  SP4COL          = $d02b
+        &ubyte  SP5COL          = $d02c
+        &ubyte  SP6COL          = $d02d
+        &ubyte  SP7COL          = $d02e
+        &ubyte[8]  SPCOL        = $d027
+
+
+; ---- end of VIC-II registers ----
+
+; ---- CIA 6526 1 & 2 registers ----
+
+        &ubyte  CIA1PRA         = $DC00        ; CIA 1 DRA, keyboard column drive (and joystick control port #2)
+        &ubyte  CIA1PRB         = $DC01        ; CIA 1 DRB, keyboard row port (and joystick control port #1)
+        &ubyte  CIA1DDRA        = $DC02        ; CIA 1 DDRA, keyboard column
+        &ubyte  CIA1DDRB        = $DC03        ; CIA 1 DDRB, keyboard row
+        &ubyte  CIA1TAL         = $DC04        ; CIA 1 timer A low byte
+        &ubyte  CIA1TAH         = $DC05        ; CIA 1 timer A high byte
+        &ubyte  CIA1TBL         = $DC06        ; CIA 1 timer B low byte
+        &ubyte  CIA1TBH         = $DC07        ; CIA 1 timer B high byte
+        &ubyte  CIA1TOD10       = $DC08        ; time of day, 1/10 sec.
+        &ubyte  CIA1TODSEC      = $DC09        ; time of day, seconds
+        &ubyte  CIA1TODMMIN     = $DC0A        ; time of day, minutes
+        &ubyte  CIA1TODHR       = $DC0B        ; time of day, hours
+        &ubyte  CIA1SDR         = $DC0C        ; Serial Data Register
+        &ubyte  CIA1ICR         = $DC0D
+        &ubyte  CIA1CRA         = $DC0E
+        &ubyte  CIA1CRB         = $DC0F
+
+        &ubyte  CIA2PRA         = $DD00        ; CIA 2 DRA, serial port and video address
+        &ubyte  CIA2PRB         = $DD01        ; CIA 2 DRB, RS232 port / USERPORT
+        &ubyte  CIA2DDRA        = $DD02        ; CIA 2 DDRA, serial port and video address
+        &ubyte  CIA2DDRB        = $DD03        ; CIA 2 DDRB, RS232 port / USERPORT
+        &ubyte  CIA2TAL         = $DD04        ; CIA 2 timer A low byte
+        &ubyte  CIA2TAH         = $DD05        ; CIA 2 timer A high byte
+        &ubyte  CIA2TBL         = $DD06        ; CIA 2 timer B low byte
+        &ubyte  CIA2TBH         = $DD07        ; CIA 2 timer B high byte
+        &ubyte  CIA2TOD10       = $DD08        ; time of day, 1/10 sec.
+        &ubyte  CIA2TODSEC      = $DD09        ; time of day, seconds
+        &ubyte  CIA2TODMIN      = $DD0A        ; time of day, minutes
+        &ubyte  CIA2TODHR       = $DD0B        ; time of day, hours
+        &ubyte  CIA2SDR         = $DD0C        ; Serial Data Register
+        &ubyte  CIA2ICR         = $DD0D
+        &ubyte  CIA2CRA         = $DD0E
+        &ubyte  CIA2CRB         = $DD0F
+
+; ---- end of CIA registers ----
+
+; ---- SID 6581/8580 registers ----
+
+        &ubyte  FREQLO1         = $D400        ; channel 1 freq lo
+        &ubyte  FREQHI1         = $D401        ; channel 1 freq hi
+        &uword  FREQ1           = $D400        ; channel 1 freq (word)
+        &ubyte  PWLO1           = $D402        ; channel 1 pulse width lo (7-0)
+        &ubyte  PWHI1           = $D403        ; channel 1 pulse width hi (11-8)
+        &uword  PW1             = $D402        ; channel 1 pulse width (word)
+        &ubyte  CR1             = $D404        ; channel 1 voice control register
+        &ubyte  AD1             = $D405        ; channel 1 attack & decay
+        &ubyte  SR1             = $D406        ; channel 1 sustain & release
+        &ubyte  FREQLO2         = $D407        ; channel 2 freq lo
+        &ubyte  FREQHI2         = $D408        ; channel 2 freq hi
+        &uword  FREQ2           = $D407        ; channel 2 freq (word)
+        &ubyte  PWLO2           = $D409        ; channel 2 pulse width lo (7-0)
+        &ubyte  PWHI2           = $D40A        ; channel 2 pulse width hi (11-8)
+        &uword  PW2             = $D409        ; channel 2 pulse width (word)
+        &ubyte  CR2             = $D40B        ; channel 2 voice control register
+        &ubyte  AD2             = $D40C        ; channel 2 attack & decay
+        &ubyte  SR2             = $D40D        ; channel 2 sustain & release
+        &ubyte  FREQLO3         = $D40E        ; channel 3 freq lo
+        &ubyte  FREQHI3         = $D40F        ; channel 3 freq hi
+        &uword  FREQ3           = $D40E        ; channel 3 freq (word)
+        &ubyte  PWLO3           = $D410        ; channel 3 pulse width lo (7-0)
+        &ubyte  PWHI3           = $D411        ; channel 3 pulse width hi (11-8)
+        &uword  PW3             = $D410        ; channel 3 pulse width (word)
+        &ubyte  CR3             = $D412        ; channel 3 voice control register
+        &ubyte  AD3             = $D413        ; channel 3 attack & decay
+        &ubyte  SR3             = $D414        ; channel 3 sustain & release
+        &ubyte  FCLO            = $D415        ; filter cutoff lo (2-0)
+        &ubyte  FCHI            = $D416        ; filter cutoff hi (10-3)
+        &uword  FC              = $D415        ; filter cutoff (word)
+        &ubyte  RESFILT         = $D417        ; filter resonance and routing
+        &ubyte  MVOL            = $D418        ; filter mode and main volume control
+        &ubyte  POTX            = $D419        ; potentiometer X
+        &ubyte  POTY            = $D41A        ; potentiometer Y
+        &ubyte  OSC3            = $D41B        ; channel 3 oscillator value read
+        &ubyte  ENV3            = $D41C        ; channel 3 envelope value read
+
+; ---- end of SID registers ----
+
+
+; ---- C64 ROM kernal routines ----
+
+romsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use txt.print instead)
+romsub $E544 = CLEARSCR() clobbers(A,X,Y)                       ; clear the screen
+romsub $E566 = HOMECRSR() clobbers(A,X,Y)                       ; cursor to top left of screen
+romsub $EA31 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
+romsub $EA81 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
+romsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
+romsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
+romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
+romsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
+romsub $FF8D = VECTOR(uword userptr @ XY, ubyte dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
+romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
+romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
+romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
+romsub $FF99 = MEMTOP(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
+romsub $FF9C = MEMBOT(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
+romsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
+romsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
+romsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
+romsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
+romsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
+romsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
+romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
+romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
+romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word
+romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
+romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
+romsub $FFC0 = OPEN() clobbers(X,Y) -> ubyte @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
+romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
+romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> ubyte @Pc    ; (via 798 ($31E)) define an input channel
+romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
+romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
+romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
+romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
+romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
+romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
+romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
+romsub $FFE1 = STOP() clobbers(X) -> ubyte @ Pz, ubyte @ A      ; (via 808 ($328)) check the STOP key (and some others in A)
+romsub $FFE4 = GETIN() clobbers(X,Y) -> ubyte @Pc, ubyte @ A    ; (via 810 ($32A)) get a character
+romsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
+romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; read number of screen rows and columns
+romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, ubyte dir @ Pc) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
+romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
+
+; ---- end of C64 ROM kernal routines ----
+
+; ---- utilities -----
+
+asmsub STOP2() -> ubyte @A  {
+    ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
+    %asm {{
+        txa
+        pha
+        jsr  c64.STOP
+        beq  +
+        pla
+        tax
+        lda  #0
+        rts
++       pla
+        tax
+        lda  #1
+        rts
+    }}
+}
+
+asmsub RDTIM16() -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    %asm {{
+        stx  P8ZP_SCRATCH_REG
+        jsr  c64.RDTIM
+        pha
+        txa
+        tay
+        pla
+        ldx  P8ZP_SCRATCH_REG
+        rts
+    }}
+}
+
+
+
+; ---- C64 specific system utility routines: ----
+
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
+    ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
+    ; Also a different color scheme is chosen to identify ourselves a little.
+    ; Uppercase charset is activated, and all three registers set to 0, status flags cleared.
+    %asm {{
+        sei
+        cld
+        lda  #%00101111
+        sta  $00
+        lda  #%00100111
+        sta  $01
+        jsr  c64.IOINIT
+        jsr  c64.RESTOR
+        jsr  c64.CINT
+        lda  #6
+        sta  c64.EXTCOL
+        lda  #7
+        sta  c64.COLOR
+        lda  #0
+        sta  c64.BGCOL0
+        jsr  disable_runstop_and_charsetswitch
+        clc
+        clv
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        rts     ; no phase 2 steps on the C64
+    }}
+}
+
+asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #$80
+        sta  657    ; disable charset switching
+        lda  #239
+        sta  808    ; disable run/stop key
+        rts
+    }}
+}
+
+asmsub  set_irq(uword handler @AY, ubyte useKernal @Pc) clobbers(A)  {
+	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  _use_kernal
+		sei
+		lda  #<_irq_handler
+		sta  c64.CINV
+		lda  #>_irq_handler
+		sta  c64.CINV+1
+		cli
+		rts
+_irq_handler    jsr  _irq_handler_init
+_modified	jsr  $ffff                      ; modified
+		jsr  _irq_handler_end
+		lda  _use_kernal
+		bne  +
+		lda  #$ff
+		sta  c64.VICIRQ			; acknowledge raster irq
+		lda  c64.CIA1ICR		; acknowledge CIA1 interrupt
+		; end irq processing - don't use kernal's irq handling
+		pla
+		tay
+		pla
+		tax
+		pla
+		rti
++		jmp  c64.IRQDFRT		; continue with normal kernal irq routine
+
+_use_kernal     .byte  0
+
+_irq_handler_init
+		; save all zp scratch registers and the X register as these might be clobbered by the irq routine
+		stx  IRQ_X_REG
+		lda  P8ZP_SCRATCH_B1
+		sta  IRQ_SCRATCH_ZPB1
+		lda  P8ZP_SCRATCH_REG
+		sta  IRQ_SCRATCH_ZPREG
+		lda  P8ZP_SCRATCH_W1
+		sta  IRQ_SCRATCH_ZPWORD1
+		lda  P8ZP_SCRATCH_W1+1
+		sta  IRQ_SCRATCH_ZPWORD1+1
+		lda  P8ZP_SCRATCH_W2
+		sta  IRQ_SCRATCH_ZPWORD2
+		lda  P8ZP_SCRATCH_W2+1
+		sta  IRQ_SCRATCH_ZPWORD2+1
+		; stack protector; make sure we don't clobber the top of the evaluation stack
+		dex
+		dex
+		dex
+		dex
+		dex
+		dex
+		cld
+		rts
+
+_irq_handler_end
+		; restore all zp scratch registers and the X register
+		lda  IRQ_SCRATCH_ZPB1
+		sta  P8ZP_SCRATCH_B1
+		lda  IRQ_SCRATCH_ZPREG
+		sta  P8ZP_SCRATCH_REG
+		lda  IRQ_SCRATCH_ZPWORD1
+		sta  P8ZP_SCRATCH_W1
+		lda  IRQ_SCRATCH_ZPWORD1+1
+		sta  P8ZP_SCRATCH_W1+1
+		lda  IRQ_SCRATCH_ZPWORD2
+		sta  P8ZP_SCRATCH_W2
+		lda  IRQ_SCRATCH_ZPWORD2+1
+		sta  P8ZP_SCRATCH_W2+1
+		ldx  IRQ_X_REG
+		rts
+
+IRQ_X_REG		.byte  0
+IRQ_SCRATCH_ZPB1	.byte  0
+IRQ_SCRATCH_ZPREG	.byte  0
+IRQ_SCRATCH_ZPWORD1	.word  0
+IRQ_SCRATCH_ZPWORD2	.word  0
+
+		}}
+}
+
+asmsub  restore_irq() clobbers(A) {
+	%asm {{
+		sei
+		lda  #<c64.IRQDFRT
+		sta  c64.CINV
+		lda  #>c64.IRQDFRT
+		sta  c64.CINV+1
+		lda  #0
+		sta  c64.IREQMASK	; disable raster irq
+		lda  #%10000001
+		sta  c64.CIA1ICR	; restore CIA1 irq
+		cli
+		rts
+	}}
+}
+
+asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0, ubyte useKernal @Pc) clobbers(A) {
+	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  set_irq._use_kernal
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		sei
+		jsr  _setup_raster_irq
+		lda  #<_raster_irq_handler
+		sta  c64.CINV
+		lda  #>_raster_irq_handler
+		sta  c64.CINV+1
+		cli
+		rts
+
+_raster_irq_handler
+		jsr  set_irq._irq_handler_init
+_modified	jsr  $ffff              ; modified
+		jsr  set_irq._irq_handler_end
+                lda  #$ff
+                sta  c64.VICIRQ			; acknowledge raster irq
+		lda  set_irq._use_kernal
+		bne  +
+		; end irq processing - don't use kernal's irq handling
+		pla
+		tay
+		pla
+		tax
+		pla
+		rti
++		jmp  c64.IRQDFRT                ; continue with kernal irq routine
+
+_setup_raster_irq
+		pha
+		lda  #%01111111
+		sta  c64.CIA1ICR    ; "switch off" interrupts signals from cia-1
+		sta  c64.CIA2ICR    ; "switch off" interrupts signals from cia-2
+		and  c64.SCROLY
+		sta  c64.SCROLY     ; clear most significant bit of raster position
+		lda  c64.CIA1ICR    ; ack previous irq
+		lda  c64.CIA2ICR    ; ack previous irq
+		pla
+		sta  c64.RASTER     ; set the raster line number where interrupt should occur
+		cpy  #0
+		beq  +
+		lda  c64.SCROLY
+		ora  #%10000000
+		sta  c64.SCROLY     ; set most significant bit of raster position
++		lda  #%00000001
+		sta  c64.IREQMASK   ;enable raster interrupt signals from vic
+		rts
+	}}
+}
+
+; ---- end of C64 specific system utility routines ----
+
+}
+
+sys {
+    ; ------- lowlevel system routines --------
+
+    const ubyte target = 64         ;  compilation target specifier.  64 = C64,  16 = CommanderX16.
+
+
+    asmsub  reset_system()  {
+        ; Soft-reset the system back to Basic prompt.
+        %asm {{
+            sei
+            lda  #14
+            sta  $01        ; bank the kernal in
+            jmp  (c64.RESET_VEC)
+        }}
+    }
+
+    sub wait(uword jiffies) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds)
+        repeat jiffies {
+            ubyte jiff = lsb(c64.RDTIM16())
+            while jiff==lsb(c64.RDTIM16()) {
+                ; wait until 1 jiffy has passed
+            }
+        }
+    }
+
+    asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1        ; source in ZP
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2        ; target in ZP
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            cpy  #0
+            bne  _longcopy
+
+            ; copy <= 255 bytes
+            tay
+            bne  _copyshort
+            rts     ; nothing to copy
+
+_copyshort
+            ; decrease source and target pointers so we can simply index by Y
+            lda  P8ZP_SCRATCH_W1
+            bne  +
+            dec  P8ZP_SCRATCH_W1+1
++           dec  P8ZP_SCRATCH_W1
+            lda  P8ZP_SCRATCH_W2
+            bne  +
+            dec  P8ZP_SCRATCH_W2+1
++           dec  P8ZP_SCRATCH_W2
+
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            dey
+            bne  -
+            rts
+
+_longcopy
+            sta  P8ZP_SCRATCH_B1        ; lsb(count) = remainder in last page
+            tya
+            tax                         ; x = num pages (1+)
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dex
+            bne  -
+            ldy  P8ZP_SCRATCH_B1
+            bne  _copyshort
+            rts
+        }}
+    }
+
+    asmsub memset(uword mem @R0, uword numbytes @R1, ubyte value @A) clobbers(A,X,Y) {
+        %asm {{
+            ldy  cx16.r0
+            sty  P8ZP_SCRATCH_W1
+            ldy  cx16.r0+1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            ldy  cx16.r1+1
+            jmp  prog8_lib.memset
+        }}
+    }
+
+    asmsub memsetw(uword mem @R0, uword numwords @R1, uword value @AY) clobbers(A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            jmp  prog8_lib.memsetw
+        }}
+    }
+
+
+    inline asmsub rsave() {
+        ; save cpu status flag and all registers A, X, Y.
+        ; see http://6502.org/tutorials/register_preservation.html
+        %asm {{
+            php
+            sta  P8ZP_SCRATCH_REG
+            pha
+            txa
+            pha
+            tya
+            pha
+            lda  P8ZP_SCRATCH_REG
+        }}
+    }
+
+    inline asmsub rrestore() {
+        ; restore all registers and cpu status flag
+        %asm {{
+            pla
+            tay
+            pla
+            tax
+            pla
+            plp
+        }}
+    }
+
+    inline asmsub read_flags() -> ubyte @A {
+        %asm {{
+            php
+            pla
+        }}
+    }
+
+    inline asmsub clear_carry() {
+        %asm {{
+            clc
+        }}
+    }
+
+    inline asmsub set_carry() {
+        %asm {{
+            sec
+        }}
+    }
+
+    inline asmsub clear_irqd() {
+        %asm {{
+            cli
+        }}
+    }
+
+    inline asmsub set_irqd() {
+        %asm {{
+            sei
+        }}
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            jsr  c64.CLRCHN		; reset i/o channels
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            rts		; return to original caller
+        }}
+    }
+
+    inline asmsub progend() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_end
+            ldy  #>prog8_program_end
+        }}
+    }
+
+}
+
+cx16 {
+
+    ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
+    ; they are simulated on the C64 as well but their location in memory is different
+    ; (because there's no room for them in the zeropage)
+    ; they are allocated at the bottom of the eval-stack (should be ample space unless
+    ; you're doing insane nesting of expressions...)
+    &uword r0  = $cf00
+    &uword r1  = $cf02
+    &uword r2  = $cf04
+    &uword r3  = $cf06
+    &uword r4  = $cf08
+    &uword r5  = $cf0a
+    &uword r6  = $cf0c
+    &uword r7  = $cf0e
+    &uword r8  = $cf10
+    &uword r9  = $cf12
+    &uword r10 = $cf14
+    &uword r11 = $cf16
+    &uword r12 = $cf18
+    &uword r13 = $cf1a
+    &uword r14 = $cf1c
+    &uword r15 = $cf1e
+
+}

compiler/res/prog8lib/c64/textio.p8

@@ -0,0 +1,618 @@
+; Prog8 definitions for the Text I/O and Screen routines for the Commodore-64
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target c64
+%import syslib
+%import conv
+
+
+txt {
+
+const ubyte DEFAULT_WIDTH = 40
+const ubyte DEFAULT_HEIGHT = 25
+
+
+sub  clear_screen() {
+    txt.chrout(147)
+}
+
+sub home() {
+    txt.chrout(19)
+}
+
+sub nl() {
+    txt.chrout('\n')
+}
+
+sub spc() {
+    txt.chrout(' ')
+}
+
+asmsub column(ubyte col @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given column (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  c64.PLOT
+        tay
+        clc
+        jmp  c64.PLOT
+    }}
+}
+
+asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+	; ---- fill the character screen with the given fill character and character color.
+	;      (assumes screen and color matrix are at their default addresses)
+
+	%asm {{
+		pha
+		tya
+		jsr  clear_screencolors
+		pla
+		jsr  clear_screenchars
+		rts
+        }}
+
+}
+
+asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+	; ---- clear the character screen with the given fill character (leaves colors)
+	;      (assumes screen matrix is at the default address)
+	%asm {{
+		ldy  #250
+-		sta  c64.Screen+250*0-1,y
+		sta  c64.Screen+250*1-1,y
+		sta  c64.Screen+250*2-1,y
+		sta  c64.Screen+250*3-1,y
+		dey
+		bne  -
+		rts
+        }}
+}
+
+asmsub  clear_screencolors (ubyte color @ A) clobbers(Y)  {
+	; ---- clear the character screen colors with the given color (leaves characters).
+	;      (assumes color matrix is at the default address)
+	%asm {{
+		ldy  #250
+-		sta  c64.Colors+250*0-1,y
+		sta  c64.Colors+250*1-1,y
+		sta  c64.Colors+250*2-1,y
+		sta  c64.Colors+250*3-1,y
+		dey
+		bne  -
+		rts
+        }}
+}
+
+sub color (ubyte txtcol) {
+    c64.COLOR = txtcol
+}
+
+sub lowercase() {
+    c64.VMCSB |= 2
+}
+
+sub uppercase() {
+    c64.VMCSB &= ~2
+}
+
+asmsub  scroll_left  (ubyte alsocolors @ Pc) clobbers(A, Y)  {
+	; ---- scroll the whole screen 1 character to the left
+	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
+	;      Carry flag determines if screen color data must be scrolled too
+
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		bcc _scroll_screen
+
++               ; scroll the screen and the color memory
+		ldx  #0
+		ldy  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 1,x
+            sta  c64.Screen + 40*row + 0,x
+            lda  c64.Colors + 40*row + 1,x
+            sta  c64.Colors + 40*row + 0,x
+        .next
+		inx
+		dey
+		bpl  -
+		rts
+
+_scroll_screen  ; scroll only the screen memory
+		ldx  #0
+		ldy  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 1,x
+            sta  c64.Screen + 40*row + 0,x
+        .next
+		inx
+		dey
+		bpl  -
+
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  scroll_right  (ubyte alsocolors @ Pc) clobbers(A)  {
+	; ---- scroll the whole screen 1 character to the right
+	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
+	;      Carry flag determines if screen color data must be scrolled too
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		bcc  _scroll_screen
+
++               ; scroll the screen and the color memory
+		ldx  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 0,x
+            sta  c64.Screen + 40*row + 1,x
+            lda  c64.Colors + 40*row + 0,x
+            sta  c64.Colors + 40*row + 1,x
+        .next
+		dex
+		bpl  -
+		rts
+
+_scroll_screen  ; scroll only the screen memory
+		ldx  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 0,x
+            sta  c64.Screen + 40*row + 1,x
+        .next
+		dex
+		bpl  -
+
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  scroll_up  (ubyte alsocolors @ Pc) clobbers(A)  {
+	; ---- scroll the whole screen 1 character up
+	;      contents of the bottom row are unchanged, you should refill/clear this yourself
+	;      Carry flag determines if screen color data must be scrolled too
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		bcc  _scroll_screen
+
++               ; scroll the screen and the color memory
+		ldx #39
+-
+        .for row=1, row<=24, row+=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row-1),x
+            lda  c64.Colors + 40*row,x
+            sta  c64.Colors + 40*(row-1),x
+        .next
+		dex
+		bpl  -
+		rts
+
+_scroll_screen  ; scroll only the screen memory
+		ldx #39
+-
+        .for row=1, row<=24, row+=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row-1),x
+        .next
+		dex
+		bpl  -
+
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  scroll_down  (ubyte alsocolors @ Pc) clobbers(A)  {
+	; ---- scroll the whole screen 1 character down
+	;      contents of the top row are unchanged, you should refill/clear this yourself
+	;      Carry flag determines if screen color data must be scrolled too
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		bcc  _scroll_screen
+
++               ; scroll the screen and the color memory
+		ldx #39
+-
+        .for row=23, row>=0, row-=1
+            lda  c64.Colors + 40*row,x
+            sta  c64.Colors + 40*(row+1),x
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row+1),x
+        .next
+		dex
+		bpl  -
+		rts
+
+_scroll_screen  ; scroll only the screen memory
+		ldx #39
+-
+        .for row=23, row>=0, row-=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row+1),x
+        .next
+		dex
+		bpl  -
+
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use c64.CHROUT directly ofcourse.
+
+asmsub  print (str text @ AY) clobbers(A,Y)  {
+	; ---- print null terminated string from A/Y
+	; note: the compiler contains an optimization that will replace
+	;       a call to this subroutine with a string argument of just one char,
+	;       by just one call to c64.CHROUT of that single char.
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		sty  P8ZP_SCRATCH_REG
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_B1),y
+		beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		jsr  conv.ubyte2decimal
+		pha
+		tya
+		jsr  c64.CHROUT
+		pla
+		jsr  c64.CHROUT
+		txa
+		jsr  c64.CHROUT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- print the ubyte in A in decimal form, without left padding 0s
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		jsr  conv.ubyte2decimal
+_print_byte_digits
+		pha
+		cpy  #'0'
+		beq  +
+		tya
+		jsr  c64.CHROUT
+		pla
+		jsr  c64.CHROUT
+		jmp  _ones
++       pla
+        cmp  #'0'
+        beq  _ones
+        jsr  c64.CHROUT
+_ones   txa
+		jsr  c64.CHROUT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+	; ---- print the byte in A in decimal form, without left padding 0s
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		pha
+		cmp  #0
+		bpl  +
+		lda  #'-'
+		jsr  c64.CHROUT
++		pla
+		jsr  conv.byte2decimal
+		jmp  print_ub._print_byte_digits
+	}}
+}
+
+asmsub  print_ubhex  (ubyte value @ A, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		bcc  +
+		pha
+		lda  #'$'
+		jsr  c64.CHROUT
+		pla
++		jsr  conv.ubyte2hex
+		jsr  c64.CHROUT
+		tya
+		jsr  c64.CHROUT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  print_ubbin  (ubyte value @ A, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'%'
+		jsr  c64.CHROUT
++		ldy  #8
+-		lda  #'0'
+		asl  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'1'
++		jsr  c64.CHROUT
+		dey
+		bne  -
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  print_uwbin  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubbin
+		pla
+		clc
+		jmp  print_ubbin
+	}}
+}
+
+asmsub  print_uwhex  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in hexadecimal form (4 digits)
+	;      (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubhex
+		pla
+		clc
+		jmp  print_ubhex
+	}}
+}
+
+asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
+	%asm {{
+	    stx  P8ZP_SCRATCH_REG
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+        beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in decimal form, without left padding 0s
+	%asm {{
+	    stx  P8ZP_SCRATCH_REG
+		jsr  conv.uword2decimal
+		ldx  P8ZP_SCRATCH_REG
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+		beq  _allzero
+		cmp  #'0'
+		bne  _gotdigit
+		iny
+		bne  -
+
+_gotdigit
+		jsr  c64.CHROUT
+		iny
+		lda  conv.uword2decimal.decTenThousands,y
+		bne  _gotdigit
+		rts
+_allzero
+        lda  #'0'
+        jmp  c64.CHROUT
+	}}
+}
+
+asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
+	%asm {{
+		cpy  #0
+		bpl  +
+		pha
+		lda  #'-'
+		jsr  c64.CHROUT
+		tya
+		eor  #255
+		tay
+		pla
+		eor  #255
+		clc
+		adc  #1
+		bcc  +
+		iny
++		jmp  print_uw
+	}}
+}
+
+asmsub  input_chars  (uword buffer @ AY) clobbers(A) -> ubyte @ Y  {
+	; ---- Input a string (max. 80 chars) from the keyboard. Returns length in Y. (string is terminated with a 0 byte as well)
+	;      It assumes the keyboard is selected as I/O channel!
+
+	%asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0				; char counter = 0
+-		jsr  c64.CHRIN
+		cmp  #$0d			; return (ascii 13) pressed?
+		beq  +				; yes, end.
+		sta  (P8ZP_SCRATCH_W1),y	; else store char in buffer
+		iny
+		bne  -
++		lda  #0
+		sta  (P8ZP_SCRATCH_W1),y	; finish string with 0 byte
+		rts
+
+	}}
+}
+
+asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)  {
+	; ---- sets the character in the screen matrix at the given position
+	%asm {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  _screenrows+1,y
+		sta  _mod+2
+		txa
+		clc
+		adc  _screenrows,y
+		sta  _mod+1
+		bcc  +
+		inc  _mod+2
++		pla
+_mod		sta  $ffff		; modified
+		rts
+
+_screenrows	.word  $0400 + range(0, 1000, 40)
+	}}
+}
+
+asmsub  getchr  (ubyte col @A, ubyte row @Y) clobbers(Y) -> ubyte @ A {
+	; ---- get the character in the screen matrix at the given location
+	%asm  {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  setchr._screenrows+1,y
+		sta  _mod+2
+		pla
+		clc
+		adc  setchr._screenrows,y
+		sta  _mod+1
+		bcc  _mod
+		inc  _mod+2
+_mod		lda  $ffff		; modified
+		rts
+	}}
+}
+
+asmsub  setclr  (ubyte col @X, ubyte row @Y, ubyte color @A) clobbers(A, Y)  {
+	; ---- set the color in A on the screen matrix at the given position
+	%asm {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  _colorrows+1,y
+		sta  _mod+2
+		txa
+		clc
+		adc  _colorrows,y
+		sta  _mod+1
+		bcc  +
+		inc  _mod+2
++		pla
+_mod		sta  $ffff		; modified
+		rts
+
+_colorrows	.word  $d800 + range(0, 1000, 40)
+	}}
+}
+
+asmsub  getclr  (ubyte col @A, ubyte row @Y) clobbers(Y) -> ubyte @ A {
+	; ---- get the color in the screen color matrix at the given location
+	%asm  {{
+		pha
+		tya
+		asl  a
+		tay
+		lda  setclr._colorrows+1,y
+		sta  _mod+2
+		pla
+		clc
+		adc  setclr._colorrows,y
+		sta  _mod+1
+		bcc  _mod
+		inc  _mod+2
+_mod		lda  $ffff		; modified
+		rts
+	}}
+}
+
+sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+	; ---- set char+color at the given position on the screen
+	%asm {{
+		lda  row
+		asl  a
+		tay
+		lda  setchr._screenrows+1,y
+		sta  _charmod+2
+		adc  #$d4
+		sta  _colormod+2
+		lda  setchr._screenrows,y
+		clc
+		adc  column
+		sta  _charmod+1
+		sta  _colormod+1
+		bcc  +
+		inc  _charmod+2
+		inc  _colormod+2
++		lda  char
+_charmod	sta  $ffff		; modified
+		lda  charcolor
+_colormod	sta  $ffff		; modified
+		rts
+	}}
+}
+
+asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
+	%asm  {{
+		stx  P8ZP_SCRATCH_REG
+		tax
+		clc
+		jsr  c64.PLOT
+		ldx  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+asmsub width() clobbers(X,Y) -> ubyte @A {
+    ; -- returns the text screen width (number of columns)
+    %asm {{
+        jsr  c64.SCREEN
+        txa
+        rts
+    }}
+}
+
+asmsub height() clobbers(X, Y) -> ubyte @A {
+    ; -- returns the text screen height (number of rows)
+    %asm {{
+        jsr  c64.SCREEN
+        tya
+        rts
+    }}
+}
+
+}

compiler/res/prog8lib/c64flt.p8

@@ -1,945 +0,0 @@
-; Prog8 definitions for floating point handling on the Commodore-64
-;
-; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
-
-%option enable_floats
-
-
-~ c64flt {
-	; ---- this block contains C-64 floating point related functions ----
-	
-		const  float  PI	= 3.141592653589793
-		const  float  TWOPI	= 6.283185307179586
-
-	
-; ---- C64 basic and kernal ROM float constants and functions ----
-
-		; note: the fac1 and fac2 are working registers and take 6 bytes each,
-		; floats in memory  (and rom) are stored in 5-byte MFLPT packed format.
-
-		; constants in five-byte "mflpt" format in the BASIC ROM
-		memory  float  FL_PIVAL		= $aea8  ; 3.1415926...
-		memory  float  FL_N32768	= $b1a5  ; -32768
-		memory  float  FL_FONE		= $b9bc  ; 1
-		memory  float  FL_SQRHLF	= $b9d6  ; SQR(2) / 2
-		memory  float  FL_SQRTWO	= $b9db  ; SQR(2)
-		memory  float  FL_NEGHLF	= $b9e0  ; -.5
-		memory  float  FL_LOG2		= $b9e5  ; LOG(2)
-		memory  float  FL_TENC		= $baf9  ; 10
-		memory  float  FL_NZMIL		= $bdbd  ; 1e9 (1 billion)
-		memory  float  FL_FHALF		= $bf11  ; .5
-		memory  float  FL_LOGEB2	= $bfbf  ; 1 / LOG(2)
-		memory  float  FL_PIHALF	= $e2e0  ; PI / 2
-		memory  float  FL_TWOPI		= $e2e5  ; 2 * PI
-		memory  float  FL_FR4		= $e2ea  ; .25
-
-
-; note: fac1/2 might get clobbered even if not mentioned in the function's name.
-; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
-
-; checked functions below:
-asmsub	MOVFM		(uword mflpt @ AY) -> clobbers(A,Y) -> ()	= $bba2		; load mflpt value from memory  in A/Y into fac1
-asmsub	FREADMEM	() -> clobbers(A,Y) -> ()			= $bba6		; load mflpt value from memory  in $22/$23 into fac1
-asmsub	CONUPK		(uword mflpt @ AY) -> clobbers(A,Y) -> ()	= $ba8c		; load mflpt value from memory  in A/Y into fac2
-asmsub	FAREADMEM	() -> clobbers(A,Y) -> ()			= $ba90		; load mflpt value from memory  in $22/$23 into fac2
-asmsub	MOVFA		() -> clobbers(A,X) -> ()			= $bbfc		; copy fac2 to fac1
-asmsub	MOVAF		() -> clobbers(A,X) -> ()			= $bc0c		; copy fac1 to fac2  (rounded)
-asmsub	MOVEF		() -> clobbers(A,X) -> ()			= $bc0f		; copy fac1 to fac2
-asmsub	MOVMF		(uword mflpt @ XY) -> clobbers(A,Y) -> ()	= $bbd4		; store fac1 to memory  X/Y as 5-byte mflpt
-
-; fac1-> signed word in Y/A (might throw ILLEGAL QUANTITY)
-; (tip: use c64flt.FTOSWRDAY to get A/Y output; lo/hi switched to normal little endian order)
-asmsub	FTOSWORDYA	() -> clobbers(X) -> (ubyte @ Y, ubyte @ A)	= $b1aa
-
-; fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
-; (tip: use c64flt.GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
-asmsub	GETADR		() -> clobbers(X) -> (ubyte @ Y, ubyte @ A)	= $b7f7
-
-asmsub	QINT		() -> clobbers(A,X,Y) -> ()			= $bc9b		; fac1 -> 4-byte signed integer in 98-101 ($62-$65), with the MSB FIRST.
-asmsub	AYINT		() -> clobbers(A,X,Y) -> ()			= $b1bf		; fac1-> signed word in 100-101 ($64-$65) MSB FIRST. (might throw ILLEGAL QUANTITY)
-
-; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1
-; (tip: use c64flt.GIVAYFAY to use A/Y input; lo/hi switched to normal order)
-; there is also c64flt.GIVUAYFAY - unsigned word in A/Y (lo/hi) to fac1
-; there is also c64flt.FREADS32  that reads from 98-101 ($62-$65) MSB FIRST
-; there is also c64flt.FREADUS32  that reads from 98-101 ($62-$65) MSB FIRST
-; there is also c64flt.FREADS24AXY  that reads signed int24 into fac1 from A/X/Y (lo/mid/hi bytes)
-asmsub	GIVAYF		(ubyte lo @ Y, ubyte hi @ A) -> clobbers(A,X,Y) -> ()	= $b391
-
-asmsub	FREADUY		(ubyte value @ Y) -> clobbers(A,X,Y) -> ()	= $b3a2		; 8 bit unsigned Y -> float in fac1
-asmsub	FREADSA		(byte value @ A) -> clobbers(A,X,Y) -> ()	= $bc3c		; 8 bit signed A -> float in fac1
-asmsub	FREADSTR	(ubyte length @ A) -> clobbers(A,X,Y) -> ()	= $b7b5		; str -> fac1, $22/23 must point to string, A=string length
-asmsub	FPRINTLN	() -> clobbers(A,X,Y) -> ()			= $aabc		; print string of fac1, on one line (= with newline) destroys fac1.  (consider FOUT + STROUT as well)
-asmsub	FOUT		() -> clobbers(X) -> (uword @ AY)		= $bddd		; fac1 -> string, address returned in AY ($0100)
-
-asmsub	FADDH		() -> clobbers(A,X,Y) -> ()			= $b849		; fac1 += 0.5, for rounding- call this before INT
-asmsub	MUL10		() -> clobbers(A,X,Y) -> ()			= $bae2		; fac1 *= 10
-asmsub	DIV10		() -> clobbers(A,X,Y) -> ()			= $bafe		; fac1 /= 10 , CAUTION: result is always positive!
-asmsub	FCOMP		(uword mflpt @ AY) -> clobbers(X,Y) -> (ubyte @ A) = $bc5b		; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
-
-asmsub	FADDT		() -> clobbers(A,X,Y) -> ()			= $b86a		; fac1 += fac2
-asmsub	FADD		(uword mflpt @ AY) -> clobbers(A,X,Y) -> ()	= $b867		; fac1 += mflpt value from A/Y
-asmsub	FSUBT		() -> clobbers(A,X,Y) -> ()			= $b853		; fac1 = fac2-fac1   mind the order of the operands
-asmsub	FSUB		(uword mflpt @ AY) -> clobbers(A,X,Y) -> ()	= $b850		; fac1 = mflpt from A/Y - fac1
-asmsub	FMULTT 		() -> clobbers(A,X,Y) -> ()			= $ba2b		; fac1 *= fac2
-asmsub	FMULT		(uword mflpt @ AY) -> clobbers(A,X,Y) -> ()	= $ba28		; fac1 *= mflpt value from A/Y
-asmsub	FDIVT 		() -> clobbers(A,X,Y) -> ()			= $bb12		; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
-asmsub	FDIV  		(uword mflpt @ AY) -> clobbers(A,X,Y) -> ()	= $bb0f		; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
-asmsub	FPWRT		() -> clobbers(A,X,Y) -> ()			= $bf7b		; fac1 = fac2 ** fac1
-asmsub	FPWR		(uword mflpt @ AY) -> clobbers(A,X,Y) -> ()	= $bf78		; fac1 = fac2 ** mflpt from A/Y
-
-asmsub	NOTOP		() -> clobbers(A,X,Y) -> ()			= $aed4		; fac1 = NOT(fac1)
-asmsub	INT		() -> clobbers(A,X,Y) -> ()			= $bccc		; INT() truncates, use FADDH first to round instead of trunc
-asmsub	LOG		() -> clobbers(A,X,Y) -> ()			= $b9ea		; fac1 = LN(fac1)  (natural log)
-asmsub	SGN		() -> clobbers(A,X,Y) -> ()			= $bc39		; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
-asmsub	SIGN		() -> clobbers() -> (ubyte @ A)			= $bc2b		; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
-asmsub	ABS		() -> clobbers() -> ()				= $bc58		; fac1 = ABS(fac1)
-asmsub	SQR		() -> clobbers(A,X,Y) -> ()			= $bf71		; fac1 = SQRT(fac1)
-asmsub	SQRA		() -> clobbers(A,X,Y) -> ()			= $bf74		; fac1 = SQRT(fac2)
-asmsub	EXP		() -> clobbers(A,X,Y) -> ()			= $bfed		; fac1 = EXP(fac1)  (e ** fac1)
-asmsub	NEGOP		() -> clobbers(A) -> ()				= $bfb4		; switch the sign of fac1
-asmsub	RND		() -> clobbers(A,X,Y) -> ()			= $e097		; fac1 = RND(fac1) float random number generator
-asmsub	COS		() -> clobbers(A,X,Y) -> ()			= $e264		; fac1 = COS(fac1)
-asmsub	SIN		() -> clobbers(A,X,Y) -> ()			= $e26b		; fac1 = SIN(fac1)
-asmsub	TAN		() -> clobbers(A,X,Y) -> ()			= $e2b4		; fac1 = TAN(fac1)
-asmsub	ATN		() -> clobbers(A,X,Y) -> ()			= $e30e		; fac1 = ATN(fac1)
-
-
-
-
-asmsub  FREADS32  () -> clobbers(A,X,Y) -> ()  {
-	; ---- fac1 = signed int32 from $62-$65 big endian (MSB FIRST)
-	%asm {{
-		lda  $62
-		eor  #$ff
-		asl  a
-		lda  #0
-		ldx  #$a0
-		jmp  $bc4f		; internal BASIC routine
-	}}
-}
-
-asmsub  FREADUS32  () -> clobbers(A,X,Y) -> ()  {
-	; ---- fac1 = uint32 from $62-$65 big endian (MSB FIRST)
-	%asm {{
-		sec
-		lda  #0
-		ldx  #$a0
-		jmp  $bc4f		; internal BASIC routine
-	}}
-}
-
-asmsub  FREADS24AXY  (ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) -> clobbers(A,X,Y) -> ()  {
-	; ---- fac1 = signed int24 (A/X/Y contain lo/mid/hi bytes)
-	;      note: there is no FREADU24AXY (unsigned), use FREADUS32 instead.
-	%asm {{
-		sty  $62
-		stx  $63
-		sta  $64
-		lda  $62
-		eor  #$FF
-		asl  a
-		lda  #0
-		sta  $65
-		ldx  #$98
-		jmp  $bc4f		; internal BASIC routine
-	}}
-}
-
-asmsub  GIVUAYFAY  (uword value @ AY) -> clobbers(A,X,Y) -> ()  {
-	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
-	%asm {{
-		sty  $62
-		sta  $63
-		ldx  #$90
-		sec
-		jmp  $bc49		; internal BASIC routine
-	}}
-}
-
-asmsub  GIVAYFAY  (uword value @ AY) -> clobbers(A,X,Y) -> ()  {
-	; ---- signed 16 bit word in A/Y (lo/hi) to float in fac1
-	%asm {{
-		sta  c64.SCRATCH_ZPREG
-		tya
-		ldy  c64.SCRATCH_ZPREG
-		jmp  c64flt.GIVAYF		; this uses the inverse order, Y/A
-	}}
-}
-
-asmsub  FTOSWRDAY  () -> clobbers(X) -> (uword @ AY)  {
-	; ---- fac1 to signed word in A/Y
-	%asm {{
-		jsr  c64flt.FTOSWORDYA	; note the inverse Y/A order
-		sta  c64.SCRATCH_ZPREG
-		tya
-		ldy  c64.SCRATCH_ZPREG
-		rts
-	}}
-}
-
-asmsub  GETADRAY  () -> clobbers(X) -> (uword @ AY)  {
-	; ---- fac1 to unsigned word in A/Y
-	%asm {{
-		jsr  c64flt.GETADR		; this uses the inverse order, Y/A
-		sta  c64.SCRATCH_ZPB1
-		tya
-		ldy  c64.SCRATCH_ZPB1
-		rts
-	}}
-}
-
-sub  print_f  (float value) {
-	; ---- prints the floating point value (without a newline) using basic rom routines. 
-	%asm {{
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<print_f_value
-		ldy  #>print_f_value
-		jsr  c64flt.MOVFM		; load float into fac1
-		jsr  c64flt.FOUT		; fac1 to string in A/Y
-		jsr  c64.STROUT			; print string in A/Y
-		ldx  c64.SCRATCH_ZPREGX
-		rts
-	}}
-}
-
-sub  print_fln  (float value) {
-	; ---- prints the floating point value (with a newline at the end) using basic rom routines
-	%asm {{
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<print_fln_value
-		ldy  #>print_fln_value
-		jsr  c64flt.MOVFM		; load float into fac1
-		jsr  c64flt.FPRINTLN		; print fac1 with newline
-		ldx  c64.SCRATCH_ZPREGX
-		rts
-	}}
-        
-}
-
-
-; --- low level floating point assembly routines
-%asm {{
-ub2float	.proc
-		; -- convert ubyte in SCRATCH_ZPB1 to float at address A/Y
-		;    clobbers A, Y
-		stx  c64.SCRATCH_ZPREGX
-		sta  c64.SCRATCH_ZPWORD2
-		sty  c64.SCRATCH_ZPWORD2+1
-		ldy  c64.SCRATCH_ZPB1
-		jsr  c64flt.FREADUY
-_fac_to_mem	ldx  c64.SCRATCH_ZPWORD2
-		ldy  c64.SCRATCH_ZPWORD2+1
-		jsr  c64flt.MOVMF
-		ldx  c64.SCRATCH_ZPREGX
-		rts
-		.pend
-
-b2float		.proc
-		; -- convert byte in SCRATCH_ZPB1 to float at address A/Y
-		;    clobbers A, Y
-		stx  c64.SCRATCH_ZPREGX
-		sta  c64.SCRATCH_ZPWORD2
-		sty  c64.SCRATCH_ZPWORD2+1
-		lda  c64.SCRATCH_ZPB1
-		jsr  c64flt.FREADSA
-		jmp  ub2float._fac_to_mem
-		.pend
-
-uw2float	.proc
-		; -- convert uword in SCRATCH_ZPWORD1 to float at address A/Y
-		stx  c64.SCRATCH_ZPREGX
-		sta  c64.SCRATCH_ZPWORD2
-		sty  c64.SCRATCH_ZPWORD2+1
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.GIVUAYFAY
-		jmp  ub2float._fac_to_mem
-		.pend
-
-w2float		.proc
-		; -- convert word in SCRATCH_ZPWORD1 to float at address A/Y
-		stx  c64.SCRATCH_ZPREGX
-		sta  c64.SCRATCH_ZPWORD2
-		sty  c64.SCRATCH_ZPWORD2+1
-		ldy  c64.SCRATCH_ZPWORD1
-		lda  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.GIVAYF
-		jmp  ub2float._fac_to_mem
-		.pend
-		
-stack_b2float	.proc
-		; -- b2float operating on the stack
-		inx
-		lda  c64.ESTACK_LO,x
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.FREADSA
-		jmp  push_fac1_as_result
-		.pend
-		
-stack_w2float	.proc
-		; -- w2float operating on the stack
-		inx
-		ldy  c64.ESTACK_LO,x
-		lda  c64.ESTACK_HI,x
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.GIVAYF
-		jmp  push_fac1_as_result
-		.pend
-
-stack_ub2float	.proc
-		; -- ub2float operating on the stack
-		inx
-		lda  c64.ESTACK_LO,x
-		stx  c64.SCRATCH_ZPREGX
-		tay
-		jsr  c64flt.FREADUY
-		jmp  push_fac1_as_result
-		.pend
-
-stack_uw2float	.proc
-		; -- uw2float operating on the stack
-		inx
-		lda  c64.ESTACK_LO,x
-		ldy  c64.ESTACK_HI,x
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.GIVUAYFAY
-		jmp  push_fac1_as_result
-		.pend
-		
-stack_float2w	.proc	
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.AYINT
-		ldx  c64.SCRATCH_ZPREGX
-		lda  $64
-		sta  c64.ESTACK_HI,x
-		lda  $65
-		sta  c64.ESTACK_LO,x
-		dex
-		rts
-		.pend
-		
-stack_float2uw	.proc	
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.GETADR
-		ldx  c64.SCRATCH_ZPREGX
-		sta  c64.ESTACK_HI,x
-		tya
-		sta  c64.ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-push_float	.proc
-		; ---- push mflpt5 in A/Y onto stack 
-		; (taking 3 stack positions = 6 bytes of which 1 is padding)
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		ldy  #0
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  c64.ESTACK_LO,x
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  c64.ESTACK_HI,x
-		dex
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  c64.ESTACK_LO,x
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  c64.ESTACK_HI,x
-		dex
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  c64.ESTACK_LO,x
-		dex
-		rts
-		.pend
-		
-func_rndf	.proc
-		; -- put a random floating point value on the stack
-		stx  c64.SCRATCH_ZPREG
-		lda  #1
-		jsr  c64flt.FREADSA
-		jsr  c64flt.RND		; rng into fac1
-		ldx  #<_rndf_rnum5
-		ldy  #>_rndf_rnum5
-		jsr  c64flt.MOVMF	; fac1 to mem X/Y
-		ldx  c64.SCRATCH_ZPREG
-		lda  #<_rndf_rnum5
-		ldy  #>_rndf_rnum5
-		jmp  push_float
-_rndf_rnum5	.byte  0,0,0,0,0
-		.pend
-		
-push_float_from_indexed_var	.proc
-		; -- push the float from the array at A/Y with index on stack, onto the stack.
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		jsr  prog8_lib.pop_index_times_5
-		jsr  prog8_lib.add_a_to_zpword
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jmp  push_float
-		.pend
-
-pop_float	.proc
-		; ---- pops mflpt5 from stack to memory A/Y
-		; (frees 3 stack positions = 6 bytes of which 1 is padding)
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		ldy  #4
-		inx
-		lda  c64.ESTACK_LO,x
-		sta  (c64.SCRATCH_ZPWORD1),y
-		dey
-		inx
-		lda  c64.ESTACK_HI,x
-		sta  (c64.SCRATCH_ZPWORD1),y
-		dey
-		lda  c64.ESTACK_LO,x
-		sta  (c64.SCRATCH_ZPWORD1),y
-		dey
-		inx
-		lda  c64.ESTACK_HI,x
-		sta  (c64.SCRATCH_ZPWORD1),y
-		dey
-		lda  c64.ESTACK_LO,x
-		sta  (c64.SCRATCH_ZPWORD1),y
-		rts
-		.pend
-		
-pop_float_fac1	.proc
-		; -- pops float from stack into FAC1
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jmp  c64flt.MOVFM
-		.pend
-		
-pop_float_to_indexed_var	.proc
-		; -- pop the float on the stack, to the memory in the array at A/Y indexed by the byte on stack
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		jsr  prog8_lib.pop_index_times_5
-		jsr  prog8_lib.add_a_to_zpword
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jmp  pop_float
-		.pend
-
-copy_float	.proc
-		; -- copies the 5 bytes of the mflt value pointed to by SCRATCH_ZPWORD1, 
-		;    into the 5 bytes pointed to by A/Y.  Clobbers A,Y.
-		sta  c64.SCRATCH_ZPWORD2
-		sty  c64.SCRATCH_ZPWORD2+1
-		ldy  #0
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  (c64.SCRATCH_ZPWORD2),y
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  (c64.SCRATCH_ZPWORD2),y
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  (c64.SCRATCH_ZPWORD2),y
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  (c64.SCRATCH_ZPWORD2),y
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		sta  (c64.SCRATCH_ZPWORD2),y
-		rts
-		.pend
-
-inc_var_f	.proc
-		; -- add 1 to float pointed to by A/Y
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.MOVFM
-		lda  #<FL_FONE
-		ldy  #>FL_FONE
-		jsr  c64flt.FADD
-		ldx  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.MOVMF
-		ldx  c64.SCRATCH_ZPREGX
-		rts
-		.pend
-                
-dec_var_f	.proc
-		; -- subtract 1 from float pointed to by A/Y
-		sta  c64.SCRATCH_ZPWORD1
-		sty  c64.SCRATCH_ZPWORD1+1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<FL_FONE
-		ldy  #>FL_FONE
-		jsr  c64flt.MOVFM
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.FSUB
-		ldx  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.MOVMF
-		ldx  c64.SCRATCH_ZPREGX
-		rts
-		.pend
-		
-inc_indexed_var_f	.proc
-		; -- add 1 to float in array pointed to by A/Y, at index X
-		pha
-		txa
-		sta  c64.SCRATCH_ZPB1
-		asl  a
-		asl  a
-		clc
-		adc  c64.SCRATCH_ZPB1
-		sta  c64.SCRATCH_ZPB1
-		pla
-		clc
-		adc  c64.SCRATCH_ZPB1
-		bcc  +
-		iny
-+		jmp  inc_var_f
-		.pend
-		
-dec_indexed_var_f	.proc
-		; -- subtract 1 to float in array pointed to by A/Y, at index X
-		pha
-		txa
-		sta  c64.SCRATCH_ZPB1
-		asl  a
-		asl  a
-		clc
-		adc  c64.SCRATCH_ZPB1
-		sta  c64.SCRATCH_ZPB1
-		pla
-		clc
-		adc  c64.SCRATCH_ZPB1
-		bcc  +
-		iny
-+		jmp  dec_var_f
-		.pend
-		
-
-pop_2_floats_f2_in_fac1	.proc
-		; -- pop 2 floats from stack, load the second one in FAC1 as well
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jmp  c64flt.MOVFM
-		.pend
-		
-fmath_float1	.byte 0,0,0,0,0	; storage for a mflpt5 value
-fmath_float2	.byte 0,0,0,0,0	; storage for a mflpt5 value
-
-push_fac1_as_result	.proc
-		; -- push the float in FAC1 onto the stack, and return from calculation
-		ldx  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.MOVMF
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		ldx  c64.SCRATCH_ZPREGX
-		jmp  push_float
-		.pend
-		
-
-div_f		.proc
-		; -- push f1/f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.FDIV
-		jmp  push_fac1_as_result
-		.pend
-
-add_f		.proc
-		; -- push f1+f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.FADD
-		jmp  push_fac1_as_result
-		.pend
-
-sub_f		.proc
-		; -- push f1-f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.FSUB
-		jmp  push_fac1_as_result
-		.pend
-
-mul_f		.proc
-		; -- push f1*f2 on stack
-		jsr  pop_2_floats_f2_in_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.FMULT
-		jmp  push_fac1_as_result
-		.pend
-		
-neg_f		.proc
-		; -- push -flt back on stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.NEGOP
-		jmp  push_fac1_as_result
-		.pend
-
-abs_f		.proc
-		; -- push abs(float) on stack (as float)
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.ABS
-		jmp  push_fac1_as_result
-		.pend
-
-equal_f		.proc
-		; -- are the two mflpt5 numbers on the stack identical?
-		inx
-		inx
-		inx
-		inx
-		lda  c64.ESTACK_LO-3,x
-		cmp  c64.ESTACK_LO,x
-		bne  _equals_false
-		lda  c64.ESTACK_LO-2,x
-		cmp  c64.ESTACK_LO+1,x
-		bne  _equals_false
-		lda  c64.ESTACK_LO-1,x
-		cmp  c64.ESTACK_LO+2,x
-		bne  _equals_false
-		lda  c64.ESTACK_HI-2,x
-		cmp  c64.ESTACK_HI+1,x
-		bne  _equals_false
-		lda  c64.ESTACK_HI-1,x
-		cmp  c64.ESTACK_HI+2,x
-		bne  _equals_false
-_equals_true	lda  #1
-_equals_store	inx
-		sta  c64.ESTACK_LO+1,x
-		rts
-_equals_false	lda  #0
-		beq  _equals_store		
-		.pend
-
-notequal_f	.proc
-		; -- are the two mflpt5 numbers on the stack different?
-		jsr  equal_f
-		eor  #1		; invert the result
-		sta  c64.ESTACK_LO+1,x
-		rts
-		.pend
-
-less_f		.proc
-		; -- is f1 < f2?
-		jsr  compare_floats
-		cmp  #255
-		beq  compare_floats._return_true
-		bne  compare_floats._return_false
-		.pend
-		
-
-lesseq_f	.proc
-		; -- is f1 <= f2?
-		jsr  compare_floats
-		cmp  #255
-		beq  compare_floats._return_true
-		cmp  #0
-		beq  compare_floats._return_true
-		bne  compare_floats._return_false
-		.pend
-
-greater_f	.proc
-		; -- is f1 > f2?
-		jsr  compare_floats
-		cmp  #1
-		beq  compare_floats._return_true
-		bne  compare_floats._return_false
-		.pend
-
-greatereq_f	.proc
-		; -- is f1 >= f2?
-		jsr  compare_floats
-		cmp  #1
-		beq  compare_floats._return_true
-		cmp  #0
-		beq  compare_floats._return_true
-		bne  compare_floats._return_false
-		.pend
-
-compare_floats	.proc
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  pop_float
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.MOVFM		; fac1 = flt1
-		lda  #<fmath_float2
-		ldy  #>fmath_float2
-		stx  c64.SCRATCH_ZPREG
-		jsr  c64flt.FCOMP		; A = flt1 compared with flt2 (0=equal, 1=flt1>flt2, 255=flt1<flt2)
-		ldx  c64.SCRATCH_ZPREG
-		rts
-_return_false	lda  #0
-_return_result  sta  c64.ESTACK_LO,x
-		dex
-		rts
-_return_true	lda  #1
-		bne  _return_result
-		.pend		
-
-func_sin	.proc
-		; -- push sin(f) back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.SIN
-		jmp  push_fac1_as_result
-		.pend
-
-func_cos	.proc
-		; -- push cos(f) back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.COS
-		jmp  push_fac1_as_result
-		.pend
-
-func_tan	.proc
-		; -- push tan(f) back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.TAN
-		jmp  push_fac1_as_result
-		.pend
-		
-func_atan	.proc
-		; -- push atan(f) back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.ATN
-		jmp  push_fac1_as_result
-		.pend
-		
-func_ln		.proc
-		; -- push ln(f) back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.LOG
-		jmp  push_fac1_as_result
-		.pend
-		
-func_log2	.proc
-		; -- push log base 2, ln(f)/ln(2), back onto stack
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.LOG
-		jsr  c64flt.MOVEF
-		lda  #<c64.FL_LOG2
-		ldy  #>c64.FL_LOG2
-		jsr  c64flt.MOVFM
-		jsr  c64flt.FDIVT
-		jmp  push_fac1_as_result
-		.pend
-		
-func_sqrt	.proc
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.SQR
-		jmp  push_fac1_as_result
-		.pend
-		
-func_rad	.proc
-		; -- convert degrees to radians (d * pi / 180)
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<_pi_div_180
-		ldy  #>_pi_div_180
-		jsr  c64flt.FMULT
-		jmp  push_fac1_as_result
-_pi_div_180	.byte 123, 14, 250, 53, 18		; pi / 180
-		.pend
-		
-func_deg	.proc
-		; -- convert radians to degrees (d * (1/ pi * 180))
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<_one_over_pi_div_180
-		ldy  #>_one_over_pi_div_180
-		jsr  c64flt.FMULT
-		jmp  push_fac1_as_result
-_one_over_pi_div_180	.byte 134, 101, 46, 224, 211		; 1 / (pi * 180)
-		.pend
-		
-func_round	.proc
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.FADDH
-		jsr  c64flt.INT
-		jmp  push_fac1_as_result
-		.pend
-		
-func_floor	.proc
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		jsr  c64flt.INT
-		jmp  push_fac1_as_result
-		.pend
-		
-func_ceil	.proc
-		; -- ceil: tr = int(f); if tr==f -> return  else return tr+1
-		jsr  pop_float_fac1
-		stx  c64.SCRATCH_ZPREGX
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.MOVMF
-		jsr  c64flt.INT
-		lda  #<fmath_float1
-		ldy  #>fmath_float1
-		jsr  c64flt.FCOMP
-		cmp  #0
-		beq  +
-		lda  #<FL_FONE
-		ldy  #>FL_FONE
-		jsr  c64flt.FADD
-+		jmp  push_fac1_as_result
-		.pend
-
-func_any_f	.proc
-		inx
-		lda  c64.ESTACK_LO,x	; array size
-		sta  c64.SCRATCH_ZPB1
-		asl  a
-		asl  a
-		clc
-		adc  c64.SCRATCH_ZPB1	; times 5 because of float
-		jmp  func_any_b._entry
-		.pend
-
-func_all_f	.proc
-		inx
-		lda  c64.ESTACK_LO,x	; array size
-		sta  c64.SCRATCH_ZPB1
-		asl  a
-		asl  a
-		clc
-		adc  c64.SCRATCH_ZPB1	; times 5 because of float
-		sta  _cmp_mod+1		; self-modifying code
-		jsr  peek_address
-		ldy  #0
--		lda  (c64.SCRATCH_ZPWORD1),y
-		bne  +
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		bne  +
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		bne  +
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		bne  +
-		iny
-		lda  (c64.SCRATCH_ZPWORD1),y
-		bne  +
-		lda  #0
-		sta  c64.ESTACK_LO+1,x
-		rts
-+		iny
-_cmp_mod	cpy  #255		; modified
-		bne  -
-		lda  #1
-		sta  c64.ESTACK_LO+1,x
-		rts
-		.pend
-
-func_max_f	.proc
-		lda  #<_min_float
-		ldy  #>_min_float
-		jsr  c64flt.MOVFM			; fac1=min(float)
-		lda  #255
-		sta  _cmp_mod+1			; compare using 255 so we keep larger values
-_minmax_entry	jsr  pop_array_and_lengthmin1Y
-		stx  c64.SCRATCH_ZPREGX
--		sty  c64.SCRATCH_ZPREG
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.FCOMP
-_cmp_mod	cmp  #255			; will be modified
-		bne  +
-		; fac1 is smaller/larger, so store the new value instead
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.MOVFM
-		ldy  c64.SCRATCH_ZPREG
-		dey
-		cmp  #255
-		beq  +
-		lda  c64.SCRATCH_ZPWORD1
-		clc
-		adc  #5
-		sta  c64.SCRATCH_ZPWORD1
-		bcc  -
-		inc  c64.SCRATCH_ZPWORD1+1
-		bne  -
-+		jmp  push_fac1_as_result
-_min_float	.byte  255,255,255,255,255	; -1.7014118345e+38
-		.pend
-
-func_min_f	.proc
-		lda  #<_max_float
-		ldy  #>_max_float
-		jsr  c64flt.MOVFM			; fac1=max(float)
-		lda  #1
-		sta  func_max_f._cmp_mod+1	; compare using 1 so we keep smaller values
-		jmp  func_max_f._minmax_entry
-_max_float	.byte  255,127,255,255,255	; 1.7014118345e+38
-		.pend
-
-func_sum_f	.proc
-		lda  #<c64.FL_NEGHLF
-		ldy  #>c64.FL_NEGHLF
-		jsr  c64flt.MOVFM
-		jsr  pop_array_and_lengthmin1Y
-		stx  c64.SCRATCH_ZPREGX
--		sty  c64.SCRATCH_ZPREG
-		lda  c64.SCRATCH_ZPWORD1
-		ldy  c64.SCRATCH_ZPWORD1+1
-		jsr  c64flt.FADD
-		ldy  c64.SCRATCH_ZPREG
-		dey
-		cpy  #255
-		beq  +
-		lda  c64.SCRATCH_ZPWORD1
-		clc
-		adc  #5
-		sta  c64.SCRATCH_ZPWORD1
-		bcc  -
-		inc  c64.SCRATCH_ZPWORD1+1
-		bne  -
-+		jsr  c64flt.FADDH
-		jmp  push_fac1_as_result
-		.pend
-}}
-
-}  ; ------ end of block c64flt

compiler/res/prog8lib/c64lib.p8

@@ -1,243 +0,0 @@
-; Prog8 definitions for the Commodore-64
-; Including memory registers, I/O registers, Basic and Kernal subroutines.
-;
-; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
-
-
-~ c64 {
-		memory  ubyte  SCRATCH_ZPB1	= $02		; scratch byte 1 in ZP
-		memory  ubyte  SCRATCH_ZPREG	= $03		; scratch register in ZP
-		memory  ubyte  SCRATCH_ZPREGX	= $fa		; temp storage for X register (stack pointer)
-		memory  uword  SCRATCH_ZPWORD1	= $fb		; scratch word in ZP ($fb/$fc)
-		memory  uword  SCRATCH_ZPWORD2	= $fd		; scratch word in ZP ($fd/$fe)
-		const   uword  ESTACK_LO	= $ce00		; evaluation stack (lsb)
-		const   uword  ESTACK_HI	= $cf00		; evaluation stack (msb)
-
-
-		memory  ubyte  TIME_HI		= $a0		; software jiffy clock, hi byte
-		memory  ubyte  TIME_MID		= $a1		;  .. mid byte
-		memory  ubyte  TIME_LO		= $a2		;    .. lo byte. Updated by IRQ every 1/60 sec
-		memory  ubyte  STKEY		= $91		; various keyboard statuses (updated by IRQ)
-		memory  ubyte  SFDX		= $cb		; current key pressed (matrix value) (updated by IRQ)
-
-		memory  ubyte  COLOR		= $0286		; cursor color
-		memory  ubyte  HIBASE		= $0288		; screen base address / 256 (hi-byte of screen memory address)
-		memory  uword  CINV		= $0314		; IRQ vector
-		memory  uword  NMI_VEC		= $FFFA		; 6502 nmi vector, determined by the kernal if banked in
-		memory  uword  RESET_VEC	= $FFFC		; 6502 reset vector, determined by the kernal if banked in
-		memory  uword  IRQ_VEC		= $FFFE		; 6502 interrupt vector, determined by the kernal if banked in
-
-		; the default addresses for the character screen chars and colors
-		const   uword  Screen		= $0400		; to have this as an array[40*25] the compiler would have to support array size > 255
-		const   uword  Colors		= $d800		; to have this as an array[40*25] the compiler would have to support array size > 255
-
-		; the default locations of the 8 sprite pointers (store address of sprite / 64)
-		memory  ubyte  SPRPTR0		= 2040
-		memory  ubyte  SPRPTR1		= 2041
-		memory  ubyte  SPRPTR2		= 2042
-		memory  ubyte  SPRPTR3		= 2043
-		memory  ubyte  SPRPTR4		= 2044
-		memory  ubyte  SPRPTR5		= 2045
-		memory  ubyte  SPRPTR6		= 2046
-		memory  ubyte  SPRPTR7		= 2047
-		memory  ubyte[8]  SPRPTR	= 2040		; the 8 sprite pointers as an array.
- 
-
-; ---- VIC-II 6567/6569/856x registers ----
-
-		memory  ubyte  SP0X		= $d000
-		memory  ubyte  SP0Y		= $d001
-		memory  ubyte  SP1X		= $d002
-		memory  ubyte  SP1Y		= $d003
-		memory  ubyte  SP2X		= $d004
-		memory  ubyte  SP2Y		= $d005
-		memory  ubyte  SP3X		= $d006
-		memory  ubyte  SP3Y		= $d007
-		memory  ubyte  SP4X		= $d008
-		memory  ubyte  SP4Y		= $d009
-		memory  ubyte  SP5X		= $d00a
-		memory  ubyte  SP5Y		= $d00b
-		memory  ubyte  SP6X		= $d00c
-		memory  ubyte  SP6Y		= $d00d
-		memory  ubyte  SP7X		= $d00e
-		memory  ubyte  SP7Y		= $d00f
-		memory  ubyte[16]  SPXY		= $d000		; the 8 sprite X and Y registers as an array.
-		memory  uword[8]  SPXYW		= $d000		; the 8 sprite X and Y registers as a combined xy word array.
-
-		memory  ubyte  MSIGX		= $d010
-		memory  ubyte  SCROLY		= $d011
-		memory  ubyte  RASTER		= $d012
-		memory  ubyte  LPENX		= $d013
-		memory  ubyte  LPENY		= $d014
-		memory  ubyte  SPENA		= $d015
-		memory  ubyte  SCROLX		= $d016
-		memory  ubyte  YXPAND		= $d017
-		memory  ubyte  VMCSB		= $d018
-		memory  ubyte  VICIRQ		= $d019
-		memory  ubyte  IREQMASK		= $d01a
-		memory  ubyte  SPBGPR		= $d01b
-		memory  ubyte  SPMC		= $d01c
-		memory  ubyte  XXPAND		= $d01d
-		memory  ubyte  SPSPCL		= $d01e
-		memory  ubyte  SPBGCL		= $d01f
-
-		memory  ubyte  EXTCOL		= $d020		; border color
-		memory  ubyte  BGCOL0		= $d021		; screen color
-		memory  ubyte  BGCOL1		= $d022
-		memory  ubyte  BGCOL2		= $d023
-		memory  ubyte  BGCOL4		= $d024
-		memory  ubyte  SPMC0		= $d025
-		memory  ubyte  SPMC1		= $d026
-		memory  ubyte  SP0COL		= $d027
-		memory  ubyte  SP1COL		= $d028
-		memory  ubyte  SP2COL		= $d029
-		memory  ubyte  SP3COL		= $d02a
-		memory  ubyte  SP4COL		= $d02b
-		memory  ubyte  SP5COL		= $d02c
-		memory  ubyte  SP6COL		= $d02d
-		memory  ubyte  SP7COL		= $d02e
-		memory  ubyte[8]  SPCOL		= $d027
-		
-
-; ---- end of VIC-II registers ----
-
-; ---- CIA 6526 1 & 2 registers ----
-
-		memory  ubyte  CIA1PRA		= $DC00		; CIA 1 DRA, keyboard column drive
-		memory  ubyte  CIA1PRB		= $DC01		; CIA 1 DRB, keyboard row port
-		memory  ubyte  CIA1DDRA		= $DC02		; CIA 1 DDRA, keyboard column
-		memory  ubyte  CIA1DDRB		= $DC03		; CIA 1 DDRB, keyboard row
-		memory  ubyte  CIA1TAL		= $DC04		; CIA 1 timer A low byte
-		memory  ubyte  CIA1TAH		= $DC05		; CIA 1 timer A high byte
-		memory  ubyte  CIA1TBL		= $DC06		; CIA 1 timer B low byte
-		memory  ubyte  CIA1TBH		= $DC07		; CIA 1 timer B high byte
-		memory  ubyte  CIA1TOD10	= $DC08		; time of day, 1/10 sec.
-		memory  ubyte  CIA1TODSEC	= $DC09		; time of day, seconds
-		memory  ubyte  CIA1TODMMIN	= $DC0A		; time of day, minutes
-		memory  ubyte  CIA1TODHR	= $DC0B		; time of day, hours
-		memory  ubyte  CIA1SDR		= $DC0C		; Serial Data Register
-		memory  ubyte  CIA1ICR		= $DC0D
-		memory  ubyte  CIA1CRA		= $DC0E
-		memory  ubyte  CIA1CRB		= $DC0F
-
-		memory  ubyte  CIA2PRA		= $DD00		; CIA 2 DRA, serial port and video address
-		memory  ubyte  CIA2PRB		= $DD01		; CIA 2 DRB, RS232 port / USERPORT
-		memory  ubyte  CIA2DDRA		= $DD02		; CIA 2 DDRA, serial port and video address
-		memory  ubyte  CIA2DDRB		= $DD03		; CIA 2 DDRB, RS232 port / USERPORT
-		memory  ubyte  CIA2TAL		= $DD04		; CIA 2 timer A low byte
-		memory  ubyte  CIA2TAH		= $DD05		; CIA 2 timer A high byte
-		memory  ubyte  CIA2TBL		= $DD06		; CIA 2 timer B low byte
-		memory  ubyte  CIA2TBH		= $DD07		; CIA 2 timer B high byte
-		memory  ubyte  CIA2TOD10	= $DD08		; time of day, 1/10 sec.
-		memory  ubyte  CIA2TODSEC	= $DD09		; time of day, seconds
-		memory  ubyte  CIA2TODMIN	= $DD0A		; time of day, minutes
-		memory  ubyte  CIA2TODHR	= $DD0B		; time of day, hours
-		memory  ubyte  CIA2SDR		= $DD0C		; Serial Data Register
-		memory  ubyte  CIA2ICR		= $DD0D
-		memory  ubyte  CIA2CRA		= $DD0E
-		memory  ubyte  CIA2CRB		= $DD0F
-
-; ---- end of CIA registers ----
-
-; ---- SID 6581/8580 registers ----
-
-		memory  ubyte  FREQLO1		= $D400		; channel 1 freq lo
-		memory	ubyte  FREQHI1		= $D401		; channel 1 freq hi
-		memory  uword  FREQ1		= $D400		; channel 1 freq (word)
-		memory  ubyte  PWLO1		= $D402		; channel 1 pulse width lo (7-0)
-		memory  ubyte  PWHI1		= $D403		; channel 1 pulse width hi (11-8)
-		memory  uword  PW1		= $D402		; channel 1 pulse width (word)
-		memory  ubyte  CR1		= $D404		; channel 1 voice control register
-		memory  ubyte  AD1		= $D405		; channel 1 attack & decay
-		memory  ubyte  SR1		= $D406		; channel 1 sustain & release
-		memory  ubyte  FREQLO2		= $D407		; channel 2 freq lo
-		memory	ubyte  FREQHI2		= $D408		; channel 2 freq hi
-		memory  uword  FREQ2		= $D407		; channel 2 freq (word)
-		memory  ubyte  PWLO2		= $D409		; channel 2 pulse width lo (7-0)
-		memory  ubyte  PWHI2		= $D40A		; channel 2 pulse width hi (11-8)
-		memory  uword  PW2		= $D409		; channel 2 pulse width (word)
-		memory  ubyte  CR2		= $D40B		; channel 2 voice control register
-		memory  ubyte  AD2		= $D40C		; channel 2 attack & decay
-		memory  ubyte  SR2		= $D40D		; channel 2 sustain & release
-		memory  ubyte  FREQLO3		= $D40E		; channel 3 freq lo
-		memory	ubyte  FREQHI3		= $D40F		; channel 3 freq hi
-		memory  uword  FREQ3		= $D40E		; channel 3 freq (word)
-		memory  ubyte  PWLO3		= $D410		; channel 3 pulse width lo (7-0)
-		memory  ubyte  PWHI3		= $D411		; channel 3 pulse width hi (11-8)
-		memory  uword  PW3		= $D410		; channel 3 pulse width (word)
-		memory  ubyte  CR3		= $D412		; channel 3 voice control register
-		memory  ubyte  AD3		= $D413		; channel 3 attack & decay
-		memory  ubyte  SR3		= $D414		; channel 3 sustain & release
-		memory  ubyte  FCLO		= $D415		; filter cutoff lo (2-0)
-		memory  ubyte  FCHI		= $D416		; filter cutoff hi (10-3)
-		memory  uword  FC		= $D415		; filter cutoff (word)
-		memory  ubyte  RESFILT		= $D417		; filter resonance and routing
-		memory  ubyte  MVOL		= $D418		; filter mode and main volume control
-		memory  ubyte  POTX		= $D419		; potentiometer X
-		memory  ubyte  POTY		= $D41A		; potentiometer Y
-		memory  ubyte  OSC3		= $D41B		; channel 3 oscillator value read
-		memory  ubyte  ENV3		= $D41C		; channel 3 envelope value read
-
-; ---- end of SID registers ----
-
-
-
-; ---- C64 basic routines ----
-
-asmsub	CLEARSCR	() -> clobbers(A,X,Y) -> ()		= $E544		; clear the screen
-asmsub	HOMECRSR	() -> clobbers(A,X,Y) -> ()		= $E566		; cursor to top left of screen
-
-
-; ---- end of C64 basic routines ----
-
-
-; ---- C64 kernal routines ----
-
-asmsub	STROUT   (uword strptr @ AY) -> clobbers(A, X, Y) -> ()	= $AB1E		; print null-terminated string (use c64scr.print instead)
-asmsub	IRQDFRT  () -> clobbers(A,X,Y) -> ()			= $EA31		; default IRQ routine
-asmsub	IRQDFEND () -> clobbers(A,X,Y) -> ()			= $EA81		; default IRQ end/cleanup
-asmsub	CINT     () -> clobbers(A,X,Y) -> ()			= $FF81		; (alias: SCINIT) initialize screen editor and video chip
-asmsub	IOINIT   () -> clobbers(A, X) -> ()			= $FF84		; initialize I/O devices (CIA, SID, IRQ)
-asmsub	RAMTAS   () -> clobbers(A,X,Y) -> ()			= $FF87		; initialize RAM, tape buffer, screen
-asmsub	RESTOR   () -> clobbers(A,X,Y) -> ()			= $FF8A		; restore default I/O vectors
-asmsub	VECTOR   (ubyte dir @ Pc, uword userptr @ XY) -> clobbers(A,Y) -> ()	= $FF8D		; read/set I/O vector table
-asmsub	SETMSG   (ubyte value @ A) -> clobbers() -> ()		= $FF90		; set Kernal message control flag
-asmsub	SECOND   (ubyte address @ A) -> clobbers(A) -> ()	= $FF93		; (alias: LSTNSA) send secondary address after LISTEN
-asmsub	TKSA     (ubyte address @ A) -> clobbers(A) -> ()	= $FF96		; (alias: TALKSA) send secondary address after TALK
-asmsub	MEMTOP   (ubyte dir @ Pc, uword address @ XY) -> clobbers() -> (uword @ XY)	= $FF99		; read/set top of memory  pointer
-asmsub	MEMBOT   (ubyte dir @ Pc, uword address @ XY) -> clobbers() -> (uword @ XY)	= $FF9C		; read/set bottom of memory  pointer
-asmsub	SCNKEY   () -> clobbers(A,X,Y) -> ()			= $FF9F		; scan the keyboard
-asmsub	SETTMO   (ubyte timeout @ A) -> clobbers() -> ()	= $FFA2		; set time-out flag for IEEE bus
-asmsub	ACPTR    () -> clobbers() -> (ubyte @ A)			= $FFA5		; (alias: IECIN) input byte from serial bus
-asmsub	CIOUT    (ubyte databyte @ A) -> clobbers() -> ()	= $FFA8		; (alias: IECOUT) output byte to serial bus
-asmsub	UNTLK    () -> clobbers(A) -> ()			= $FFAB		; command serial bus device to UNTALK
-asmsub	UNLSN    () -> clobbers(A) -> ()			= $FFAE		; command serial bus device to UNLISTEN
-asmsub	LISTEN   (ubyte device @ A) -> clobbers(A) -> ()	= $FFB1		; command serial bus device to LISTEN
-asmsub	TALK     (ubyte device @ A) -> clobbers(A) -> ()	= $FFB4		; command serial bus device to TALK
-asmsub	READST   () -> clobbers() -> (ubyte @ A)			= $FFB7		; read I/O status word
-asmsub	SETLFS   (ubyte logical @ A, ubyte device @ X, ubyte address @ Y) -> clobbers() -> () = $FFBA	; set logical file parameters
-asmsub	SETNAM   (ubyte namelen @ A, str filename @ XY) -> clobbers() -> ()	= $FFBD		; set filename parameters
-asmsub	OPEN     () -> clobbers(A,X,Y) -> ()			= $FFC0		; (via 794 ($31A)) open a logical file
-asmsub	CLOSE    (ubyte logical @ A) -> clobbers(A,X,Y) -> ()	= $FFC3		; (via 796 ($31C)) close a logical file
-asmsub	CHKIN    (ubyte logical @ X) -> clobbers(A,X) -> ()	= $FFC6		; (via 798 ($31E)) define an input channel
-asmsub	CHKOUT   (ubyte logical @ X) -> clobbers(A,X) -> ()	= $FFC9		; (via 800 ($320)) define an output channel
-asmsub	CLRCHN   () -> clobbers(A,X) -> ()			= $FFCC		; (via 802 ($322)) restore default devices
-asmsub	CHRIN    () -> clobbers(Y) -> (ubyte @ A)		= $FFCF		; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-asmsub	CHROUT   (ubyte char @ A) -> clobbers() -> ()		= $FFD2		; (via 806 ($326)) output a character
-asmsub	LOAD     (ubyte verify @ A, uword address @ XY) -> clobbers() -> (ubyte @Pc, ubyte @ A, ubyte @ X, ubyte @ Y) = $FFD5	; (via 816 ($330)) load from device
-asmsub	SAVE     (ubyte zp_startaddr @ A, uword endaddr @ XY) -> clobbers() -> (ubyte @ Pc, ubyte @ A) = $FFD8	; (via 818 ($332)) save to a device
-asmsub	SETTIM   (ubyte low @ A, ubyte middle @ X, ubyte high @ Y) -> clobbers() -> ()	= $FFDB		; set the software clock
-asmsub	RDTIM    () -> clobbers() -> (ubyte @ A, ubyte @ X, ubyte @ Y) = $FFDE	; read the software clock
-asmsub	STOP     () -> clobbers(A,X) -> (ubyte @ Pz, ubyte @ Pc)	= $FFE1		; (via 808 ($328)) check the STOP key
-asmsub	GETIN    () -> clobbers(X,Y) -> (ubyte @ A)		= $FFE4		; (via 810 ($32A)) get a character
-asmsub	CLALL    () -> clobbers(A,X) -> ()			= $FFE7		; (via 812 ($32C)) close all files
-asmsub	UDTIM    () -> clobbers(A,X) -> ()			= $FFEA		; update the software clock
-asmsub	SCREEN   () -> clobbers() -> (ubyte @ X, ubyte @ Y)	= $FFED		; read number of screen rows and columns
-asmsub	PLOT     (ubyte dir @ Pc, ubyte col @ Y, ubyte row @ X) -> clobbers() -> (ubyte @ X, ubyte @ Y)	= $FFF0		; read/set position of cursor on screen.  See c64scr.PLOT for a 'safe' wrapper that preserves X.
-asmsub	IOBASE   () -> clobbers() -> (uword @ XY)		= $FFF3		; read base address of I/O devices
-
-; ---- end of C64 kernal routines ----
-
-}

compiler/res/prog8lib/conv.p8

@@ -0,0 +1,736 @@
+; Number conversions routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+
+conv {
+
+; ----- number conversions to decimal strings ----
+
+    str  string_out = "????????????????"       ; result buffer for the string conversion routines
+
+asmsub  str_ub0  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in decimal string form, with left padding 0s (3 positions total)
+	%asm {{
+            phx
+            jsr  conv.ubyte2decimal
+            sty  string_out
+            sta  string_out+1
+            stx  string_out+2
+            lda  #0
+            sta  string_out+3
+            plx
+            rts
+	}}
+}
+
+asmsub  str_ub  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in decimal string form, without left padding 0s
+	%asm {{
+		phx
+		ldy  #0
+		sty  P8ZP_SCRATCH_B1
+		jsr  conv.ubyte2decimal
+_output_byte_digits
+                ; hundreds?
+		cpy  #'0'
+		beq  +
+		pha
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		sta  string_out,y
+		pla
+		inc  P8ZP_SCRATCH_B1
+		; tens?
++		ldy  P8ZP_SCRATCH_B1
+                cmp  #'0'
+		beq  +
+		sta  string_out,y
+		iny
++               ; ones.
+                txa
+                sta  string_out,y
+                iny
+                lda  #0
+                sta  string_out,y
+                plx
+                rts
+	}}
+}
+
+asmsub  str_b  (byte value @ A) clobbers(A,Y)  {
+	; ---- convert the byte in A in decimal string form, without left padding 0s
+	%asm {{
+            phx
+            ldy  #0
+            sty  P8ZP_SCRATCH_B1
+            cmp  #0
+            bpl  +
+            pha
+            lda  #'-'
+            sta  string_out
+            inc  P8ZP_SCRATCH_B1
+            pla
++	    jsr  conv.byte2decimal
+            bra  str_ub._output_byte_digits
+	}}
+}
+
+asmsub  str_ubhex  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in hex string form
+	%asm {{
+            jsr  conv.ubyte2hex
+            sta  string_out
+            sty  string_out+1
+            lda  #0
+            sta  string_out+2
+            rts
+	}}
+}
+
+asmsub  str_ubbin  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in binary string form
+	%asm {{
+	    sta  P8ZP_SCRATCH_B1
+	    ldy  #0
+	    sty  string_out+8
+	    ldy  #7
+-	    lsr  P8ZP_SCRATCH_B1
+            bcc  +
+            lda  #'1'
+            bne  _digit
++           lda  #'0'
+_digit      sta  string_out,y
+            dey
+	    bpl  -
+	    rts
+	}}
+}
+
+asmsub  str_uwbin  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in binary string form
+	%asm {{
+	    sta  P8ZP_SCRATCH_REG
+	    tya
+	    jsr  str_ubbin
+	    ldy  #0
+	    sty  string_out+16
+	    ldy  #7
+-	    lsr  P8ZP_SCRATCH_REG
+            bcc  +
+            lda  #'1'
+            bne  _digit
++           lda  #'0'
+_digit      sta  string_out+8,y
+            dey
+	    bpl  -
+	    rts
+	}}
+}
+
+asmsub  str_uwhex  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in hexadecimal string form (4 digits)
+	%asm {{
+            pha
+            tya
+            jsr  conv.ubyte2hex
+            sta  string_out
+            sty  string_out+1
+            pla
+            jsr  conv.ubyte2hex
+            sta  string_out+2
+            sty  string_out+3
+            lda  #0
+            sta  string_out+4
+            rts
+	}}
+}
+
+asmsub  str_uw0  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
+	%asm {{
+	    phx
+	    jsr  conv.uword2decimal
+	    ldy  #0
+-           lda  conv.uword2decimal.decTenThousands,y
+            sta  string_out,y
+            beq  +
+            iny
+            bne  -
++           plx
+	    rts
+	}}
+}
+
+asmsub  str_uw  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in decimal string form, without left padding 0s
+	%asm {{
+	    phx
+	    jsr  conv.uword2decimal
+	    ldx  #0
+_output_digits
+	    ldy  #0
+-           lda  conv.uword2decimal.decTenThousands,y
+            beq  _allzero
+            cmp  #'0'
+            bne  _gotdigit
+            iny
+            bne  -
+_gotdigit   sta  string_out,x
+            inx
+            iny
+            lda  conv.uword2decimal.decTenThousands,y
+            bne  _gotdigit
+_end        lda  #0
+            sta  string_out,x
+            plx
+            rts
+
+_allzero    lda  #'0'
+            sta  string_out,x
+            inx
+            bne  _end
+	}}
+}
+
+asmsub  str_w  (word value @ AY) clobbers(A,Y)  {
+	; ---- convert the (signed) word in A/Y in decimal string form, without left padding 0's
+	%asm {{
+	    cpy  #0
+	    bpl  str_uw
+	    phx
+	    pha
+	    lda  #'-'
+	    sta  string_out
+            tya
+            eor  #255
+            tay
+            pla
+            eor  #255
+            clc
+            adc  #1
+            bcc  +
+            iny
++	    jsr  conv.uword2decimal
+	    ldx  #1
+	    bne  str_uw._output_digits
+	}}
+}
+
+
+; ---- string conversion to numbers -----
+
+asmsub  any2uword(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- parses a string into a 16 bit unsigned number. String may be in decimal, hex or binary format.
+	;    (the latter two require a $ or % prefix to be recognised)
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    returns amount of processed characters in A, and the parsed number will be in cx16.r15.
+	;    if the string was invalid, 0 will be returned in A.
+	%asm {{
+	pha
+	sta  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	ldy  #0
+	lda  (P8ZP_SCRATCH_W1),y
+	ldy  P8ZP_SCRATCH_W1+1
+	cmp  #'$'
+	beq  _hex
+	cmp  #'%'
+	beq  _bin
+	pla
+	jsr  str2uword
+	jmp  _result
+_hex	pla
+	jsr  hex2uword
+	jmp  _result
+_bin	pla
+	jsr  bin2uword
+_result
+        pha
+        lda  cx16.r15
+        sta  P8ZP_SCRATCH_B1        ; result value
+        pla
+        sta  cx16.r15
+        sty  cx16.r15+1
+        lda  P8ZP_SCRATCH_B1
+        rts
+	}}
+}
+
+inline asmsub  str2ubyte(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- returns in A the unsigned byte value of the string number argument in AY
+	;    the number may NOT be preceded by a + sign and may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in A,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+		jsr  conv.str2uword
+	}}
+}
+
+inline asmsub  str2byte(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- returns in A the signed byte value of the string number argument in AY
+	;    the number may be preceded by a + or - sign but may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in A,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+		jsr  conv.str2word
+	}}
+}
+
+asmsub  str2uword(str string @AY) -> uword @AY {
+	; -- returns the unsigned word value of the string number argument in AY
+	;    the number may NOT be preceded by a + sign and may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+_result = P8ZP_SCRATCH_W1
+        	sta  P8ZP_SCRATCH_W2
+        	sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		sty  _result
+		sty  _result+1
+		sty  cx16.r15+1
+_loop
+		lda  (P8ZP_SCRATCH_W2),y
+		sec
+		sbc  #48
+		bpl  _digit
+_done
+		sty  cx16.r15
+		lda  _result
+		ldy  _result+1
+		rts
+_digit
+		cmp  #10
+		bcs  _done
+		; add digit to result
+		pha
+		jsr  _result_times_10
+		pla
+		clc
+		adc  _result
+		sta  _result
+		bcc  +
+		inc  _result+1
++		iny
+		bne  _loop
+		; never reached
+
+_result_times_10     ; (W*4 + W)*2
+		lda  _result+1
+		sta  P8ZP_SCRATCH_REG
+		lda  _result
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		clc
+		adc  _result
+		sta  _result
+		lda  P8ZP_SCRATCH_REG
+		adc  _result+1
+		asl  _result
+		rol  a
+		sta  _result+1
+		rts
+	}}
+}
+
+asmsub  str2word(str string @AY) -> word @AY {
+	; -- returns the signed word value of the string number argument in AY
+	;    the number may be preceded by a + or - sign but may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+_result = P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		sty  _result
+		sty  _result+1
+		sty  _negative
+		sty  cx16.r15+1
+		lda  (P8ZP_SCRATCH_W2),y
+		cmp  #'+'
+		bne  +
+		iny
++		cmp  #'-'
+		bne  _parse
+		inc  _negative
+		iny
+_parse		lda  (P8ZP_SCRATCH_W2),y
+		sec
+		sbc  #48
+		bpl  _digit
+_done
+		sty  cx16.r15
+		lda  _negative
+		beq  +
+		sec
+		lda  #0
+		sbc  _result
+		sta  _result
+		lda  #0
+		sbc  _result+1
+		sta  _result+1
++		lda  _result
+		ldy  _result+1
+		rts
+_digit
+		cmp  #10
+		bcs  _done
+		; add digit to result
+		pha
+		jsr  str2uword._result_times_10
+		pla
+		clc
+		adc  _result
+		sta  _result
+		bcc  +
+		inc  _result+1
++		iny
+		bne  _parse
+		; never reached
+_negative	.byte  0
+	}}
+}
+
+asmsub  hex2uword(str string @AY) -> uword @AY {
+	; -- hexadecimal string (with or without '$') to uword.
+	;    string may be in petscii or c64-screencode encoding.
+	;    stops parsing at the first character that's not a hex digit (except leading $)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+	sta  P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_W2+1
+	ldy  #0
+	sty  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	sty  cx16.r15+1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #'$'
+	bne  _loop
+	iny
+_loop
+	lda  #0
+	sta  P8ZP_SCRATCH_B1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #7                 ; screencode letters A-F are 1-6
+	bcc  _add_letter
+	cmp  #'g'
+	bcs  _stop
+	cmp  #'a'
+	bcs  _add_letter
+	cmp  #'0'
+	bcc  _stop
+	cmp  #'9'+1
+	bcs  _stop
+_calc
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	and  #$0f
+	clc
+	adc  P8ZP_SCRATCH_B1
+	ora  P8ZP_SCRATCH_W1
+	sta  P8ZP_SCRATCH_W1
+	iny
+	bne  _loop
+_stop
+	sty  cx16.r15
+	lda  P8ZP_SCRATCH_W1
+	ldy  P8ZP_SCRATCH_W1+1
+	rts
+_add_letter
+	pha
+	lda  #9
+	sta  P8ZP_SCRATCH_B1
+	pla
+	jmp  _calc
+	}}
+}
+
+asmsub  bin2uword(str string @AY) -> uword @AY {
+	; -- binary string (with or without '%') to uword.
+	;    stops parsing at the first character that's not a 0 or 1. (except leading %)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+	sta  P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_W2+1
+	ldy  #0
+	sty  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	sty  cx16.r15+1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #'%'
+	bne  _loop
+	iny
+_loop
+	lda  (P8ZP_SCRATCH_W2),y
+	cmp  #'0'
+	bcc  _stop
+	cmp  #'2'
+	bcs  _stop
+_first  asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	and  #1
+	ora  P8ZP_SCRATCH_W1
+	sta  P8ZP_SCRATCH_W1
+	iny
+	bne  _loop
+_stop
+	sty  cx16.r15
+	lda  P8ZP_SCRATCH_W1
+	ldy  P8ZP_SCRATCH_W1+1
+	rts
+	}}
+}
+
+
+; ----- low level number conversions to decimal strings ----
+
+asmsub  ubyte2decimal  (ubyte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
+	; ---- A to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
+	%asm {{
+		ldy  #uword2decimal.ASCII_0_OFFSET
+		bne  uword2decimal.hex_try200
+		rts
+	}}
+}
+
+asmsub  uword2decimal  (uword value @AY) -> ubyte @Y, ubyte @A, ubyte @X  {
+	;  ---- convert 16 bit uword in A/Y to decimal
+	;  output in uword2decimal.decTenThousands, decThousands, decHundreds, decTens, decOnes
+	;  (these are terminated by a zero byte so they can be easily printed)
+	;  also returns Y = 100's, A = 10's, X = 1's
+
+	%asm {{
+
+;Convert 16 bit Hex to Decimal (0-65535) Rev 2
+;By Omegamatrix    Further optimizations by tepples
+; routine from http://forums.nesdev.com/viewtopic.php?f=2&t=11341&start=15
+
+;HexToDec99
+; start in A
+; end with A = 10's, decOnes (also in X)
+
+;HexToDec255
+; start in A
+; end with Y = 100's, A = 10's, decOnes (also in X)
+
+;HexToDec999
+; start with A = high byte, Y = low byte
+; end with Y = 100's, A = 10's, decOnes (also in X)
+; requires 1 extra temp register on top of decOnes, could combine
+; these two if HexToDec65535 was eliminated...
+
+;HexToDec65535
+; start with A/Y (low/high) as 16 bit value
+; end with decTenThousand, decThousand, Y = 100's, A = 10's, decOnes (also in X)
+; (irmen: I store Y and A in decHundreds and decTens too, so all of it can be easily printed)
+
+
+ASCII_0_OFFSET 	= $30
+temp       	    = P8ZP_SCRATCH_B1	; byte in zeropage
+hexHigh      	= P8ZP_SCRATCH_W1	; byte in zeropage
+hexLow       	= P8ZP_SCRATCH_W1+1	; byte in zeropage
+
+
+HexToDec65535; SUBROUTINE
+    sty    hexHigh               ;3  @9
+    sta    hexLow                ;3  @12
+    tya
+    tax                          ;2  @14
+    lsr    a                     ;2  @16
+    lsr    a                     ;2  @18   integer divide 1024 (result 0-63)
+
+    cpx    #$A7                  ;2  @20   account for overflow of multiplying 24 from 43,000 ($A7F8) onward,
+    adc    #1                    ;2  @22   we can just round it to $A700, and the divide by 1024 is fine...
+
+    ;at this point we have a number 1-65 that we have to times by 24,
+    ;add to original sum, and Mod 1024 to get a remainder 0-999
+
+
+    sta    temp                  ;3  @25
+    asl    a                     ;2  @27
+    adc    temp                  ;3  @30  x3
+    tay                          ;2  @32
+    lsr    a                     ;2  @34
+    lsr    a                     ;2  @36
+    lsr    a                     ;2  @38
+    lsr    a                     ;2  @40
+    lsr    a                     ;2  @42
+    tax                          ;2  @44
+    tya                          ;2  @46
+    asl    a                     ;2  @48
+    asl    a                     ;2  @50
+    asl    a                     ;2  @52
+    clc                          ;2  @54
+    adc    hexLow                ;3  @57
+    sta    hexLow                ;3  @60
+    txa                          ;2  @62
+    adc    hexHigh               ;3  @65
+    sta    hexHigh               ;3  @68
+    ror    a                     ;2  @70
+    lsr    a                     ;2  @72
+    tay                          ;2  @74    integer divide 1,000 (result 0-65)
+
+    lsr    a                     ;2  @76    split the 1,000 and 10,000 digit
+    tax                          ;2  @78
+    lda    ShiftedBcdTab,x       ;4  @82
+    tax                          ;2  @84
+    rol    a                     ;2  @86
+    and    #$0F                  ;2  @88
+    ora    #ASCII_0_OFFSET
+    sta    decThousands          ;3  @91
+    txa                          ;2  @93
+    lsr    a                     ;2  @95
+    lsr    a                     ;2  @97
+    lsr    a                     ;2  @99
+    ora    #ASCII_0_OFFSET
+    sta    decTenThousands       ;3  @102
+
+    lda    hexLow                ;3  @105
+    cpy    temp                  ;3  @108
+    bmi    _doSubtract           ;2³ @110/111
+    beq    _useZero               ;2³ @112/113
+    adc    #23 + 24              ;2  @114
+_doSubtract
+    sbc    #23                   ;2  @116
+    sta    hexLow                ;3  @119
+_useZero
+    lda    hexHigh               ;3  @122
+    sbc    #0                    ;2  @124
+
+Start100s
+    and    #$03                  ;2  @126
+    tax                          ;2  @128   0,1,2,3
+    cmp    #2                    ;2  @130
+    rol    a                     ;2  @132   0,2,5,7
+    ora    #ASCII_0_OFFSET
+    tay                          ;2  @134   Y = Hundreds digit
+
+    lda    hexLow                ;3  @137
+    adc    Mod100Tab,x           ;4  @141    adding remainder of 256, 512, and 256+512 (all mod 100)
+    bcs    hex_doSub200             ;2³ @143/144
+
+hex_try200
+    cmp    #200                  ;2  @145
+    bcc    hex_try100               ;2³ @147/148
+hex_doSub200
+    iny                          ;2  @149
+    iny                          ;2  @151
+    sbc    #200                  ;2  @153
+hex_try100
+    cmp    #100                  ;2  @155
+    bcc    HexToDec99            ;2³ @157/158
+    iny                          ;2  @159
+    sbc    #100                  ;2  @161
+
+HexToDec99; SUBROUTINE
+    lsr    a                     ;2  @163
+    tax                          ;2  @165
+    lda    ShiftedBcdTab,x       ;4  @169
+    tax                          ;2  @171
+    rol    a                     ;2  @173
+    and    #$0F                  ;2  @175
+    ora    #ASCII_0_OFFSET
+    sta    decOnes               ;3  @178
+    txa                          ;2  @180
+    lsr    a                     ;2  @182
+    lsr    a                     ;2  @184
+    lsr    a                     ;2  @186
+    ora    #ASCII_0_OFFSET
+
+    ; irmen: load X with ones, and store Y and A too, for easy printing afterwards
+    sty  decHundreds
+    sta  decTens
+    ldx  decOnes
+    rts                          ;6  @192   Y=hundreds, A = tens digit, X=ones digit
+
+
+HexToDec999; SUBROUTINE
+    sty    hexLow                ;3  @9
+    jmp    Start100s             ;3  @12
+
+Mod100Tab
+    .byte 0,56,12,56+12
+
+ShiftedBcdTab
+    .byte $00,$01,$02,$03,$04,$08,$09,$0A,$0B,$0C
+    .byte $10,$11,$12,$13,$14,$18,$19,$1A,$1B,$1C
+    .byte $20,$21,$22,$23,$24,$28,$29,$2A,$2B,$2C
+    .byte $30,$31,$32,$33,$34,$38,$39,$3A,$3B,$3C
+    .byte $40,$41,$42,$43,$44,$48,$49,$4A,$4B,$4C
+
+decTenThousands   	.byte  0
+decThousands    	.byte  0
+decHundreds		.byte  0
+decTens			.byte  0
+decOnes   		.byte  0
+			.byte  0		; zero-terminate the decimal output string
+
+    }}
+}
+
+asmsub  byte2decimal  (byte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
+	; ---- A (signed byte) to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
+	;      note: if the number is negative, you have to deal with the '-' yourself!
+	%asm {{
+		cmp  #0
+		bpl  +
+		eor  #255
+		clc
+		adc  #1
++		jmp  ubyte2decimal
+	}}
+}
+
+asmsub  ubyte2hex  (ubyte value @A) -> ubyte @A, ubyte @Y  {
+	; ---- A to hex petscii string in AY (first hex char in A, second hex char in Y)
+	%asm {{
+		stx  P8ZP_SCRATCH_REG
+		pha
+		and  #$0f
+		tax
+		ldy  _hex_digits,x
+		pla
+		lsr  a
+		lsr  a
+		lsr  a
+		lsr  a
+		tax
+		lda  _hex_digits,x
+		ldx  P8ZP_SCRATCH_REG
+		rts
+
+_hex_digits	.text "0123456789abcdef"	; can probably be reused for other stuff as well
+	}}
+}
+
+asmsub  uword2hex  (uword value @AY) clobbers(A,Y)  {
+	; ---- convert 16 bit uword in A/Y into 4-character hexadecimal string 'uword2hex.output' (0-terminated)
+	%asm {{
+		sta  P8ZP_SCRATCH_REG
+		tya
+		jsr  ubyte2hex
+		sta  output
+		sty  output+1
+		lda  P8ZP_SCRATCH_REG
+		jsr  ubyte2hex
+		sta  output+2
+		sty  output+3
+		rts
+output		.text  "0000", $00      ; 0-terminated output buffer (to make printing easier)
+	}}
+}
+
+}

compiler/res/prog8lib/cx16/floats.p8

@@ -0,0 +1,164 @@
+; Prog8 definitions for floating point handling on the CommanderX16
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target cx16
+%option enable_floats
+
+floats {
+	; ---- this block contains C-64 floating point related functions ----
+
+        const float  PI     = 3.141592653589793
+        const float  TWOPI  = 6.283185307179586
+
+
+; ---- ROM float functions ----
+
+		; note: the fac1 and fac2 are working registers and take 6 bytes each,
+		; floats in memory  (and rom) are stored in 5-byte MFLPT packed format.
+
+; note: fac1/2 might get clobbered even if not mentioned in the function's name.
+; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
+
+romsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101 ($64-$65) MSB FIRST. (might throw ILLEGAL QUANTITY)
+
+; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1
+; there is also floats.GIVUAYFAY - unsigned word in A/Y (lo/hi) to fac1
+; (tip: use GIVAYFAY to use A/Y input; lo/hi switched to normal order)
+romsub $fe03 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
+
+; fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
+; (tip: use GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
+romsub $fe06 = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
+
+romsub $fe09 = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for rounding- call this before INT
+romsub $fe0c = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
+romsub $fe0f = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
+romsub $fe12 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
+romsub $fe15 = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
+romsub $fe1b = ZEROFC() clobbers(A,X,Y)                     ; fac1 = 0
+romsub $fe1e = NORMAL() clobbers(A,X,Y)                     ; normalize fac1 (?)
+romsub $fe24 = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
+romsub $fe27 = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
+romsub $fe2a = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
+romsub $fe33 = CONUPK(uword mflpt @ AY) clobbers(A,Y)       ; load mflpt value from memory  in A/Y into fac2
+romsub $fe36 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
+romsub $fe3c = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive!
+romsub $fe3f = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
+romsub $fe42 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
+
+romsub $fe48 = MOVFM(uword mflpt @ AY) clobbers(A,Y)        ; load mflpt value from memory  in A/Y into fac1
+romsub $fe4b = MOVMF(uword mflpt @ XY) clobbers(A,Y)        ; store fac1 to memory  X/Y as 5-byte mflpt
+romsub $fe4e = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+romsub $fe51 = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded)
+romsub $fe54 = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+romsub $fe5a = SIGN() -> ubyte @ A                          ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+romsub $fe5d = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
+romsub $fe60 = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
+romsub $fe6c = ABS()                                        ; fac1 = ABS(fac1)
+romsub $fe6f = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
+romsub $fe78 = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
+romsub $fe7e = FINLOG(byte value @A) clobbers (A, X, Y)           ; fac1 += signed byte in A
+romsub $fe81 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
+romsub $fe8a = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+romsub $fe8d = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+; note: there is no FPWR() on the Cx16
+romsub $fe93 = NEGOP() clobbers(A)                          ; switch the sign of fac1
+romsub $fe96 = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+romsub $fe9f = RND2(byte value @A) clobbers(A,X,Y)                ; fac1 = RND(A) float random number generator
+romsub $fea2 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
+romsub $fea5 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+romsub $fea8 = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+romsub $feab = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+romsub $feae = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+
+
+asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
+	%asm {{
+        phx
+        sta  _tmp
+        sty  P8ZP_SCRATCH_B1
+        tya
+        ldy  _tmp
+        jsr  GIVAYF                 ; load it as signed... correct afterwards
+        lda  P8ZP_SCRATCH_B1
+	    bpl  +
+	    lda  #<_flt65536
+	    ldy  #>_flt65536
+	    jsr  FADD
++	    plx
+        rts
+_tmp        .byte 0
+_flt65536    .byte 145,0,0,0,0       ; 65536.0
+	}}
+}
+
+
+asmsub  GIVAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- signed 16 bit word in A/Y (lo/hi) to float in fac1
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		jmp  GIVAYF		; this uses the inverse order, Y/A
+	}}
+}
+
+asmsub  FTOSWRDAY  () clobbers(X) -> uword @ AY  {
+	; ---- fac1 to signed word in A/Y
+	%asm {{
+		jsr  FTOSWORDYA	; note the inverse Y/A order
+		sta  P8ZP_SCRATCH_B1
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		rts
+	}}
+}
+
+asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
+	; ---- fac1 to unsigned word in A/Y
+	%asm {{
+		jsr  GETADR		; this uses the inverse order, Y/A
+		sta  P8ZP_SCRATCH_B1
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		rts
+	}}
+}
+
+asmsub  FREADUY (ubyte value @Y) {
+    ; -- 8 bit unsigned Y -> float in fac1
+    %asm {{
+        lda  #0
+        jmp  GIVAYF
+    }}
+}
+
+sub  print_f  (float value) {
+	; ---- prints the floating point value (without a newline).
+	%asm {{
+		phx
+		lda  #<value
+		ldy  #>value
+		jsr  MOVFM		; load float into fac1
+		jsr  FOUT		; fac1 to string in A/Y
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		plx
+		rts
+	}}
+}
+
+%asminclude "library:c64/floats.asm", ""
+%asminclude "library:c64/floats_funcs.asm", ""
+
+}

compiler/res/prog8lib/cx16/gfx2.p8

@@ -0,0 +1,974 @@
+%target cx16
+
+; Bitmap pixel graphics routines for the CommanderX16
+; Custom routines to use the full-screen 640x480 and 320x240 screen modes.
+; (These modes are not supported by the documented GRAPH_xxxx kernal routines)
+;
+; No text layer is currently shown, text can be drawn as part of the bitmap itself.
+; Note: for similar graphics routines that also work on the C-64, use the "graphics" module instead.
+; Note: for color palette manipulation, use the "palette" module or write Vera registers yourself.
+; Note: this library implements code for various resolutions and color depths. This takes up memory.
+;       If you're memory constrained you should probably not use this built-in library,
+;       but make a copy in your project only containing the code for the required resolution.
+;
+;
+; SCREEN MODE LIST:
+;   mode 0 = reset back to default text mode
+;   mode 1 = bitmap 320 x 240 monochrome
+;   mode 2 = bitmap 320 x 240 x 4c (TODO not yet implemented)
+;   mode 3 = bitmap 320 x 240 x 16c (TODO not yet implemented)
+;   mode 4 = bitmap 320 x 240 x 256c
+;   mode 5 = bitmap 640 x 480 monochrome
+;   mode 6 = bitmap 640 x 480 x 4c
+;   higher color dephts in highres are not supported due to lack of VRAM
+
+; TODO can we make a FB vector table and emulation routines for the Cx16s' GRAPH_init() call? to replace the builtin 320x200 fb driver?
+
+gfx2 {
+
+    ; read-only control variables:
+    ubyte active_mode = 0
+    uword width = 0
+    uword height = 0
+    ubyte bpp = 0
+    ubyte monochrome_dont_stipple_flag = false            ; set to false to enable stippling mode in monochrome displaymodes
+
+    sub screen_mode(ubyte mode) {
+        when mode {
+            1 -> {
+                ; lores monochrome
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 64
+                cx16.VERA_DC_VSCALE = 64
+                cx16.VERA_L1_CONFIG = %00000100
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = 0
+                width = 320
+                height = 240
+                bpp = 1
+            }
+            ; TODO modes 2, 3 not yet implemented
+            4 -> {
+                ; lores 256c
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 64
+                cx16.VERA_DC_VSCALE = 64
+                cx16.VERA_L1_CONFIG = %00000111
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = 0
+                width = 320
+                height = 240
+                bpp = 8
+            }
+            5 -> {
+                ; highres monochrome
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 128
+                cx16.VERA_DC_VSCALE = 128
+                cx16.VERA_L1_CONFIG = %00000100
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = %00000001
+                width = 640
+                height = 480
+                bpp = 1
+            }
+            6 -> {
+                ; highres 4c
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 128
+                cx16.VERA_DC_VSCALE = 128
+                cx16.VERA_L1_CONFIG = %00000101
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = %00000001
+                width = 640
+                height = 480
+                bpp = 2
+            }
+            else -> {
+                ; back to default text mode and colors
+                cx16.VERA_CTRL = %10000000      ; reset VERA and palette
+                c64.CINT()      ; back to text mode
+                width = 0
+                height = 0
+                bpp = 0
+                mode = 0
+            }
+        }
+
+        active_mode = mode
+        if bpp
+            clear_screen()
+    }
+
+    sub clear_screen() {
+        monochrome_stipple(false)
+        position(0, 0)
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                repeat 240/2/8
+                    cs_innerloop640()
+            }
+            ; TODO mode 2, 3
+            4 -> {
+                ; lores 256c
+                repeat 240/2
+                    cs_innerloop640()
+            }
+            5 -> {
+                ; highres monochrome
+                repeat 480/8
+                    cs_innerloop640()
+            }
+            6 -> {
+                ; highres 4c
+                repeat 480/4
+                    cs_innerloop640()
+            }
+            ; modes 7 and 8 not supported due to lack of VRAM
+        }
+        position(0, 0)
+    }
+
+    sub monochrome_stipple(ubyte enable) {
+        monochrome_dont_stipple_flag = not enable
+    }
+
+    sub rect(uword x, uword y, uword width, uword height, ubyte color) {
+        if width==0 or height==0
+            return
+        horizontal_line(x, y, width, color)
+        if height==1
+            return
+        horizontal_line(x, y+height-1, width, color)
+        vertical_line(x, y+1, height-2, color)
+        if width==1
+            return
+        vertical_line(x+width-1, y+1, height-2, color)
+    }
+
+    sub fillrect(uword x, uword y, uword width, uword height, ubyte color) {
+        if width==0
+            return
+        repeat height {
+            horizontal_line(x, y, width, color)
+            y++
+        }
+    }
+
+    sub horizontal_line(uword x, uword y, uword length, ubyte color) {
+        if length==0
+            return
+        when active_mode {
+            1, 5 -> {
+                ; monochrome modes, either resolution
+                ubyte separate_pixels = (8-lsb(x)) & 7
+                if separate_pixels as uword > length
+                    separate_pixels = lsb(length)
+                repeat separate_pixels {
+                    ; TODO optimize this by writing a masked byte in 1 go
+                    plot(x, y, color)
+                    x++
+                }
+                length -= separate_pixels
+                if length {
+                    position(x, y)
+                    separate_pixels = lsb(length) & 7
+                    x += length & $fff8
+                    %asm {{
+                        lsr  length+1
+                        ror  length
+                        lsr  length+1
+                        ror  length
+                        lsr  length+1
+                        ror  length
+                        lda  color
+                        bne  +
+                        ldy  #0     ; black
+                        bra  _loop
++                       lda  monochrome_dont_stipple_flag
+                        beq  _stipple
+                        ldy  #255       ; don't stipple
+                        bra  _loop
+_stipple                lda  y
+                        and  #1         ; determine stipple pattern to use
+                        bne  +
+                        ldy  #%01010101
+                        bra  _loop
++                       ldy  #%10101010
+_loop                   lda  length
+                        ora  length+1
+                        beq  _done
+                        sty  cx16.VERA_DATA0
+                        lda  length
+                        bne  +
+                        dec  length+1
++                       dec  length
+                        bra  _loop
+_done
+                    }}
+                    repeat separate_pixels {
+                        ; TODO optimize this by writing a masked byte in 1 go
+                        plot(x, y, color)
+                        x++
+                    }
+                }
+                cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off again
+            }
+            4 -> {
+                ; lores 256c
+                position(x, y)
+                %asm {{
+                    lda  color
+                    phx
+                    ldx  length+1
+                    beq  +
+                    ldy  #0
+-                   sta  cx16.VERA_DATA0
+                    iny
+                    bne  -
+                    dex
+                    bne  -
++                   ldy  length     ; remaining
+                    beq  +
+-                   sta  cx16.VERA_DATA0
+                    dey
+                    bne  -
++                   plx
+                }}
+            }
+            6 -> {
+                ; highres 4c
+                ; TODO also mostly usable for lores 4c?
+                color &= 3
+                ubyte[4] colorbits
+                ubyte ii
+                for ii in 3 downto 0 {
+                    colorbits[ii] = color
+                    color <<= 2
+                }
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                %asm {{
+                    lda  cx16.VERA_ADDR_H
+                    and  #%00000111         ; no auto advance
+                    sta  cx16.VERA_ADDR_H
+                    stz  cx16.VERA_CTRL     ; setup vera addr 0
+                    lda  cx16.r1
+                    and  #1
+                    sta  cx16.VERA_ADDR_H
+                    lda  cx16.r0
+                    sta  cx16.VERA_ADDR_L
+                    lda  cx16.r0+1
+                    sta  cx16.VERA_ADDR_M
+                    phx
+                    ldx  x
+                }}
+
+                repeat length {
+                    %asm {{
+                        txa
+                        and  #3
+                        tay
+                        lda  cx16.VERA_DATA0
+                        and  gfx2.plot.mask4c,y
+                        ora  colorbits,y
+                        sta  cx16.VERA_DATA0
+                        cpy  #%00000011         ; next vera byte?
+                        bne  +
+                        inc  cx16.VERA_ADDR_L
+                        bne  +
+                        inc  cx16.VERA_ADDR_M
++                       bne  +
+                        inc  cx16.VERA_ADDR_H
++                       inx                     ; next pixel
+                    }}
+                }
+
+                %asm {{
+                    plx
+                }}
+            }
+        }
+    }
+
+    sub vertical_line(uword x, uword y, uword height, ubyte color) {
+        position(x,y)
+        when active_mode {
+            1, 5 -> {
+                ; monochrome, either resolution
+                ; note for the 1 bpp modes we can't use vera's auto increment mode because we have to 'or' the pixel data in place.
+                ; TODO use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment IS possible
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                cx16.r15 = gfx2.plot.bits[x as ubyte & 7]           ; bitmask
+                if active_mode>=5
+                    cx16.r14 = 640/8
+                else
+                    cx16.r14 = 320/8
+                if color {
+                    if monochrome_dont_stipple_flag {
+                        repeat height {
+                            %asm {{
+                                lda  cx16.VERA_DATA0
+                                ora  cx16.r15
+                                sta  cx16.VERA_DATA0
+                                lda  cx16.VERA_ADDR_L
+                                clc
+                                adc  cx16.r14                 ; advance vera ptr to go to the next line
+                                sta  cx16.VERA_ADDR_L
+                                lda  cx16.VERA_ADDR_M
+                                adc  #0
+                                sta  cx16.VERA_ADDR_M
+                                ; lda  cx16.VERA_ADDR_H     ; the bitmap size is small enough to not have to deal with the _H part.
+                                ; adc  #0
+                                ; sta  cx16.VERA_ADDR_H
+                            }}
+                        }
+                    } else {
+                        ; stippling.
+                        height = (height+1)/2       ; TODO is the line sometimes 1 pixel too long now because of rounding?
+                        %asm {{
+                            lda x
+                            eor y
+                            and #1
+                            bne +
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14                ; advance vera ptr to go to the next line for correct stipple pattern
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+        +
+                            asl  cx16.r14
+                            ldy  height
+                            beq  +
+        -                   lda  cx16.VERA_DATA0
+                            ora  cx16.r15
+                            sta  cx16.VERA_DATA0
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14               ; advance vera data ptr to go to the next-next line
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+                            ; lda  cx16.VERA_ADDR_H      ; the bitmap size is small enough to not have to deal with the _H part.
+                            ; adc  #0
+                            ; sta  cx16.VERA_ADDR_H
+                            dey
+                            bne  -
+        +
+                        }}
+                    }
+                } else {
+                    cx16.r15 = ~cx16.r15
+                    repeat height {
+                        %asm {{
+                            lda  cx16.VERA_DATA0
+                            and  cx16.r15
+                            sta  cx16.VERA_DATA0
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14             ; advance vera data ptr to go to the next line
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+                            ; the bitmap size is small enough to not have to deal with the _H part:
+                            ; lda  cx16.VERA_ADDR_H
+                            ; adc  #0
+                            ; sta  cx16.VERA_ADDR_H
+                        }}
+                    }
+                }
+            }
+            4 -> {
+                ; lores 256c
+                ; set vera auto-increment to 320 pixel increment (=next line)
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
+                %asm {{
+                    ldy  height
+                    beq  +
+                    lda  color
+-                   sta  cx16.VERA_DATA0
+                    dey
+                    bne  -
++
+                }}
+            }
+            6 -> {
+                ; highres 4c
+                ; note for this mode we can't use vera's auto increment mode because we have to 'or' the pixel data in place.
+                ; TODO use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment IS possible
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                ; TODO also mostly usable for lores 4c?
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+
+                ; TODO optimize this vertical line loop in pure assembly
+                color &= 3
+                color <<= gfx2.plot.shift4c[lsb(x) & 3]
+                ubyte mask = gfx2.plot.mask4c[lsb(x) & 3]
+                repeat height {
+                    ubyte value = cx16.vpeek(lsb(cx16.r1), cx16.r0) & mask | color
+                    cx16.vpoke(lsb(cx16.r1), cx16.r0, value)
+                    %asm {{
+                        ; 24 bits add 160 (640/4)
+                        clc
+                        lda  cx16.r0
+                        adc  #640/4
+                        sta  cx16.r0
+                        lda  cx16.r0+1
+                        adc  #0
+                        sta  cx16.r0+1
+                        bcc  +
+                        inc  cx16.r1
++
+                    }}
+                }
+            }
+        }
+
+    }
+
+    sub line(uword @zp x1, uword @zp y1, uword @zp x2, uword @zp y2, ubyte color) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        if y1>y2 {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            swap(x1, x2)
+            swap(y1, y2)
+        }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as uword +1, color)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1, color)
+            return
+        }
+
+        word @zp d = 0
+        cx16.r13 = true      ; 'positive_ix'
+        if dx < 0 {
+            dx = -dx
+            cx16.r13 = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+        cx16.r14 = x1       ; internal plot X
+
+        if dx >= dy {
+            if cx16.r13 {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14++
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14--
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if cx16.r13 {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+    }
+
+    sub circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, ubyte color) {
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - yy
+            plot(cx16.r14, cx16.r15, color)
+
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                xx--
+                decisionOver2 += (yy as word -xx)*2+1
+            }
+        }
+    }
+
+    sub disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
+            }
+        }
+    }
+
+    sub plot(uword @zp x, uword y, ubyte color) {
+        ubyte[8] bits = [128, 64, 32, 16, 8, 4, 2, 1]
+        ubyte[4] mask4c = [%00111111, %11001111, %11110011, %11111100]
+        ubyte[4] shift4c = [6,4,2,0]
+        uword addr
+        ubyte value
+
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    addr = x/8 + y*(320/8)
+                    value = bits[lsb(x)&7]
+                    if color
+                        cx16.vpoke_or(0, addr, value)
+                    else {
+                        value = ~value
+                        cx16.vpoke_and(0, addr, value)
+                    }
+                }
+            }
+            4 -> {
+                ; lores 256c
+                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                cx16.vpoke(lsb(cx16.r1), cx16.r0, color)
+                ; activate vera auto-increment mode so next_pixel() can be used after this
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | %00010000
+                color = cx16.VERA_DATA0
+            }
+            5 -> {
+                ; highres monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    addr = x/8 + y*(640/8)
+                    value = bits[lsb(x)&7]
+                    if color
+                        cx16.vpoke_or(0, addr, value)
+                    else {
+                        value = ~value
+                        cx16.vpoke_and(0, addr, value)
+                    }
+                }
+            }
+            6 -> {
+                ; highres 4c
+                ; TODO also mostly usable for lores 4c?
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                color &= 3
+                color <<= shift4c[lsb(x) & 3]
+                ; TODO optimize the vera memory manipulation in pure assembly
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                value = cx16.vpeek(lsb(cx16.r1), cx16.r0) & mask4c[lsb(x) & 3] | color
+                cx16.vpoke(lsb(cx16.r1), cx16.r0, value)
+            }
+        }
+    }
+
+    sub position(uword @zp x, uword y) {
+        ubyte bank
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                cx16.r0 = y*(320/8) + x/8
+                cx16.vaddr(0, cx16.r0, 0, 1)
+            }
+            ; TODO modes 2,3
+            4 -> {
+                ; lores 256c
+                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                bank = lsb(cx16.r1)
+                cx16.vaddr(bank, cx16.r0, 0, 1)
+            }
+            5 -> {
+                ; highres monochrome
+                cx16.r0 = y*(640/8) + x/8
+                cx16.vaddr(0, cx16.r0, 0, 1)
+            }
+            6 -> {
+                ; highres 4c
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                bank = lsb(cx16.r1)
+                cx16.vaddr(bank, cx16.r0, 0, 1)
+            }
+        }
+    }
+
+    inline asmsub next_pixel(ubyte color @A) {
+        ; -- sets the next pixel byte to the graphics chip.
+        ;    for 8 bpp screens this will plot 1 pixel.
+        ;    for 1 bpp screens it will plot 8 pixels at once (color = bit pattern).
+        ;    for 2 bpp screens it will plot 4 pixels at once (color = bit pattern).
+        %asm {{
+            sta  cx16.VERA_DATA0
+        }}
+    }
+
+    asmsub next_pixels(uword pixels @AY, uword amount @R0) clobbers(A, Y)  {
+        ; -- sets the next bunch of pixels from a prepared array of bytes.
+        ;    for 8 bpp screens this will plot 1 pixel per byte.
+        ;    for 1 bpp screens it will plot 8 pixels at once (colors are the bit patterns per byte).
+        ;    for 2 bpp screens it will plot 4 pixels at once (colors are the bit patterns per byte).
+        %asm {{
+            phx
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r0+1
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1       ; next page of 256 pixels
+            dex
+            bne  -
+
++           ldx  cx16.r0           ; remaining pixels
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            dex
+            bne  -
++           plx
+        }}
+    }
+
+    asmsub set_8_pixels_from_bits(ubyte bits @R0, ubyte oncolor @A, ubyte offcolor @Y) {
+        ; this is only useful in 256 color mode where one pixel equals one byte value.
+        %asm {{
+            phx
+            ldx  #8
+-           asl  cx16.r0
+            bcc  +
+            sta  cx16.VERA_DATA0
+            bra  ++
++           sty  cx16.VERA_DATA0
++           dex
+            bne  -
+            plx
+            rts
+        }}
+    }
+
+    const ubyte charset_orig_bank = $0
+    const uword charset_orig_addr = $f800        ; in bank 0, so $0f800
+    const ubyte charset_bank = $1
+    const uword charset_addr = $f000       ; in bank 1, so $1f000
+
+    sub text_charset(ubyte charset) {
+        ; -- make a copy of the selected character set to use with text()
+        ;    the charset number is the same as for the cx16.screen_set_charset() ROM function.
+        ;    1 = ISO charset, 2 = PETSCII uppercase+graphs, 3= PETSCII uppercase+lowercase.
+        cx16.screen_set_charset(charset, 0)
+        cx16.vaddr(charset_orig_bank, charset_orig_addr, 0, 1)
+        cx16.vaddr(charset_bank, charset_addr, 1, 1)
+        repeat 256*8 {
+            cx16.VERA_DATA1 = cx16.VERA_DATA0
+        }
+    }
+
+    sub text(uword @zp x, uword y, ubyte color, uword sctextptr) {
+        ; -- Write some text at the given pixel position. The text string must be in screencode encoding (not petscii!).
+        ;    You must also have called text_charset() first to select and prepare the character set to use.
+        ;    NOTE: in monochrome (1bpp) screen modes, x position is currently constrained to mulitples of 8 !  TODO allow per-pixel horizontal positioning
+        uword chardataptr
+        when active_mode {
+            1, 5 -> {
+                ; monochrome mode, either resolution
+                cx16.r2 = 40
+                if active_mode>=5
+                    cx16.r2 = 80
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
+                    position(x,y)
+                    %asm {{
+                        lda  cx16.VERA_ADDR_H
+                        and  #%111              ; don't auto-increment, we have to do that manually because of the ora
+                        sta  cx16.VERA_ADDR_H
+                        lda  color
+                        sta  P8ZP_SCRATCH_B1
+                        ldy  #8
+-                       lda  P8ZP_SCRATCH_B1
+                        bne  +                  ; white color, plot normally
+                        lda  cx16.VERA_DATA1
+                        eor  #255               ; black color, keep only the other pixels
+                        and  cx16.VERA_DATA0
+                        bra  ++
++                       lda  cx16.VERA_DATA0
+                        ora  cx16.VERA_DATA1
++                       sta  cx16.VERA_DATA0
+                        lda  cx16.VERA_ADDR_L
+                        clc
+                        adc  cx16.r2
+                        sta  cx16.VERA_ADDR_L
+                        bcc  +
+                        inc  cx16.VERA_ADDR_M
++                       lda  x
+                        clc
+                        adc  #1
+                        sta  x
+                        bcc  +
+                        inc  x+1
++                       dey
+                        bne  -
+                    }}
+                    sctextptr++
+                }
+            }
+            4 -> {
+                ; lores 256c
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
+                    repeat 8 {
+                        ; TODO rewrite this inner loop fully in assembly
+                        position(x,y)
+                        y++
+                        %asm {{
+                            phx
+                            ldx  #1
+                            lda  cx16.VERA_DATA1
+                            sta  P8ZP_SCRATCH_B1
+                            ldy  #8
+-                           asl  P8ZP_SCRATCH_B1
+                            bcc  +
+                            stx  cx16.VERA_DATA0    ; write a pixel
+                            bra  ++
++                           lda  cx16.VERA_DATA0    ; don't write a pixel, but do advance to the next address
++                           dey
+                            bne  -
+                            plx
+                        }}
+                    }
+                    x+=8
+                    y-=8
+                    sctextptr++
+                }
+            }
+            6 -> {
+                ; hires 4c
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    repeat 8 {
+                        ; TODO rewrite this inner loop fully in assembly
+                        ubyte charbits = cx16.vpeek(charset_bank, chardataptr)
+                        repeat 8 {
+                            charbits <<= 1
+                            if_cs
+                                plot(x, y, color)
+                            x++
+                        }
+                        x-=8
+                        chardataptr++
+                        y++
+                    }
+                    x+=8
+                    y-=8
+                    sctextptr++
+                }
+            }
+        }
+    }
+
+    asmsub cs_innerloop640() clobbers(Y) {
+        %asm {{
+            ldy  #80
+-           stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            dey
+            bne  -
+            rts
+        }}
+    }
+
+    asmsub addr_mul_24_for_highres_4c(uword yy @R2, uword xx @R3)  clobbers(A, Y) -> uword @R0, uword @R1 {
+        ; yy * 160 + xx/4  (24 bits calculation)
+        ; 24 bits result is in r0 and r1L (highest byte)
+        %asm {{
+            ldy  #5
+-           asl  cx16.r2
+            rol  cx16.r2+1
+            dey
+            bne  -
+            lda  cx16.r2
+            sta  cx16.r0
+            lda  cx16.r2+1
+            sta  cx16.r0+1
+            asl  cx16.r0
+            rol  cx16.r0+1
+            asl  cx16.r0
+            rol  cx16.r0+1
+
+            ; xx >>= 2  (xx=R3)
+            lsr  cx16.r3+1
+            ror  cx16.r3
+            lsr  cx16.r3+1
+            ror  cx16.r3
+
+            ; add r2 and xx (r3) to r0 (24-bits)
+            stz  cx16.r1
+            clc
+            lda  cx16.r0
+            adc  cx16.r2
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  cx16.r2+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           clc
+            lda  cx16.r0
+            adc  cx16.r3
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  cx16.r3+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++
+            rts
+        }}
+    }
+
+    asmsub addr_mul_24_for_lores_256c(uword yy @R0, uword xx @AY) clobbers(A) -> uword @R0, ubyte @R1  {
+        ; yy * 320 + xx (24 bits calculation)
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r0
+            sta  P8ZP_SCRATCH_B1
+            lda  cx16.r0+1
+            sta  cx16.r1
+            sta  P8ZP_SCRATCH_REG
+            lda  cx16.r0
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            sta  cx16.r0
+            lda  P8ZP_SCRATCH_B1
+            clc
+            adc  P8ZP_SCRATCH_REG
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           ; now add the value to this 24-bits number
+            lda  cx16.r0
+            clc
+            adc  P8ZP_SCRATCH_W1
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  P8ZP_SCRATCH_W1+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           lda  cx16.r1
+            rts
+        }}
+    }
+
+}

compiler/res/prog8lib/cx16/graphics.p8

@@ -0,0 +1,138 @@
+%target cx16
+%import syslib
+%import textio
+
+; Bitmap pixel graphics module for the CommanderX16
+; wraps the graphics functions that are in ROM.
+; only black/white monochrome 320x200 for now. (i.e. truncated at the bottom)
+; For full-screen 640x480 or 320x240 graphics, use the "gfx2" module instead. (but that is Cx16-specific)
+; Note: there is no color palette manipulation here, you have to do that yourself or use the "palette" module.
+
+
+graphics {
+    const uword WIDTH = 320
+    const ubyte HEIGHT = 200
+
+    sub enable_bitmap_mode() {
+        ; enable bitmap screen, erase it and set colors to black/white.
+        void cx16.screen_set_mode($80)
+        cx16.GRAPH_init(0)
+        clear_screen(1, 0)
+    }
+
+    sub disable_bitmap_mode() {
+        ; enables text mode, erase the text screen, color white
+        void cx16.screen_set_mode(2)
+        txt.fill_screen(' ', 1)     ; doesn't seem to fully clear the text screen after returning from gfx mode
+    }
+
+
+    sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
+        cx16.GRAPH_set_colors(pixelcolor, pixelcolor, bgcolor)
+        cx16.GRAPH_clear()
+    }
+
+    sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
+        cx16.GRAPH_draw_line(x1, y1, x2, y2)
+    }
+
+    sub fillrect(uword x, uword y, uword width, uword height) {
+        cx16.GRAPH_draw_rect(x, y, width, height, 0, 1)
+    }
+
+    sub rect(uword x, uword y, uword width, uword height) {
+        cx16.GRAPH_draw_rect(x, y, width, height, 0, 0)
+    }
+
+    sub horizontal_line(uword x, uword y, uword length) {
+        if length
+            cx16.GRAPH_draw_line(x, y, x+length-1, y)
+    }
+
+    sub vertical_line(uword x, uword y, uword height) {
+        if height
+            cx16.GRAPH_draw_line(x, y, x, y+height-1)
+    }
+
+    sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
+        ;cx16.r0 = xcenter - radius/2
+        ;cx16.r1 = ycenter - radius/2
+        ;cx16.r2 = radius*2
+        ;cx16.r3 = radius*2
+        ;cx16.GRAPH_draw_oval(false)          ; currently this call is not implemented on cx16, does a BRK
+
+        ; Midpoint algorithm
+        if radius==0
+            return
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+
+        while xx>=yy {
+            cx16.r0 = xcenter + xx
+            cx16.r1 = ycenter + yy
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter - xx
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter + xx
+            cx16.r1 = ycenter - yy
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter - xx
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter + yy
+            cx16.r1 = ycenter + xx
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter - yy
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter + yy
+            cx16.r1 = ycenter - xx
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            cx16.r0 = xcenter - yy
+            cx16.FB_cursor_position2()
+            cx16.FB_set_pixel(1)
+            yy++
+            if decisionOver2<=0 {
+                decisionOver2 += (yy as word)*2+1
+            } else {
+                xx--
+                decisionOver2 += (yy as word -xx)*2+1
+            }
+        }
+    }
+
+    sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
+            }
+        }
+    }
+
+    inline asmsub  plot(uword plotx @R0, uword ploty @R1) clobbers(A, X, Y) {
+        %asm {{
+            jsr  cx16.FB_cursor_position
+            lda  #1
+            jsr  cx16.FB_set_pixel
+        }}
+    }
+}

compiler/res/prog8lib/cx16/palette.p8

@@ -0,0 +1,183 @@
+%target cx16
+
+; Manipulate the Commander X16's display color palette.
+; Should you want to restore the default palette, you have to reinitialize the Vera yourself.
+
+palette {
+
+    uword vera_palette_ptr
+    ubyte c
+
+    sub set_color(ubyte index, uword color) {
+        vera_palette_ptr = $fa00+index*2
+        cx16.vpoke(1, vera_palette_ptr, lsb(color))
+        vera_palette_ptr++
+        cx16.vpoke(1, vera_palette_ptr, msb(color))
+    }
+
+    sub set_rgb4(uword palette_bytes_ptr, uword num_colors) {
+        ; 2 bytes per color entry, the Vera uses this, but the R/GB bytes order is swapped
+        vera_palette_ptr = $fa00
+        repeat num_colors {
+            cx16.vpoke(1, vera_palette_ptr+1, @(palette_bytes_ptr))
+            palette_bytes_ptr++
+            cx16.vpoke(1, vera_palette_ptr, @(palette_bytes_ptr))
+            palette_bytes_ptr++
+            vera_palette_ptr+=2
+        }
+    }
+
+    sub set_rgb(uword palette_words_ptr, uword num_colors) {
+        ; 1 word per color entry (in little endian format so $gb0r)
+        vera_palette_ptr = $fa00
+        repeat num_colors*2 {
+            cx16.vpoke(1, vera_palette_ptr, @(palette_words_ptr))
+            palette_words_ptr++
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_rgb8(uword palette_bytes_ptr, uword num_colors) {
+        ; 3 bytes per color entry, adjust color depth from 8 to 4 bits per channel.
+        vera_palette_ptr = $fa00
+        ubyte red
+        ubyte greenblue
+        repeat num_colors {
+            red = @(palette_bytes_ptr) >> 4
+            palette_bytes_ptr++
+            greenblue = @(palette_bytes_ptr) & %11110000
+            palette_bytes_ptr++
+            greenblue |= @(palette_bytes_ptr) >> 4    ; add Blue
+            palette_bytes_ptr++
+            cx16.vpoke(1, vera_palette_ptr, greenblue)
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, red)
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_monochrome(uword screencolorRGB, uword drawcolorRGB) {
+        vera_palette_ptr = $fa00
+        cx16.vpoke(1, vera_palette_ptr, lsb(screencolorRGB))   ; G,B
+        vera_palette_ptr++
+        cx16.vpoke(1, vera_palette_ptr, msb(screencolorRGB))   ; R
+        vera_palette_ptr++
+        repeat 255 {
+            cx16.vpoke(1, vera_palette_ptr, lsb(drawcolorRGB)) ; G,B
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, msb(drawcolorRGB)) ; R
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_grayscale() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            c=0
+            repeat 16 {
+                cx16.vpoke(1, vera_palette_ptr, c)
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, c)
+                vera_palette_ptr++
+                c += $11
+            }
+        }
+    }
+
+    uword[] C64_colorpalette_dark = [   ; this is a darker palette with more contrast
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $632,  ; 2 = red
+        $7AB,  ; 3 = cyan
+        $638,  ; 4 = purple
+        $584,  ; 5 = green
+        $327,  ; 6 = blue
+        $BC6,  ; 7 = yellow
+        $642,  ; 8 = orange
+        $430,  ; 9 = brown
+        $965,  ; 10 = light red
+        $444,  ; 11 = dark grey
+        $666,  ; 12 = medium grey
+        $9D8,  ; 13 = light green
+        $65B,  ; 14 = light blue
+        $999   ; 15 = light grey
+    ]
+
+    uword[] C64_colorpalette_pepto = [  ; # this is Pepto's Commodore-64 palette  http://www.pepto.de/projects/colorvic/
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $833,  ; 2 = red
+        $7cc,  ; 3 = cyan
+        $839,  ; 4 = purple
+        $5a4,  ; 5 = green
+        $229,  ; 6 = blue
+        $ef7,  ; 7 = yellow
+        $852,  ; 8 = orange
+        $530,  ; 9 = brown
+        $c67,  ; 10 = light red
+        $444,  ; 11 = dark grey
+        $777,  ; 12 = medium grey
+        $af9,  ; 13 = light green
+        $76e,  ; 14 = light blue
+        $bbb   ; 15 = light grey
+    ]
+
+    uword[] C64_colorpalette_light = [  ; this is a lighter palette
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $944,  ; 2 = red
+        $7CC,  ; 3 = cyan
+        $95A,  ; 4 = purple
+        $6A5,  ; 5 = green
+        $549,  ; 6 = blue
+        $CD8,  ; 7 = yellow
+        $963,  ; 8 = orange
+        $650,  ; 9 = brown
+        $C77,  ; 10 = light red
+        $666,  ; 11 = dark grey
+        $888,  ; 12 = medium grey
+        $AE9,  ; 13 = light green
+        $87C,  ; 14 = light blue
+        $AAA   ; 15 = light grey
+    ]
+
+    sub set_c64pepto() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_pepto[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+    sub set_c64light() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_light[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+    sub set_c64dark() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_dark[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+}

compiler/res/prog8lib/cx16/syslib.p8

@@ -0,0 +1,822 @@
+; Prog8 definitions for the CommanderX16
+; Including memory registers, I/O registers, Basic and Kernal subroutines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target cx16
+
+
+c64 {
+
+; ---- kernal routines, these are the same as on the Commodore-64 (hence the same block name) ----
+
+; STROUT --> use txt.print
+; CLEARSCR -> use txt.clear_screen
+; HOMECRSR -> use txt.plot
+
+romsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
+romsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
+romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
+romsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
+romsub $FF8D = VECTOR(uword userptr @ XY, ubyte dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
+romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
+romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
+romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
+romsub $FF99 = MEMTOP(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer.   NOTE: as a Cx16 extension, also returns the number of RAM memory banks in register A !  See MEMTOP2
+romsub $FF9C = MEMBOT(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
+romsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
+romsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
+romsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
+romsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
+romsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
+romsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
+romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
+romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
+romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word
+romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
+romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
+romsub $FFC0 = OPEN() clobbers(X,Y) -> ubyte @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
+romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
+romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> ubyte @Pc    ; (via 798 ($31E)) define an input channel
+romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
+romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
+romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
+romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
+romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
+romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
+romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
+romsub $FFE1 = STOP() clobbers(X) -> ubyte @ Pz, ubyte @ A      ; (via 808 ($328)) check the STOP key (and some others in A)
+romsub $FFE4 = GETIN() clobbers(X,Y) -> ubyte @Pc, ubyte @ A    ; (via 810 ($32A)) get a character
+romsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
+romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; read number of screen rows and columns
+romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, ubyte dir @ Pc) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
+romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
+
+; ---- utility
+
+asmsub STOP2() -> ubyte @A  {
+    ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
+    %asm {{
+        phx
+        jsr  c64.STOP
+        beq  +
+        plx
+        lda  #0
+        rts
++       plx
+        lda  #1
+        rts
+    }}
+}
+
+asmsub RDTIM16() -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    %asm {{
+        phx
+        jsr  c64.RDTIM
+        pha
+        txa
+        tay
+        pla
+        plx
+        rts
+    }}
+}
+
+asmsub MEMTOP2() -> ubyte @A {
+    ; -- uses MEMTOP's cx16 extension to query the number of available RAM banks.
+    %asm {{
+        phx
+        sec
+        jsr  c64.MEMTOP
+        plx
+        rts
+    }}
+}
+
+}
+
+cx16 {
+
+; irq and hardware vectors:
+    &uword  CINV            = $0314     ; IRQ vector (in ram)
+    &uword  NMI_VEC         = $FFFA     ; 65c02 nmi vector, determined by the kernal if banked in
+    &uword  RESET_VEC       = $FFFC     ; 65c02 reset vector, determined by the kernal if banked in
+    &uword  IRQ_VEC         = $FFFE     ; 65c02 interrupt vector, determined by the kernal if banked in
+
+
+; the sixteen virtual 16-bit registers
+    &uword r0  = $0002
+    &uword r1  = $0004
+    &uword r2  = $0006
+    &uword r3  = $0008
+    &uword r4  = $000a
+    &uword r5  = $000c
+    &uword r6  = $000e
+    &uword r7  = $0010
+    &uword r8  = $0012
+    &uword r9  = $0014
+    &uword r10 = $0016
+    &uword r11 = $0018
+    &uword r12 = $001a
+    &uword r13 = $001c
+    &uword r14 = $001e
+    &uword r15 = $0020
+
+; VERA registers
+
+    const uword VERA_BASE       = $9F20
+    &ubyte  VERA_ADDR_L         = VERA_BASE + $0000
+    &ubyte  VERA_ADDR_M         = VERA_BASE + $0001
+    &ubyte  VERA_ADDR_H         = VERA_BASE + $0002
+    &ubyte  VERA_DATA0          = VERA_BASE + $0003
+    &ubyte  VERA_DATA1          = VERA_BASE + $0004
+    &ubyte  VERA_CTRL           = VERA_BASE + $0005
+    &ubyte  VERA_IEN            = VERA_BASE + $0006
+    &ubyte  VERA_ISR            = VERA_BASE + $0007
+    &ubyte  VERA_IRQ_LINE_L     = VERA_BASE + $0008
+    &ubyte  VERA_DC_VIDEO       = VERA_BASE + $0009
+    &ubyte  VERA_DC_HSCALE      = VERA_BASE + $000A
+    &ubyte  VERA_DC_VSCALE      = VERA_BASE + $000B
+    &ubyte  VERA_DC_BORDER      = VERA_BASE + $000C
+    &ubyte  VERA_DC_HSTART      = VERA_BASE + $0009
+    &ubyte  VERA_DC_HSTOP       = VERA_BASE + $000A
+    &ubyte  VERA_DC_VSTART      = VERA_BASE + $000B
+    &ubyte  VERA_DC_VSTOP       = VERA_BASE + $000C
+    &ubyte  VERA_L0_CONFIG      = VERA_BASE + $000D
+    &ubyte  VERA_L0_MAPBASE     = VERA_BASE + $000E
+    &ubyte  VERA_L0_TILEBASE    = VERA_BASE + $000F
+    &ubyte  VERA_L0_HSCROLL_L   = VERA_BASE + $0010
+    &ubyte  VERA_L0_HSCROLL_H   = VERA_BASE + $0011
+    &ubyte  VERA_L0_VSCROLL_L   = VERA_BASE + $0012
+    &ubyte  VERA_L0_VSCROLL_H   = VERA_BASE + $0013
+    &ubyte  VERA_L1_CONFIG      = VERA_BASE + $0014
+    &ubyte  VERA_L1_MAPBASE     = VERA_BASE + $0015
+    &ubyte  VERA_L1_TILEBASE    = VERA_BASE + $0016
+    &ubyte  VERA_L1_HSCROLL_L   = VERA_BASE + $0017
+    &ubyte  VERA_L1_HSCROLL_H   = VERA_BASE + $0018
+    &ubyte  VERA_L1_VSCROLL_L   = VERA_BASE + $0019
+    &ubyte  VERA_L1_VSCROLL_H   = VERA_BASE + $001A
+    &ubyte  VERA_AUDIO_CTRL     = VERA_BASE + $001B
+    &ubyte  VERA_AUDIO_RATE     = VERA_BASE + $001C
+    &ubyte  VERA_AUDIO_DATA     = VERA_BASE + $001D
+    &ubyte  VERA_SPI_DATA       = VERA_BASE + $001E
+    &ubyte  VERA_SPI_CTRL       = VERA_BASE + $001F
+; VERA_PSG_BASE     = $1F9C0
+; VERA_PALETTE_BASE = $1FA00
+; VERA_SPRITES_BASE = $1FC00
+
+; I/O
+
+    const uword  via1   = $9f60                  ;VIA 6522 #1
+    &ubyte  d1prb	= via1+0
+    &ubyte  d1pra	= via1+1
+    &ubyte  d1ddrb	= via1+2
+    &ubyte  d1ddra	= via1+3
+    &ubyte  d1t1l	= via1+4
+    &ubyte  d1t1h	= via1+5
+    &ubyte  d1t1ll	= via1+6
+    &ubyte  d1t1lh	= via1+7
+    &ubyte  d1t2l	= via1+8
+    &ubyte  d1t2h	= via1+9
+    &ubyte  d1sr	= via1+10
+    &ubyte  d1acr	= via1+11
+    &ubyte  d1pcr	= via1+12
+    &ubyte  d1ifr	= via1+13
+    &ubyte  d1ier	= via1+14
+    &ubyte  d1ora	= via1+15
+
+    const uword  via2   = $9f70                  ;VIA 6522 #2
+    &ubyte  d2prb	= via2+0
+    &ubyte  d2pra	= via2+1
+    &ubyte  d2ddrb	= via2+2
+    &ubyte  d2ddra	= via2+3
+    &ubyte  d2t1l	= via2+4
+    &ubyte  d2t1h	= via2+5
+    &ubyte  d2t1ll	= via2+6
+    &ubyte  d2t1lh	= via2+7
+    &ubyte  d2t2l	= via2+8
+    &ubyte  d2t2h	= via2+9
+    &ubyte  d2sr	= via2+10
+    &ubyte  d2acr	= via2+11
+    &ubyte  d2pcr	= via2+12
+    &ubyte  d2ifr	= via2+13
+    &ubyte  d2ier	= via2+14
+    &ubyte  d2ora	= via2+15
+
+
+; ---- Commander X-16 additions on top of C64 kernal routines ----
+; spelling of the names is taken from the Commander X-16 rom sources
+
+; supported C128 additions
+romsub $ff4a = close_all(ubyte device @A)  clobbers(A,X,Y)
+romsub $ff59 = lkupla(ubyte la @A)  clobbers(A,X,Y)
+romsub $ff5c = lkupsa(ubyte sa @Y)  clobbers(A,X,Y)
+romsub $ff5f = screen_set_mode(ubyte mode @A)  clobbers(A, X, Y) -> ubyte @Pc
+romsub $ff62 = screen_set_charset(ubyte charset @A, uword charsetptr @XY)  clobbers(A,X,Y)      ; incompatible with C128  dlchr()
+; not yet supported: romsub $ff65 = pfkey()  clobbers(A,X,Y)
+romsub $ff6e = jsrfar()
+romsub $ff74 = fetch(ubyte bank @X, ubyte index @Y)  clobbers(X) -> ubyte @A
+romsub $ff77 = stash(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)
+romsub $ff7a = cmpare(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)
+romsub $ff7d = primm()
+
+; X16 additions
+romsub $ff44 = macptr()  clobbers(A,X,Y)
+romsub $ff47 = enter_basic(ubyte cold_or_warm @Pc)  clobbers(A,X,Y)
+romsub $ff68 = mouse_config(ubyte shape @A, ubyte scale @X)  clobbers (A, X, Y)
+romsub $ff6b = mouse_get(ubyte zpdataptr @X)  clobbers(A)
+romsub $ff71 = mouse_scan()  clobbers(A, X, Y)
+romsub $ff53 = joystick_scan()  clobbers(A, X, Y)
+romsub $ff56 = joystick_get(ubyte joynr @A) -> ubyte @A, ubyte @X, ubyte @Y
+romsub $ff4d = clock_set_date_time(uword yearmonth @R0, uword dayhours @R1, uword minsecs @R2, ubyte jiffies @R3)  clobbers(A, X, Y)
+romsub $ff50 = clock_get_date_time()  clobbers(A, X, Y)  -> uword @R0, uword @R1, uword @R2, ubyte @R3   ; result registers see clock_set_date_time()
+
+; TODO specify the correct clobbers for alle these functions below, we now assume all 3 regs are clobbered
+
+; high level graphics & fonts
+romsub $ff20 = GRAPH_init(uword vectors @R0)  clobbers(A,X,Y)
+romsub $ff23 = GRAPH_clear()  clobbers(A,X,Y)
+romsub $ff26 = GRAPH_set_window(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
+romsub $ff29 = GRAPH_set_colors(ubyte stroke @A, ubyte fill @X, ubyte background @Y)  clobbers (A,X,Y)
+romsub $ff2c = GRAPH_draw_line(uword x1 @R0, uword y1 @R1, uword x2 @R2, uword y2 @R3)  clobbers(A,X,Y)
+romsub $ff2f = GRAPH_draw_rect(uword x @R0, uword y @R1, uword width @R2, uword height @R3, uword cornerradius @R4, ubyte fill @Pc)  clobbers(A,X,Y)
+romsub $ff32 = GRAPH_move_rect(uword sx @R0, uword sy @R1, uword tx @R2, uword ty @R3, uword width @R4, uword height @R5)  clobbers(A,X,Y)
+romsub $ff35 = GRAPH_draw_oval(uword x @R0, uword y @R1, uword width @R2, uword height @R3, ubyte fill @Pc)  clobbers(A,X,Y)
+romsub $ff38 = GRAPH_draw_image(uword x @R0, uword y @R1, uword ptr @R2, uword width @R3, uword height @R4)  clobbers(A,X,Y)
+romsub $ff3b = GRAPH_set_font(uword fontptr @R0)  clobbers(A,X,Y)
+romsub $ff3e = GRAPH_get_char_size(ubyte baseline @A, ubyte width @X, ubyte height_or_style @Y, ubyte is_control @Pc)  clobbers(A,X,Y)
+romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte char @A)  clobbers(A,X,Y)
+romsub $ff41 = GRAPH_put_next_char(ubyte char @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
+
+; framebuffer
+romsub $fef6 = FB_init()  clobbers(A,X,Y)
+romsub $fef9 = FB_get_info()  clobbers(X,Y) -> byte @A, uword @R0, uword @R1    ; width=r0, height=r1
+romsub $fefc = FB_set_palette(uword pointer @R0, ubyte index @A, ubyte bytecount @X)  clobbers(A,X,Y)
+romsub $feff = FB_cursor_position(uword x @R0, uword y @R1)  clobbers(A,X,Y)
+romsub $feff = FB_cursor_position2()  clobbers(A,X,Y)           ;  alias for the previous routine, but avoiding having to respecify both x and y every time
+romsub $ff02 = FB_cursor_next_line(uword x @R0)  clobbers(A,X,Y)
+romsub $ff05 = FB_get_pixel()  clobbers(X,Y) -> ubyte @A
+romsub $ff08 = FB_get_pixels(uword pointer @R0, uword count @R1)  clobbers(A,X,Y)
+romsub $ff0b = FB_set_pixel(ubyte color @A)  clobbers(A,X,Y)
+romsub $ff0e = FB_set_pixels(uword pointer @R0, uword count @R1)  clobbers(A,X,Y)
+romsub $ff11 = FB_set_8_pixels(ubyte pattern @A, ubyte color @X)  clobbers(A,X,Y)
+romsub $ff14 = FB_set_8_pixels_opaque(ubyte pattern @R0, ubyte mask @A, ubyte color1 @X, ubyte color2 @Y)  clobbers(A,X,Y)
+romsub $ff17 = FB_fill_pixels(uword count @R0, uword pstep @R1, ubyte color @A)  clobbers(A,X,Y)
+romsub $ff1a = FB_filter_pixels(uword pointer @ R0, uword count @R1)  clobbers(A,X,Y)
+romsub $ff1d = FB_move_pixels(uword sx @R0, uword sy @R1, uword tx @R2, uword ty @R3, uword count @R4)  clobbers(A,X,Y)
+
+; misc
+romsub $fef0 = sprite_set_image(uword pixels @R0, uword mask @R1, ubyte bpp @R2, ubyte number @A, ubyte width @X, ubyte height @Y, ubyte apply_mask @Pc)  clobbers(A,X,Y) -> ubyte @Pc
+romsub $fef3 = sprite_set_position(uword x @R0, uword y @R1, ubyte number @A)  clobbers(A,X,Y)
+romsub $fee4 = memory_fill(uword address @R0, uword num_bytes @R1, ubyte value @A)  clobbers(A,X,Y)
+romsub $fee7 = memory_copy(uword source @R0, uword target @R1, uword num_bytes @R2)  clobbers(A,X,Y)
+romsub $feea = memory_crc(uword address @R0, uword num_bytes @R1)  clobbers(A,X,Y) -> uword @R2
+romsub $feed = memory_decompress(uword input @R0, uword output @R1)  clobbers(A,X,Y) -> uword @R1       ; last address +1 is result in R1
+romsub $fedb = console_init(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
+romsub $fede = console_put_char(ubyte char @A, ubyte wrapping @Pc)  clobbers(A,X,Y)
+romsub $fee1 = console_get_char()  clobbers(X,Y) -> ubyte @A
+romsub $fed8 = console_put_image(uword pointer @R0, uword width @R1, uword height @R2)  clobbers(A,X,Y)
+romsub $fed5 = console_set_paging_message(uword msgptr @R0)  clobbers(A,X,Y)
+romsub $fed2 = kbdbuf_put(ubyte key @A)  clobbers(A,X,Y)
+romsub $fecf = entropy_get() -> ubyte @A, ubyte @X, ubyte @Y
+romsub $fecc = monitor()  clobbers(A,X,Y)
+
+; ---- end of kernal routines ----
+
+
+; ---- utilities -----
+
+inline asmsub rombank(ubyte rombank @A) {
+    ; -- set the rom banks
+    %asm {{
+        sta  $01            ; rom bank register (new)
+        sta  cx16.d1prb     ; rom bank register (old)
+    }}
+}
+
+inline asmsub rambank(ubyte rambank @A) {
+    ; -- set the ram bank
+    %asm {{
+        sta  $00            ; ram bank register (new)
+        sta  cx16.d1pra     ; ram bank register (old)
+    }}
+}
+
+asmsub vpeek(ubyte bank @A, uword address @XY) -> ubyte @A {
+        ; -- get a byte from VERA's video memory
+        ;    note: inefficient when reading multiple sequential bytes!
+        %asm {{
+                pha
+                lda  #1
+                sta  cx16.VERA_CTRL
+                pla
+                and  #1
+                sta  cx16.VERA_ADDR_H
+                sty  cx16.VERA_ADDR_M
+                stx  cx16.VERA_ADDR_L
+                lda  cx16.VERA_DATA1
+                rts
+            }}
+}
+
+asmsub vaddr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, byte autoIncrOrDecrByOne @Y) clobbers(A) {
+        ; -- setup the VERA's data address register 0 or 1
+        %asm {{
+            and  #1
+            pha
+            lda  cx16.r1
+            and  #1
+            sta  cx16.VERA_CTRL
+            lda  cx16.r0
+            sta  cx16.VERA_ADDR_L
+            lda  cx16.r0+1
+            sta  cx16.VERA_ADDR_M
+            pla
+            cpy  #0
+            bmi  ++
+            beq  +
+            ora  #%00010000
++           sta  cx16.VERA_ADDR_H
+            rts
++           ora  #%00011000
+            sta  cx16.VERA_ADDR_H
+            rts
+        }}
+}
+
+asmsub vpoke(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers(A) {
+    ; -- write a single byte to VERA's video memory
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        sty  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_or(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers (A) {
+    ; -- or a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        ora  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_and(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers(A) {
+    ; -- and a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        and  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_xor(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers (A) {
+    ; -- xor a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        eor  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vload(str name @R0, ubyte device @Y, ubyte bank @A, uword address @R1) -> ubyte @A {
+    ; -- like the basic command VLOAD "filename",device,bank,address
+    ;    loads a file into video memory in the given bank:address, returns success in A
+    ;    !! NOTE !! the V38 ROMs contain a bug in the LOAD code that makes the load address not work correctly,
+    ;               it works fine when loading from local filesystem
+    %asm {{
+        ; -- load a file into video ram
+        phx
+        pha
+        tya
+        tax
+        lda  #1
+        ldy  #0
+        jsr  c64.SETLFS
+        lda  cx16.r0
+        ldy  cx16.r0+1
+        jsr  prog8_lib.strlen
+        tya
+        ldx  cx16.r0
+        ldy  cx16.r0+1
+        jsr  c64.SETNAM
+        pla
+        clc
+        adc  #2
+        ldx  cx16.r1
+        ldy  cx16.r1+1
+        stz  P8ZP_SCRATCH_B1
+        jsr  c64.LOAD
+        bcs  +
+        inc  P8ZP_SCRATCH_B1
++       jsr  c64.CLRCHN
+        lda  #1
+        jsr  c64.CLOSE
+        plx
+        lda  P8ZP_SCRATCH_B1
+        rts
+    }}
+}
+
+
+sub FB_set_pixels_from_buf(uword buffer, uword count) {
+    %asm {{
+            ; -- This is replacement code for the normal FB_set_pixels subroutine in ROM
+            ;    However that routine contains a bug in the current v38 ROM that makes it crash when count > 255.
+            ;    So the code below replaces that. Once the ROM is patched this routine is no longer necessary.
+            ;    See https://github.com/commanderx16/x16-rom/issues/179
+            phx
+            lda  buffer
+            ldy  buffer+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            jsr  _pixels
+            plx
+            rts
+
+_pixels     lda  count+1
+            beq  +
+            ldx  #0
+-           jsr  _loop
+            inc  P8ZP_SCRATCH_W1+1
+            dec  count+1
+            bne  -
+
++           ldx  count
+_loop       ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            dex
+            bne  -
+            rts
+        }}
+}
+
+; ---- system stuff -----
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    %asm {{
+        sei
+        cld
+        ;stz  $00
+        ;stz  $01
+        ;stz  d1prb      ; select rom bank 0 (enable kernal)
+        lda  #$80
+        sta  VERA_CTRL
+        jsr  c64.IOINIT
+        jsr  c64.RESTOR
+        jsr  c64.CINT
+        lda  #$90       ; black
+        jsr  c64.CHROUT
+        lda  #1         ; swap fg/bg
+        jsr  c64.CHROUT
+        lda  #$9e       ; yellow
+        jsr  c64.CHROUT
+        lda  #147       ; clear screen
+        jsr  c64.CHROUT
+        lda  #0
+        tax
+        tay
+        clc
+        clv
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        sei
+        lda  cx16.CINV
+        sta  restore_irq._orig_irqvec
+        lda  cx16.CINV+1
+        sta  restore_irq._orig_irqvec+1
+        cli
+        rts
+    }}
+}
+
+asmsub  set_irq(uword handler @AY, ubyte useKernal @Pc) clobbers(A)  {
+	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  _use_kernal
+		sei
+		lda  #<_irq_handler
+		sta  cx16.CINV
+		lda  #>_irq_handler
+		sta  cx16.CINV+1
+                lda  cx16.VERA_IEN
+                ora  #%00000001     ; enable the vsync irq
+                sta  cx16.VERA_IEN
+		cli
+		rts
+
+_irq_handler    jsr  _irq_handler_init
+_modified	jsr  $ffff                      ; modified
+		jsr  _irq_handler_end
+		lda  _use_kernal
+		bne  +
+		; end irq processing - don't use kernal's irq handling
+		lda  cx16.VERA_ISR
+		ora  #1
+		sta  cx16.VERA_ISR      ; clear Vera Vsync irq status
+		ply
+		plx
+		pla
+		rti
++		jmp  (restore_irq._orig_irqvec)   ; continue with normal kernal irq routine
+
+_use_kernal     .byte  0
+
+_irq_handler_init
+		; save all zp scratch registers and the X register as these might be clobbered by the irq routine
+		stx  IRQ_X_REG
+		lda  P8ZP_SCRATCH_B1
+		sta  IRQ_SCRATCH_ZPB1
+		lda  P8ZP_SCRATCH_REG
+		sta  IRQ_SCRATCH_ZPREG
+		lda  P8ZP_SCRATCH_W1
+		sta  IRQ_SCRATCH_ZPWORD1
+		lda  P8ZP_SCRATCH_W1+1
+		sta  IRQ_SCRATCH_ZPWORD1+1
+		lda  P8ZP_SCRATCH_W2
+		sta  IRQ_SCRATCH_ZPWORD2
+		lda  P8ZP_SCRATCH_W2+1
+		sta  IRQ_SCRATCH_ZPWORD2+1
+		; stack protector; make sure we don't clobber the top of the evaluation stack
+		dex
+		dex
+		dex
+		dex
+		dex
+		dex
+		cld
+		rts
+
+_irq_handler_end
+		; restore all zp scratch registers and the X register
+		lda  IRQ_SCRATCH_ZPB1
+		sta  P8ZP_SCRATCH_B1
+		lda  IRQ_SCRATCH_ZPREG
+		sta  P8ZP_SCRATCH_REG
+		lda  IRQ_SCRATCH_ZPWORD1
+		sta  P8ZP_SCRATCH_W1
+		lda  IRQ_SCRATCH_ZPWORD1+1
+		sta  P8ZP_SCRATCH_W1+1
+		lda  IRQ_SCRATCH_ZPWORD2
+		sta  P8ZP_SCRATCH_W2
+		lda  IRQ_SCRATCH_ZPWORD2+1
+		sta  P8ZP_SCRATCH_W2+1
+		ldx  IRQ_X_REG
+		rts
+
+IRQ_X_REG		.byte  0
+IRQ_SCRATCH_ZPB1	.byte  0
+IRQ_SCRATCH_ZPREG	.byte  0
+IRQ_SCRATCH_ZPWORD1	.word  0
+IRQ_SCRATCH_ZPWORD2	.word  0
+
+		}}
+}
+
+
+asmsub  restore_irq() clobbers(A) {
+	%asm {{
+	    sei
+	    lda  _orig_irqvec
+	    sta  cx16.CINV
+	    lda  _orig_irqvec+1
+	    sta  cx16.CINV+1
+	    lda  cx16.VERA_IEN
+	    and  #%11110000     ; disable all Vera IRQs
+	    ora  #%00000001     ; enable only the vsync Irq
+	    sta  cx16.VERA_IEN
+	    cli
+	    rts
+_orig_irqvec    .word  0
+        }}
+}
+
+asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
+	%asm {{
+            sta  _modified+1
+            sty  _modified+2
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sei
+            lda  cx16.VERA_IEN
+            and  #%11110000     ; clear other IRQs
+            ora  #%00000010     ; enable the line (raster) irq
+            sta  cx16.VERA_IEN
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            jsr  set_rasterline
+            lda  #<_raster_irq_handler
+            sta  cx16.CINV
+            lda  #>_raster_irq_handler
+            sta  cx16.CINV+1
+            cli
+            rts
+
+_raster_irq_handler
+            jsr  set_irq._irq_handler_init
+_modified   jsr  $ffff                      ; modified
+            jsr  set_irq._irq_handler_end
+            ; end irq processing - don't use kernal's irq handling
+            lda  cx16.VERA_ISR
+            ora  #%00000010
+            sta  cx16.VERA_ISR      ; clear Vera line irq status
+            ply
+            plx
+            pla
+            rti
+        }}
+}
+
+asmsub  set_rasterline(uword line @AY) {
+    %asm {{
+        sta  cx16.VERA_IRQ_LINE_L
+        lda  cx16.VERA_IEN
+        and  #%01111111
+        sta  cx16.VERA_IEN
+        tya
+        lsr  a
+        ror  a
+        and  #%10000000
+        ora  cx16.VERA_IEN
+        sta  cx16.VERA_IEN
+        rts
+    }}
+}
+
+}
+
+
+sys {
+    ; ------- lowlevel system routines --------
+
+    const ubyte target = 16         ;  compilation target specifier.  64 = C64,  16 = CommanderX16.
+
+
+    asmsub reset_system() {
+        ; Soft-reset the system back to Basic prompt.
+        %asm {{
+            sei
+            stz  $01            ; bank the kernal in (new rom bank register)
+            stz  cx16.d1prb     ; bank the kernal in (old rom bank register)
+            jmp  (cx16.RESET_VEC)
+        }}
+    }
+
+    sub wait(uword jiffies) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds)
+        repeat jiffies {
+            ubyte jiff = lsb(c64.RDTIM16())
+            while jiff==lsb(c64.RDTIM16()) {
+                ; wait until 1 jiffy has passed
+            }
+        }
+    }
+
+    inline asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        %asm {{
+            sta  cx16.r2
+            sty  cx16.r2+1
+            jsr  cx16.memory_copy
+        }}
+    }
+
+    inline asmsub memset(uword mem @R0, uword numbytes @R1, ubyte value @A) clobbers(A,X,Y) {
+        %asm {{
+            jsr  cx16.memory_fill
+        }}
+    }
+
+    asmsub memsetw(uword mem @R0, uword numwords @R1, uword value @AY) clobbers (A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            jmp  prog8_lib.memsetw
+        }}
+    }
+
+    inline asmsub rsave() {
+        ; save cpu status flag and all registers A, X, Y.
+        ; see http://6502.org/tutorials/register_preservation.html
+        %asm {{
+            php
+            pha
+            phy
+            phx
+        }}
+    }
+
+    inline asmsub rrestore() {
+        ; restore all registers and cpu status flag
+        %asm {{
+            plx
+            ply
+            pla
+            plp
+        }}
+    }
+
+    inline asmsub read_flags() -> ubyte @A {
+        %asm {{
+            php
+            pla
+        }}
+    }
+
+    inline asmsub clear_carry() {
+        %asm {{
+            clc
+        }}
+    }
+
+    inline asmsub set_carry() {
+        %asm {{
+            sec
+        }}
+    }
+
+    inline asmsub clear_irqd() {
+        %asm {{
+            cli
+        }}
+    }
+
+    inline asmsub set_irqd() {
+        %asm {{
+            sei
+        }}
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            jsr  c64.CLRCHN		; reset i/o channels
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            rts		; return to original caller
+        }}
+    }
+
+    inline asmsub progend() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_end
+            ldy  #>prog8_program_end
+        }}
+    }
+}

compiler/res/prog8lib/cx16/textio.p8

@@ -0,0 +1,749 @@
+; Prog8 definitions for the Text I/O and Screen routines for the CommanderX16
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+; indent format: TABS, size=8
+
+%target cx16
+%import syslib
+%import conv
+
+
+txt {
+
+const ubyte DEFAULT_WIDTH = 80
+const ubyte DEFAULT_HEIGHT = 60
+
+
+sub clear_screen() {
+    txt.chrout(147)
+}
+
+sub home() {
+    txt.chrout(19)
+}
+
+sub nl() {
+    txt.chrout('\n')
+}
+
+sub spc() {
+    txt.chrout(' ')
+}
+
+asmsub column(ubyte col @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given column (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  c64.PLOT
+        tay
+        clc
+        jmp  c64.PLOT
+    }}
+}
+
+asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+	; ---- fill the character screen with the given fill character and character color.
+	%asm {{
+	    sty  _ly+1
+        phx
+        pha
+        jsr  c64.SCREEN             ; get dimensions in X/Y
+        txa
+        lsr  a
+        lsr  a
+        sta  _lx+1
+        stz  cx16.VERA_CTRL
+        lda  #%00010000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 1, bank 0.
+        stz  cx16.VERA_ADDR_L       ; start at (0,0)
+        stz  cx16.VERA_ADDR_M
+        pla
+_lx     ldx  #0                     ; modified
+        phy
+_ly     ldy  #1                     ; modified
+-       sta  cx16.VERA_DATA0
+        sty  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sty  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sty  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sty  cx16.VERA_DATA0
+        dex
+        bne  -
+        ply
+        dey
+        beq  +
+        stz  cx16.VERA_ADDR_L
+        inc  cx16.VERA_ADDR_M       ; next line
+        bra  _lx
++       plx
+        rts
+        }}
+}
+
+asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+	; ---- clear the character screen with the given fill character (leaves colors)
+	;      (assumes screen matrix is at the default address)
+	%asm {{
+        phx
+        pha
+        jsr  c64.SCREEN             ; get dimensions in X/Y
+        txa
+        lsr  a
+        lsr  a
+        sta  _lx+1
+        stz  cx16.VERA_CTRL
+        lda  #%00100000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 2, bank 0.
+        stz  cx16.VERA_ADDR_L       ; start at (0,0)
+        stz  cx16.VERA_ADDR_M
+        pla
+_lx     ldx  #0                     ; modified
+-       sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        dex
+        bne  -
+        dey
+        beq  +
+        stz  cx16.VERA_ADDR_L
+        inc  cx16.VERA_ADDR_M       ; next line
+        bra  _lx
++       plx
+        rts
+        }}
+}
+
+asmsub  clear_screencolors (ubyte color @ A) clobbers(Y)  {
+	; ---- clear the character screen colors with the given color (leaves characters).
+	;      (assumes color matrix is at the default address)
+	%asm {{
+        phx
+        sta  _la+1
+        jsr  c64.SCREEN             ; get dimensions in X/Y
+        txa
+        lsr  a
+        lsr  a
+        sta  _lx+1
+        stz  cx16.VERA_CTRL
+        lda  #%00100000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 2, bank 0.
+        lda  #1
+        sta  cx16.VERA_ADDR_L       ; start at (1,0)
+        stz  cx16.VERA_ADDR_M
+_lx     ldx  #0                     ; modified
+_la     lda  #0                     ; modified
+-       sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        dex
+        bne  -
+        dey
+        beq  +
+        lda  #1
+        sta  cx16.VERA_ADDR_L
+        inc  cx16.VERA_ADDR_M       ; next line
+        bra  _lx
++       plx
+        rts
+        }}
+}
+
+
+ubyte[16] color_to_charcode = [$90,$05,$1c,$9f,$9c,$1e,$1f,$9e,$81,$95,$96,$97,$98,$99,$9a,$9b]
+
+sub color (ubyte txtcol) {
+    txtcol &= 15
+    c64.CHROUT(color_to_charcode[txtcol])
+}
+
+sub color2 (ubyte txtcol, ubyte bgcol) {
+    txtcol &= 15
+    bgcol &= 15
+    c64.CHROUT(color_to_charcode[bgcol])
+    c64.CHROUT(1)       ; switch fg and bg colors
+    c64.CHROUT(color_to_charcode[txtcol])
+}
+
+sub lowercase() {
+    cx16.screen_set_charset(3, 0)  ; lowercase charset
+}
+
+sub uppercase() {
+    cx16.screen_set_charset(2, 0)  ; uppercase charset
+}
+
+asmsub  scroll_left() clobbers(A, Y)  {
+	; ---- scroll the whole screen 1 character to the left
+	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
+	%asm {{
+	    phx
+	    jsr  c64.SCREEN
+	    dex
+	    stx  _lx+1
+        dey
+        sty  P8ZP_SCRATCH_B1    ; number of rows to scroll
+
+_nextline
+        stz  cx16.VERA_CTRL     ; data port 0: source column
+        lda  #%00010000         ; auto increment 1
+        sta  cx16.VERA_ADDR_H
+        lda  #2
+        sta  cx16.VERA_ADDR_L   ; begin in column 1
+        ldy  P8ZP_SCRATCH_B1
+        sty  cx16.VERA_ADDR_M
+        lda  #1
+        sta  cx16.VERA_CTRL     ; data port 1: destination column
+        lda  #%00010000         ; auto increment 1
+        sta  cx16.VERA_ADDR_H
+        stz  cx16.VERA_ADDR_L
+        sty  cx16.VERA_ADDR_M
+
+_lx     ldx  #0                ; modified
+-       lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1    ; copy char
+        lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1    ; copy color
+        dex
+        bne  -
+        dec  P8ZP_SCRATCH_B1
+        bpl  _nextline
+
+        lda  #0
+        sta  cx16.VERA_CTRL
+	    plx
+	    rts
+	}}
+}
+
+asmsub  scroll_right() clobbers(A)  {
+	; ---- scroll the whole screen 1 character to the right
+	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
+	%asm {{
+	    phx
+	    jsr  c64.SCREEN
+	    dex
+	    stx  _lx+1
+	    txa
+	    asl  a
+	    dea
+	    sta  _rcol+1
+	    ina
+	    ina
+	    sta  _rcol2+1
+        dey
+        sty  P8ZP_SCRATCH_B1    ; number of rows to scroll
+
+_nextline
+        stz  cx16.VERA_CTRL     ; data port 0: source column
+        lda  #%00011000         ; auto decrement 1
+        sta  cx16.VERA_ADDR_H
+_rcol   lda  #79*2-1             ; modified
+        sta  cx16.VERA_ADDR_L   ; begin in rightmost column minus one
+        ldy  P8ZP_SCRATCH_B1
+        sty  cx16.VERA_ADDR_M
+        lda  #1
+        sta  cx16.VERA_CTRL     ; data port 1: destination column
+        lda  #%00011000         ; auto decrement 1
+        sta  cx16.VERA_ADDR_H
+_rcol2  lda  #79*2+1            ; modified
+        sta  cx16.VERA_ADDR_L
+        sty  cx16.VERA_ADDR_M
+
+_lx     ldx  #0                 ; modified
+-       lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1    ; copy char
+        lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1    ; copy color
+        dex
+        bne  -
+        dec  P8ZP_SCRATCH_B1
+        bpl  _nextline
+
+        lda  #0
+        sta  cx16.VERA_CTRL
+	    plx
+	    rts
+	}}
+}
+
+asmsub  scroll_up() clobbers(A, Y)  {
+	; ---- scroll the whole screen 1 character up
+	;      contents of the bottom row are unchanged, you should refill/clear this yourself
+	%asm {{
+	    phx
+	    jsr  c64.SCREEN
+	    stx  _nextline+1
+	    dey
+        sty  P8ZP_SCRATCH_B1
+        stz  cx16.VERA_CTRL         ; data port 0 is source
+        lda  #1
+        sta  cx16.VERA_ADDR_M       ; start at second line
+        stz  cx16.VERA_ADDR_L
+        lda  #%00010000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 1, bank 0.
+
+        lda  #1
+        sta  cx16.VERA_CTRL         ; data port 1 is destination
+        stz  cx16.VERA_ADDR_M       ; start at top line
+        stz  cx16.VERA_ADDR_L
+        lda  #%00010000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 1, bank 0.
+
+_nextline
+        ldx  #80        ; modified
+-       lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1        ; copy char
+        lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1        ; copy color
+        dex
+        bne  -
+        dec  P8ZP_SCRATCH_B1
+        beq  +
+        stz  cx16.VERA_CTRL         ; data port 0
+        stz  cx16.VERA_ADDR_L
+        inc  cx16.VERA_ADDR_M
+        lda  #1
+        sta  cx16.VERA_CTRL         ; data port 1
+        stz  cx16.VERA_ADDR_L
+        inc  cx16.VERA_ADDR_M
+        bra  _nextline
+
++       lda  #0
+        sta  cx16.VERA_CTRL
+	    plx
+	    rts
+	}}
+}
+
+asmsub  scroll_down() clobbers(A, Y)  {
+	; ---- scroll the whole screen 1 character down
+	;      contents of the top row are unchanged, you should refill/clear this yourself
+	%asm {{
+	    phx
+	    jsr  c64.SCREEN
+	    stx  _nextline+1
+	    dey
+        sty  P8ZP_SCRATCH_B1
+        stz  cx16.VERA_CTRL         ; data port 0 is source
+        dey
+        sty  cx16.VERA_ADDR_M       ; start at line before bottom line
+        stz  cx16.VERA_ADDR_L
+        lda  #%00010000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 1, bank 0.
+
+        lda  #1
+        sta  cx16.VERA_CTRL         ; data port 1 is destination
+        iny
+        sty  cx16.VERA_ADDR_M       ; start at bottom line
+        stz  cx16.VERA_ADDR_L
+        lda  #%00010000
+        sta  cx16.VERA_ADDR_H       ; enable auto increment by 1, bank 0.
+
+_nextline
+        ldx  #80        ; modified
+-       lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1        ; copy char
+        lda  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA1        ; copy color
+        dex
+        bne  -
+        dec  P8ZP_SCRATCH_B1
+        beq  +
+        stz  cx16.VERA_CTRL         ; data port 0
+        stz  cx16.VERA_ADDR_L
+        dec  cx16.VERA_ADDR_M
+        lda  #1
+        sta  cx16.VERA_CTRL         ; data port 1
+        stz  cx16.VERA_ADDR_L
+        dec  cx16.VERA_ADDR_M
+        bra  _nextline
+
++       lda  #0
+        sta  cx16.VERA_CTRL
+	    plx
+	    rts
+	}}
+}
+
+romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use c64.CHROUT directly ofcourse.
+
+asmsub  print (str text @ AY) clobbers(A,Y)  {
+	; ---- print null terminated string from A/Y
+	; note: the compiler contains an optimization that will replace
+	;       a call to this subroutine with a string argument of just one char,
+	;       by just one call to c64.CHROUT of that single char.
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		sty  P8ZP_SCRATCH_REG
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_B1),y
+		beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
+	%asm {{
+		phx
+		jsr  conv.ubyte2decimal
+		pha
+		tya
+		jsr  c64.CHROUT
+		pla
+		jsr  c64.CHROUT
+		txa
+		jsr  c64.CHROUT
+		plx
+		rts
+	}}
+}
+
+asmsub  print_ub  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- print the ubyte in A in decimal form, without left padding 0s
+	%asm {{
+		phx
+		jsr  conv.ubyte2decimal
+_print_byte_digits
+		pha
+		cpy  #'0'
+		beq  +
+		tya
+		jsr  c64.CHROUT
+		pla
+		jsr  c64.CHROUT
+		bra  _ones
++       pla
+        cmp  #'0'
+        beq  _ones
+        jsr  c64.CHROUT
+_ones   txa
+		jsr  c64.CHROUT
+		plx
+		rts
+	}}
+}
+
+asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
+	; ---- print the byte in A in decimal form, without left padding 0s
+	%asm {{
+		phx
+		pha
+		cmp  #0
+		bpl  +
+		lda  #'-'
+		jsr  c64.CHROUT
++		pla
+		jsr  conv.byte2decimal
+		bra  print_ub._print_byte_digits
+	}}
+}
+
+asmsub  print_ubhex  (ubyte value @ A, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		phx
+		bcc  +
+		pha
+		lda  #'$'
+		jsr  c64.CHROUT
+		pla
++		jsr  conv.ubyte2hex
+		jsr  c64.CHROUT
+		tya
+		jsr  c64.CHROUT
+		plx
+		rts
+	}}
+}
+
+asmsub  print_ubbin  (ubyte value @ A, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		phx
+		sta  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'%'
+		jsr  c64.CHROUT
++		ldy  #8
+-		lda  #'0'
+		asl  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'1'
++		jsr  c64.CHROUT
+		dey
+		bne  -
+		plx
+		rts
+	}}
+}
+
+asmsub  print_uwbin  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubbin
+		pla
+		clc
+		bra  print_ubbin
+	}}
+}
+
+asmsub  print_uwhex  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in hexadecimal form (4 digits)
+	;      (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubhex
+		pla
+		clc
+		bra  print_ubhex
+	}}
+}
+
+asmsub  print_uw0  (uword value @ AY) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
+	%asm {{
+	    phx
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+        beq  +
+		jsr  c64.CHROUT
+		iny
+		bne  -
++		plx
+		rts
+	}}
+}
+
+asmsub  print_uw  (uword value @ AY) clobbers(A,Y)  {
+	; ---- print the uword in A/Y in decimal form, without left padding 0s
+	%asm {{
+	    phx
+		jsr  conv.uword2decimal
+		plx
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+		beq  _allzero
+		cmp  #'0'
+		bne  _gotdigit
+		iny
+		bne  -
+
+_gotdigit
+		jsr  c64.CHROUT
+		iny
+		lda  conv.uword2decimal.decTenThousands,y
+		bne  _gotdigit
+		rts
+_allzero
+        lda  #'0'
+        jmp  c64.CHROUT
+	}}
+}
+
+asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
+	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
+	%asm {{
+		cpy  #0
+		bpl  +
+		pha
+		lda  #'-'
+		jsr  c64.CHROUT
+		tya
+		eor  #255
+		tay
+		pla
+		eor  #255
+		clc
+		adc  #1
+		bcc  +
+		iny
++		bra  print_uw
+	}}
+}
+
+asmsub  input_chars  (uword buffer @ AY) clobbers(A) -> ubyte @ Y  {
+	; ---- Input a string (max. 80 chars) from the keyboard. Returns length in Y. (string is terminated with a 0 byte as well)
+	;      It assumes the keyboard is selected as I/O channel!
+
+	%asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0				; char counter = 0
+-		jsr  c64.CHRIN
+		cmp  #$0d			; return (ascii 13) pressed?
+		beq  +				; yes, end.
+		sta  (P8ZP_SCRATCH_W1),y	; else store char in buffer
+		iny
+		bne  -
++		lda  #0
+		sta  (P8ZP_SCRATCH_W1),y	; finish string with 0 byte
+		rts
+
+	}}
+}
+
+asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A)  {
+	; ---- sets the character in the screen matrix at the given position
+	%asm {{
+            pha
+            txa
+            asl  a
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            pla
+            sta  cx16.VERA_DATA0
+            rts
+	}}
+}
+
+asmsub  getchr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
+	; ---- get the character in the screen matrix at the given location
+	%asm  {{
+            asl  a
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            rts
+	}}
+}
+
+asmsub  setclr  (ubyte col @X, ubyte row @Y, ubyte color @A) clobbers(A)  {
+	; ---- set the color in A on the screen matrix at the given position
+	;      note: on the CommanderX16 this allows you to set both Fg and Bg colors;
+	;            use the high nybble in A to set the Bg color!
+	%asm {{
+            pha
+            txa
+            asl  a
+            ina
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            pla
+            sta  cx16.VERA_DATA0
+            rts
+	}}
+}
+
+asmsub  getclr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
+	; ---- get the color in the screen color matrix at the given location
+	%asm  {{
+            asl  a
+            ina
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            rts
+	}}
+}
+
+sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+	; ---- set char+color at the given position on the screen
+	;      note: color handling is the same as on the C64: it only sets the foreground color.
+	;            use setcc2 if you want Cx-16 specific feature of setting both Bg+Fg colors.
+	%asm {{
+            phx
+            lda  column
+            asl  a
+            tax
+            ldy  row
+            lda  charcolor
+            and  #$0f
+            sta  P8ZP_SCRATCH_B1
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  char
+            sta  cx16.VERA_DATA0
+            inx
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            and  #$f0
+            ora  P8ZP_SCRATCH_B1
+            sta  cx16.VERA_DATA0
+            plx
+            rts
+    }}
+}
+
+sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
+	; ---- set char+color at the given position on the screen
+	;      note: on the CommanderX16 this allows you to set both Fg and Bg colors;
+	;            use the high nybble in A to set the Bg color!
+	%asm {{
+            phx
+            lda  column
+            asl  a
+            tax
+            ldy  row
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  char
+            sta  cx16.VERA_DATA0
+            inx
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  colors
+            sta  cx16.VERA_DATA0
+            plx
+            rts
+    }}
+}
+
+asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
+	%asm  {{
+		phx
+		tax
+		clc
+		jsr  c64.PLOT
+		plx
+		rts
+	}}
+}
+
+asmsub width() clobbers(X,Y) -> ubyte @A {
+    ; -- returns the text screen width (number of columns)
+    %asm {{
+        jsr  c64.SCREEN
+        txa
+        rts
+    }}
+}
+
+asmsub height() clobbers(X, Y) -> ubyte @A {
+    ; -- returns the text screen height (number of rows)
+    %asm {{
+        jsr  c64.SCREEN
+        tya
+        rts
+    }}
+}
+
+}

compiler/res/prog8lib/cx16logo.p8

@@ -0,0 +1,31 @@
+; routine to draw the Commander X16's log in petscii.
+
+%import textio
+
+cx16logo {
+    sub logo_at(ubyte column, ubyte row) {
+        uword strptr
+        for strptr in logo_lines {
+            txt.plot(column, row)
+            txt.print(strptr)
+            row++
+        }
+    }
+
+    sub logo() {
+        uword strptr
+        for strptr in logo_lines
+            txt.print(strptr)
+        txt.nl()
+    }
+
+    str[] logo_lines = [
+            "\uf10d\uf11a\uf139\uf11b     \uf11a\uf13a\uf11b\n",
+            "\uf10b\uf11a▎\uf139\uf11b   \uf11a\uf13a\uf130\uf11b\n",
+            "\uf10f\uf11a▌ \uf139\uf11b \uf11a\uf13a \uf11b▌\n",
+            "\uf102 \uf132\uf11a▖\uf11b \uf11a▗\uf11b\uf132\n",
+            "\uf10e ▂\uf11a▘\uf11b \uf11a▝\uf11b▂\n",
+            "\uf104 \uf11a \uf11b\uf13a\uf11b \uf139\uf11a \uf11b\n",
+            "\uf101\uf130\uf13a   \uf139▎\uf100"
+        ]
+}

compiler/res/prog8lib/diskio.p8

@@ -0,0 +1,450 @@
+; C64 and Cx16 disk drive I/O routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+%import textio
+%import string
+%import syslib
+
+diskio {
+
+    sub directory(ubyte drivenumber) -> ubyte {
+        ; -- Prints the directory contents of disk drive 8-11 to the screen. Returns success.
+
+        c64.SETNAM(1, "$")
+        c64.SETLFS(13, drivenumber, 0)
+        void c64.OPEN()          ; open 13,8,0,"$"
+        if_cs
+            goto io_error
+        void c64.CHKIN(13)        ; use #13 as input channel
+        if_cs
+            goto io_error
+
+        repeat 4 {
+            void c64.CHRIN()     ; skip the 4 prologue bytes
+        }
+
+        ; while not key pressed / EOF encountered, read data.
+        ubyte status = c64.READST()
+        while not status {
+            ubyte low = c64.CHRIN()
+            ubyte high = c64.CHRIN()
+            txt.print_uw(mkword(high, low))
+            txt.spc()
+            ubyte @zp char
+            repeat {
+                char = c64.CHRIN()
+                if char==0
+                    break
+                txt.chrout(char)
+            }
+            txt.nl()
+            void c64.CHRIN()     ; skip 2 bytes
+            void c64.CHRIN()
+            status = c64.READST()
+            if c64.STOP2()
+                break
+        }
+
+io_error:
+        status = c64.READST()
+        c64.CLRCHN()        ; restore default i/o devices
+        c64.CLOSE(13)
+
+        if status and status & $40 == 0 {            ; bit 6=end of file
+            txt.print("\ni/o error, status: ")
+            txt.print_ub(status)
+            txt.nl()
+            return false
+        }
+
+        return true
+    }
+
+
+    ; internal variables for the iterative file lister / loader
+    ubyte list_skip_disk_name
+    uword list_pattern
+    uword list_blocks
+    ubyte iteration_in_progress = false
+    ubyte @zp first_byte
+    ubyte have_first_byte
+    str   list_filename = "?" * 32
+
+
+    ; ----- get a list of files (uses iteration functions internally) -----
+
+    sub list_files(ubyte drivenumber, uword pattern_ptr, uword name_ptrs, ubyte max_names) -> ubyte {
+        ; -- fill the array 'name_ptrs' with (pointers to) the names of the files requested.
+        uword names_buffer = memory("filenames", 512)
+        uword buffer_start = names_buffer
+        ubyte files_found = 0
+        if lf_start_list(drivenumber, pattern_ptr) {
+            while lf_next_entry() {
+                @(name_ptrs) = lsb(names_buffer)
+                name_ptrs++
+                @(name_ptrs) = msb(names_buffer)
+                name_ptrs++
+                names_buffer += string.copy(diskio.list_filename, names_buffer) + 1
+                files_found++
+                if names_buffer - buffer_start > 512-18
+                    break
+                if files_found == max_names
+                    break
+            }
+            lf_end_list()
+        }
+        return files_found
+    }
+
+    ; ----- iterative file lister functions (uses io channel 12) -----
+
+    sub lf_start_list(ubyte drivenumber, uword pattern_ptr) -> ubyte {
+        ; -- start an iterative file listing with optional pattern matching.
+        ;    note: only a single iteration loop can be active at a time!
+        lf_end_list()
+        list_pattern = pattern_ptr
+        list_skip_disk_name = true
+        iteration_in_progress = true
+
+        c64.SETNAM(1, "$")
+        c64.SETLFS(12, drivenumber, 0)
+        void c64.OPEN()          ; open 12,8,0,"$"
+        if_cs
+            goto io_error
+        void c64.CHKIN(12)        ; use #12 as input channel
+        if_cs
+            goto io_error
+
+        repeat 4 {
+            void c64.CHRIN()     ; skip the 4 prologue bytes
+        }
+
+        if c64.READST()==0
+            return true
+
+io_error:
+        lf_end_list()
+        return false
+    }
+
+    sub lf_next_entry() -> ubyte {
+        ; -- retrieve the next entry from an iterative file listing session.
+        ;    results will be found in list_blocks and list_filename.
+        ;    if it returns false though, there are no more entries (or an error occurred).
+
+        if not iteration_in_progress
+            return false
+
+        repeat {
+            void c64.CHKIN(12)        ; use #12 as input channel again
+
+            uword nameptr = &list_filename
+            ubyte blocks_lsb = c64.CHRIN()
+            ubyte blocks_msb = c64.CHRIN()
+
+            if c64.READST()
+                goto close_end
+
+            list_blocks = mkword(blocks_msb, blocks_lsb)
+
+            ; read until the filename starts after the first "
+            while c64.CHRIN()!='\"'  {
+                if c64.READST()
+                    goto close_end
+            }
+
+            ; read the filename
+            repeat {
+                ubyte char = c64.CHRIN()
+                if char==0
+                    break
+                if char=='\"'
+                    break
+                @(nameptr) = char
+                nameptr++
+            }
+
+            @(nameptr) = 0
+
+            while c64.CHRIN() {
+                ; read the rest of the entry until the end
+            }
+
+            void c64.CHRIN()     ; skip 2 bytes
+            void c64.CHRIN()
+
+            if not list_skip_disk_name {
+                if not list_pattern
+                    return true
+                if prog8_lib.pattern_match(list_filename, list_pattern)
+                    return true
+            }
+            list_skip_disk_name = false
+        }
+
+close_end:
+        lf_end_list()
+        return false
+    }
+
+    sub lf_end_list() {
+        ; -- end an iterative file listing session (close channels).
+        if iteration_in_progress {
+            c64.CLRCHN()
+            c64.CLOSE(12)
+            iteration_in_progress = false
+        }
+    }
+
+
+    ; ----- iterative file loader functions (uses io channel 11) -----
+
+    sub f_open(ubyte drivenumber, uword filenameptr) -> ubyte {
+        ; -- open a file for iterative reading with f_read
+        ;    note: only a single iteration loop can be active at a time!
+        f_close()
+
+        c64.SETNAM(string.length(filenameptr), filenameptr)
+        c64.SETLFS(11, drivenumber, 0)
+        void c64.OPEN()          ; open 11,8,0,"filename"
+        if_cc {
+            iteration_in_progress = true
+            have_first_byte = false
+            void c64.CHKIN(11)        ; use #11 as input channel
+            if_cc {
+                first_byte = c64.CHRIN()   ; read first byte to test for file not found
+                if not c64.READST() {
+                    have_first_byte = true
+                    return true
+                }
+            }
+        }
+        f_close()
+        return false
+    }
+
+    sub f_read(uword bufferpointer, uword num_bytes) -> uword {
+        ; -- read from the currently open file, up to the given number of bytes.
+        ;    returns the actual number of bytes read.  (checks for End-of-file and error conditions)
+        if not iteration_in_progress or not num_bytes
+            return 0
+
+        list_blocks = 0     ; we reuse this variable for the total number of bytes read
+        if have_first_byte {
+            have_first_byte=false
+            @(bufferpointer) = first_byte
+            bufferpointer++
+            list_blocks++
+            num_bytes--
+        }
+
+        void c64.CHKIN(11)        ; use #11 as input channel again
+        %asm {{
+            lda  bufferpointer
+            sta  _in_buffer+1
+            lda  bufferpointer+1
+            sta  _in_buffer+2
+        }}
+        repeat num_bytes {
+            %asm {{
+                jsr  c64.CHRIN
+                sta  cx16.r5
+_in_buffer      sta  $ffff
+                inc  _in_buffer+1
+                bne  +
+                inc  _in_buffer+2
++               inc  list_blocks
+                bne  +
+                inc  list_blocks+1
++
+            }}
+
+            if cx16.r5==$0d {   ; chance on I/o error status?
+                first_byte = c64.READST()
+                if first_byte & $40
+                    f_close()       ; end of file, close it
+                if first_byte
+                    return list_blocks
+            }
+        }
+        return list_blocks
+    }
+
+    sub f_read_all(uword bufferpointer) -> uword {
+        ; -- read the full contents of the file, returns number of bytes read.
+        if not iteration_in_progress
+            return 0
+
+        list_blocks = 0     ; we reuse this variable for the total number of bytes read
+        if have_first_byte {
+            have_first_byte=false
+            @(bufferpointer) = first_byte
+            bufferpointer++
+            list_blocks++
+        }
+
+        while not c64.READST() {
+            list_blocks += f_read(bufferpointer, 256)
+            bufferpointer += 256
+        }
+        return list_blocks
+    }
+
+    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y {
+        ; Routine to read text lines from a text file. Lines must be less than 255 characters.
+        ; Reads characters from the input file UNTIL a newline or return character (or EOF).
+        ; The line read will be 0-terminated in the buffer (and not contain the end of line character).
+        ; The length of the line is returned in Y. Note that an empty line is okay and is length 0!
+        ; I/O error status should be checked by the caller itself via READST() routine.
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  #11
+            jsr  c64.CHKIN              ; use channel 11 again for input
+            ldy  #0
+            lda  have_first_byte
+            beq  _loop
+            lda  #0
+            sta  have_first_byte
+            lda  first_byte
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+_loop       jsr  c64.CHRIN
+            sta  (P8ZP_SCRATCH_W1),y
+            beq  _end
+            iny
+            cmp  #$0a
+            beq  _line_end
+            cmp  #$0d
+            bne  _loop
+_line_end   dey     ; get rid of the trailing end-of-line char
+            lda  #0
+            sta  (P8ZP_SCRATCH_W1),y
+_end        rts
+        }}
+    }
+
+
+    sub f_close() {
+        ; -- end an iterative file loading session (close channels).
+        if iteration_in_progress {
+            c64.CLRCHN()
+            c64.CLOSE(11)
+            iteration_in_progress = false
+        }
+    }
+
+
+    sub status(ubyte drivenumber) -> uword {
+        ; -- retrieve the disk drive's current status message
+        uword messageptr = &filename
+        c64.SETNAM(0, filename)
+        c64.SETLFS(15, drivenumber, 15)
+        void c64.OPEN()          ; open 15,8,15
+        if_cs
+            goto io_error
+        void c64.CHKIN(15)        ; use #15 as input channel
+        if_cs
+            goto io_error
+
+        while not c64.READST() {
+            @(messageptr) = c64.CHRIN()
+            messageptr++
+        }
+
+io_error:
+        @(messageptr) = 0
+        c64.CLRCHN()        ; restore default i/o devices
+        c64.CLOSE(15)
+        return filename
+    }
+
+
+    sub save(ubyte drivenumber, uword filenameptr, uword address, uword size) -> ubyte {
+        c64.SETNAM(string.length(filenameptr), filenameptr)
+        c64.SETLFS(1, drivenumber, 0)
+        uword end_address = address + size
+
+        %asm {{
+            lda  address
+            sta  P8ZP_SCRATCH_W1
+            lda  address+1
+            sta  P8ZP_SCRATCH_W1+1
+            stx  P8ZP_SCRATCH_REG
+            lda  #<P8ZP_SCRATCH_W1
+            ldx  end_address
+            ldy  end_address+1
+            jsr  c64.SAVE
+            php
+            ldx  P8ZP_SCRATCH_REG
+            plp
+        }}
+
+        first_byte = 0      ; result var reuse
+        if_cc
+            first_byte = c64.READST()==0
+
+        c64.CLRCHN()
+        c64.CLOSE(1)
+
+        return first_byte
+    }
+
+    sub load(ubyte drivenumber, uword filenameptr, uword address_override) -> uword {
+        c64.SETNAM(string.length(filenameptr), filenameptr)
+        ubyte secondary = 1
+        uword end_of_load = 0
+        if address_override
+            secondary = 0
+        c64.SETLFS(1, drivenumber, secondary)
+        %asm {{
+            stx  P8ZP_SCRATCH_REG
+            lda  #0
+            ldx  address_override
+            ldy  address_override+1
+            jsr  c64.LOAD
+            bcs  +
+            stx  end_of_load
+            sty  end_of_load+1
++           ldx  P8ZP_SCRATCH_REG
+        }}
+
+        c64.CLRCHN()
+        c64.CLOSE(1)
+
+        if end_of_load
+            return end_of_load - address_override
+
+        return 0
+    }
+
+
+    str filename = "0:??????????????????????????????????????"
+
+    sub delete(ubyte drivenumber, uword filenameptr) {
+        ; -- delete a file on the drive
+        filename[0] = 's'
+        filename[1] = ':'
+        ubyte flen = string.copy(filenameptr, &filename+2)
+        c64.SETNAM(flen+2, filename)
+        c64.SETLFS(1, drivenumber, 15)
+        void c64.OPEN()
+        c64.CLRCHN()
+        c64.CLOSE(1)
+    }
+
+    sub rename(ubyte drivenumber, uword oldfileptr, uword newfileptr) {
+        ; -- rename a file on the drive
+        filename[0] = 'r'
+        filename[1] = ':'
+        ubyte flen_new = string.copy(newfileptr, &filename+2)
+        filename[flen_new+2] = '='
+        ubyte flen_old = string.copy(oldfileptr, &filename+3+flen_new)
+        c64.SETNAM(3+flen_new+flen_old, filename)
+        c64.SETLFS(1, drivenumber, 15)
+        void c64.OPEN()
+        c64.CLRCHN()
+        c64.CLOSE(1)
+    }
+}

compiler/res/prog8lib/math.p8

@@ -1,11 +1,7 @@
-; Prog8 internal Math library routines - always included by the compiler
+; Internal Math library routines - always included by the compiler
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
 
-%import c64lib
-
-~ math {
+math {
 	%asminclude "library:math.asm", ""
 }

compiler/res/prog8lib/prog8_lib.p8

@@ -0,0 +1,84 @@
+; Internal library routines - always included by the compiler
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+prog8_lib {
+	%asminclude "library:prog8_lib.asm", ""
+	%asminclude "library:prog8_funcs.asm", ""
+
+	uword @zp retval_interm_uw      ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	word @zp retval_interm_w        ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	ubyte @zp retval_interm_ub      ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	byte @zp retval_interm_b        ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+
+	asmsub pattern_match(str string @AY, str pattern @R0) clobbers(Y) -> ubyte @A {
+		%asm {{
+; pattern matching of a string.
+; Input:  cx16.r0:  A NUL-terminated, <255-length pattern
+;              AY:  A NUL-terminated, <255-length string
+;
+; Output: A = 1 if the string matches the pattern, A = 0 if not.
+;
+; Notes:  Clobbers A, X, Y. Each * in the pattern uses 4 bytes of stack.
+;
+; see http://6502.org/source/strings/patmatch.htm
+
+str = P8ZP_SCRATCH_W1
+
+	stx  P8ZP_SCRATCH_REG
+	sta  str
+	sty  str+1
+	lda  cx16.r0
+	sta  modify_pattern1+1
+	sta  modify_pattern2+1
+	lda  cx16.r0+1
+	sta  modify_pattern1+2
+	sta  modify_pattern2+2
+	jsr  _match
+	lda  #0
+	adc  #0
+	ldx  P8ZP_SCRATCH_REG
+	rts
+
+
+_match
+	ldx #$00        ; x is an index in the pattern
+	ldy #$ff        ; y is an index in the string
+modify_pattern1
+next    lda $ffff,x   ; look at next pattern character    MODIFIED
+	cmp #'*'     ; is it a star?
+	beq star        ; yes, do the complicated stuff
+	iny             ; no, let's look at the string
+	cmp #'?'     ; is the pattern caracter a ques?
+	bne reg         ; no, it's a regular character
+	lda (str),y     ; yes, so it will match anything
+	beq fail        ;  except the end of string
+reg     cmp (str),y     ; are both characters the same?
+	bne fail        ; no, so no match
+	inx             ; yes, keep checking
+	cmp #0          ; are we at end of string?
+	bne next        ; not yet, loop
+found   rts             ; success, return with c=1
+
+star    inx             ; skip star in pattern
+modify_pattern2
+	cmp $ffff,x   	; string of stars equals one star	MODIFIED
+	beq star        ;  so skip them also
+stloop  txa             ; we first try to match with * = ""
+	pha             ;  and grow it by 1 character every
+	tya             ;  time we loop
+	pha             ; save x and y on stack
+	jsr next        ; recursive call
+	pla             ; restore x and y
+	tay
+	pla
+	tax
+	bcs found       ; we found a match, return with c=1
+	iny             ; no match yet, try to grow * string
+	lda (str),y     ; are we at the end of string?
+	bne stloop      ; not yet, add a character
+fail    clc             ; yes, no match found, return with c=0
+	rts
+		}}
+	}
+}

compiler/res/prog8lib/prog8lib.p8

@@ -1,11 +0,0 @@
-; Prog8 internal library routines - always included by the compiler
-;
-; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
-
-%import c64lib
-
-~ prog8_lib {
-	%asminclude "library:prog8lib.asm", ""
-}

compiler/res/prog8lib/string.p8

@@ -0,0 +1,235 @@
+; 0-terminated string manipulation routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+
+string {
+
+    asmsub length(uword string @AY) clobbers(A) -> ubyte @Y {
+        ; Returns the number of bytes in the string.
+        ; This value is determined during runtime and counts upto the first terminating 0 byte in the string,
+        ; regardless of the size of the string during compilation time. Don’t confuse this with len and sizeof!
+
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  +
+		iny
+		bne  -
++		rts
+        }}
+    }
+
+    asmsub left(uword source @R0, ubyte length @A, uword target @R1) clobbers(A, Y) {
+        ; Copies the left side of the source string of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that length is smaller or equal to the length of the source string.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+		ldy  cx16.r0
+		sty  P8ZP_SCRATCH_W1
+		ldy  cx16.r0+1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  cx16.r1
+		sty  P8ZP_SCRATCH_W2
+		ldy  cx16.r1+1
+		sty  P8ZP_SCRATCH_W2+1
+		tay
+		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
+		rts
+_loop		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
++		rts
+        }}
+;                asmgen.out("  jsr  prog8_lib.func_leftstr")
+    }
+
+    asmsub right(uword source @R0, ubyte length @A, uword target @R1) clobbers(A,Y) {
+        ; Copies the right side of the source string of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that length is smaller or equal to the length of the source string.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+                sta  P8ZP_SCRATCH_B1
+                lda  cx16.r0
+                ldy  cx16.r0+1
+                jsr  string.length
+                tya
+                sec
+                sbc  P8ZP_SCRATCH_B1
+                clc
+                adc  cx16.r0
+		sta  P8ZP_SCRATCH_W1
+		lda  cx16.r0+1
+		adc  #0
+		sta  P8ZP_SCRATCH_W1+1
+		ldy  cx16.r1
+		sty  P8ZP_SCRATCH_W2
+		ldy  cx16.r1+1
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  P8ZP_SCRATCH_B1
+		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
+		rts
+_loop		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
++		rts
+        }}
+    }
+
+    asmsub slice(uword source @R0, ubyte start @A, ubyte length @Y, uword target @R1) clobbers(A, Y) {
+        ; Copies a segment from the source string, starting at the given index,
+        ;  and of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that start and length are within bounds of the strings.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+		; substr(source, target, start, length)
+		sta  P8ZP_SCRATCH_B1
+		lda  cx16.r0
+		sta  P8ZP_SCRATCH_W1
+		lda  cx16.r0+1
+		sta  P8ZP_SCRATCH_W1+1
+		lda  cx16.r1
+		sta  P8ZP_SCRATCH_W2
+		lda  cx16.r1+1
+		sta  P8ZP_SCRATCH_W2+1
+
+		; adjust src location
+		clc
+		lda  P8ZP_SCRATCH_W1
+		adc  P8ZP_SCRATCH_B1
+		sta  P8ZP_SCRATCH_W1
+		bcc  +
+		inc  P8ZP_SCRATCH_W1+1
++		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		beq  _startloop
+-		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+_startloop	dey
+		cpy  #$ff
+		bne  -
+		rts
+        }}
+    }
+
+    asmsub find(uword string @R0, ubyte character @A) -> uword @AY {
+        ; Locates the first position of the given character in the string,
+        ;  returns the string starting with this character or $0000 if the character is not found.
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+                sta  P8ZP_SCRATCH_B1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  _notfound
+		cmp  P8ZP_SCRATCH_B1
+		beq  _found
+		iny
+		bne  -
+_notfound	lda  #0
+		ldy  #0
+		rts
+_found		sty  P8ZP_SCRATCH_B1
+		ldy  P8ZP_SCRATCH_W1+1
+		lda  P8ZP_SCRATCH_W1
+		clc
+		adc  P8ZP_SCRATCH_B1
+		bcc  +
+		iny
++		rts
+
+        }}
+    }
+
+    asmsub copy(uword source @R0, uword target @AY) clobbers(A) -> ubyte @Y {
+        ; Copy a string to another, overwriting that one.
+        ; Returns the length of the string that was copied.
+        ; Often you don’t have to call this explicitly and can just write string1 = string2
+        ; but this function is useful if you’re dealing with addresses for instance.
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcpy
+        }}
+    }
+
+    asmsub compare(uword string1 @R0, uword string2 @AY) clobbers(Y) -> byte @A {
+        ; Compares two strings for sorting.
+        ; Returns -1 (255), 0 or 1 depeding on wether string1 sorts before, equal or after string2.
+        ; Note that you can also directly compare strings and string values with eachother using
+        ; comparison operators ==, < etcetera (it will use strcmp for you under water automatically).
+        %asm {{
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcmp_mem
+        }}
+    }
+
+    asmsub lower(uword st @AY) {
+        ; Lowercases the petscii string in-place.
+        ; (for efficiency, non-letter characters > 128 will also not be left intact,
+        ;  but regular text doesn't usually contain those characters anyway.)
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            beq  _done
+            and  #$7f
+            cmp  #97
+            bcc  +
+            cmp  #123
+            bcs  +
+            and  #%11011111
++           sta  (P8ZP_SCRATCH_W1),y
+            iny
+            bne  -
+_done       rts
+        }}
+    }
+
+    asmsub upper(uword st @AY) {
+        ; Uppercases the petscii string in-place.
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            beq  _done
+            cmp  #65
+            bcc  +
+            cmp  #91
+            bcs  +
+            ora  #%00100000
++           sta  (P8ZP_SCRATCH_W1),y
+            iny
+            bne  -
+_done       rts
+        }}
+    }
+}

compiler/res/prog8lib/test_stack.p8

@@ -0,0 +1,50 @@
+; utility debug code to print the X (evalstack) and SP (cpu stack) registers.
+
+%import textio
+
+test_stack {
+
+    asmsub test() {
+        %asm {{
+	stx  _saveX
+	lda  #13
+	jsr  txt.chrout
+	lda  #'-'
+	ldy  #12
+-	jsr  txt.chrout
+	dey
+	bne  -
+	lda  #13
+	jsr  txt.chrout
+	lda  #'x'
+	jsr  txt.chrout
+	lda  #'='
+	jsr  txt.chrout
+	lda  _saveX
+	jsr  txt.print_ub
+	lda  #' '
+	jsr  txt.chrout
+	lda  #'s'
+	jsr  txt.chrout
+	lda  #'p'
+	jsr  txt.chrout
+	lda  #'='
+	jsr  txt.chrout
+	tsx
+	txa
+	jsr  txt.print_ub
+	lda  #13
+	jsr  txt.chrout
+	lda  #'-'
+	ldy  #12
+-	jsr  txt.chrout
+	dey
+	bne  -
+	lda  #13
+	jsr  txt.chrout
+	ldx  _saveX
+	rts
+_saveX	.byte 0
+        }}
+    }
+}

compiler/res/version.txt

@@ -1 +1 @@
-1.2 (beta)
+6.4

compiler/src/prog8/CompilerMain.kt

@@ -1,216 +1,124 @@
 package prog8
 
-import prog8.ast.*
-import prog8.compiler.*
-import prog8.compiler.target.c64.AsmGen
-import prog8.compiler.target.c64.C64Zeropage
-import prog8.optimizing.constantFold
-import prog8.optimizing.optimizeStatements
-import prog8.optimizing.simplifyExpressions
+import kotlinx.cli.ArgParser
+import kotlinx.cli.ArgType
+import kotlinx.cli.default
+import kotlinx.cli.multiple
+import prog8.ast.base.AstException
+import prog8.compiler.CompilationResult
+import prog8.compiler.compileProgram
+import prog8.compiler.target.C64Target
+import prog8.compiler.target.Cx16Target
 import prog8.parser.ParsingFailedError
-import prog8.parser.importModule
-import java.io.File
-import java.io.PrintStream
-import java.lang.Exception
-import java.nio.file.Paths
+import java.nio.file.FileSystems
+import java.nio.file.Path
+import java.nio.file.StandardWatchEventKinds
+import java.time.LocalDateTime
 import kotlin.system.exitProcess
-import kotlin.system.measureTimeMillis
 
 
 fun main(args: Array<String>) {
-
-    // check if the user wants to launch the VM instead
-    if("-vm" in args) {
-        val newArgs = args.toMutableList()
-        newArgs.remove("-vm")
-        return stackVmMain(newArgs.toTypedArray())
-    }
-
     printSoftwareHeader("compiler")
 
-    if (args.isEmpty())
-        usage()
     compileMain(args)
 }
 
-fun printSoftwareHeader(what: String) {
+internal fun printSoftwareHeader(what: String) {
     val buildVersion = object {}.javaClass.getResource("/version.txt").readText().trim()
-    println("\nProg8 $what by Irmen de Jong (irmen@razorvine.net)")
-    println("Version: $buildVersion")
+    println("\nProg8 $what v$buildVersion by Irmen de Jong (irmen@razorvine.net)")
     println("This software is licensed under the GNU GPL 3.0, see https://www.gnu.org/licenses/gpl.html\n")
 }
 
 
-private fun compileMain(args: Array<String>) {
-    var emulatorToStart = ""
-    var moduleFile = ""
-    var writeVmCode = false
-    var writeAssembly = true
-    for (arg in args) {
-        if(arg=="-emu")
-            emulatorToStart = "x64"
-        else if(arg=="-emu2")
-            emulatorToStart = "x64sc"
-        else if(arg=="-writevm")
-            writeVmCode = true
-        else if(arg=="-noasm")
-            writeAssembly = false
-        else if(!arg.startsWith("-"))
-            moduleFile = arg
-        else
-            usage()
-    }
-    if(moduleFile.isBlank())
-        usage()
+fun pathFrom(stringPath: String, vararg rest: String): Path  = FileSystems.getDefault().getPath(stringPath, *rest)
 
-    val filepath = Paths.get(moduleFile).normalize()
-    var programname = "?"
+
+private fun compileMain(args: Array<String>) {
+    val cli = ArgParser("prog8compiler", prefixStyle = ArgParser.OptionPrefixStyle.JVM)
+    val startEmulator by cli.option(ArgType.Boolean, fullName = "emu", description = "auto-start emulator after successful compilation")
+    val outputDir by cli.option(ArgType.String, fullName = "out", description = "directory for output files instead of current directory").default(".")
+    val dontWriteAssembly by cli.option(ArgType.Boolean, fullName = "noasm", description="don't create assembly code")
+    val dontOptimize by cli.option(ArgType.Boolean, fullName = "noopt", description = "don't perform any optimizations")
+    val watchMode by cli.option(ArgType.Boolean, fullName = "watch", description = "continuous compilation mode (watches for file changes), greatly increases compilation speed")
+    val slowCodegenWarnings by cli.option(ArgType.Boolean, fullName = "slowwarn", description="show debug warnings about slow/problematic assembly code generation")
+    val compilationTarget by cli.option(ArgType.String, fullName = "target", description = "target output of the compiler, currently '${C64Target.name}' and '${Cx16Target.name}' available").default(C64Target.name)
+    val moduleFiles by cli.argument(ArgType.String, fullName = "modules", description = "main module file(s) to compile").multiple(999)
 
     try {
-        val totalTime = measureTimeMillis {
-            // import main module and process additional imports
-            println("Parsing...")
-            val moduleAst = importModule(filepath)
-            moduleAst.linkParents()
-            var namespace = moduleAst.definingScope()
+        cli.parse(args)
+    } catch (e: IllegalStateException) {
+        System.err.println(e.message)
+        exitProcess(1)
+    }
 
-            // determine special compiler options
+    val outputPath = pathFrom(outputDir)
+    if(!outputPath.toFile().isDirectory) {
+        System.err.println("Output path doesn't exist")
+        exitProcess(1)
+    }
 
-            val compilerOptions = determineCompilationOptions(moduleAst)
+    if(watchMode==true) {
+        val watchservice = FileSystems.getDefault().newWatchService()
+        val allImportedFiles = mutableSetOf<Path>()
 
-            if (compilerOptions.launcher == LauncherType.BASIC && compilerOptions.output != OutputType.PRG)
-                throw ParsingFailedError("${moduleAst.position} BASIC launcher requires output type PRG.")
-
-            // perform initial syntax checks and constant folding
-            println("Syntax check...")
-            val heap = HeapValues()
-            val time1= measureTimeMillis {
-                moduleAst.checkIdentifiers(heap)
-            }
-            //println(" time1: $time1")
-            val time2 = measureTimeMillis {
-                moduleAst.constantFold(namespace, heap)
-            }
-            //println(" time2: $time2")
-            val time3 = measureTimeMillis {
-                moduleAst.reorderStatements(namespace,heap)     // reorder statements to please the compiler later
-            }
-            //println(" time3: $time3")
-            val time4 = measureTimeMillis {
-                moduleAst.checkValid(namespace, compilerOptions, heap)          // check if tree is valid
-            }
-            //println(" time4: $time4")
-
-            // optimize the parse tree
-            println("Optimizing...")
-            val allScopedSymbolDefinitions = moduleAst.checkIdentifiers(heap)       // useful for checking symbol usage later?
-            while (true) {
-                // keep optimizing expressions and statements until no more steps remain
-                val optsDone1 = moduleAst.simplifyExpressions(namespace, heap)
-                val optsDone2 = moduleAst.optimizeStatements(namespace, heap)
-                if (optsDone1 + optsDone2 == 0)
-                    break
+        while(true) {
+            println("Continuous watch mode active. Modules: $moduleFiles")
+            val results = mutableListOf<CompilationResult>()
+            for(filepathRaw in moduleFiles) {
+                val filepath = pathFrom(filepathRaw).normalize()
+                val compilationResult = compileProgram(filepath, dontOptimize!=true, dontWriteAssembly!=true, slowCodegenWarnings==true, compilationTarget, outputPath)
+                results.add(compilationResult)
             }
 
-            namespace = moduleAst.definingScope()       // create it again, it could have changed in the meantime
-            moduleAst.checkValid(namespace, compilerOptions, heap)          // check if final tree is valid
-            moduleAst.checkRecursion(namespace)         // check if there are recursive subroutine calls
+            val allNewlyImportedFiles = results.flatMap { it.importedFiles }
+            allImportedFiles.addAll(allNewlyImportedFiles)
 
-            // namespace.debugPrint()
+            println("Imported files (now watching:)")
+            for (importedFile in allImportedFiles) {
+                print("  ")
+                println(importedFile)
+                val watchDir = importedFile.parent ?: Path.of(".")
+                watchDir.register(watchservice, StandardWatchEventKinds.ENTRY_MODIFY)
+            }
+            println("[${LocalDateTime.now().withNano(0)}]  Waiting for file changes.")
 
-            // compile the syntax tree into stackvmProg form, and optimize that
-            val compiler = Compiler(moduleAst, namespace, heap)
-            val intermediate = compiler.compile(compilerOptions)
-            intermediate.optimize()
-
-            if(writeVmCode) {
-                val stackVmFilename = intermediate.name + ".vm.txt"
-                val stackvmFile = PrintStream(File(stackVmFilename), "utf-8")
-                intermediate.writeCode(stackvmFile)
-                stackvmFile.close()
-                println("StackVM program code written to '$stackVmFilename'")
+            var recompile=false
+            while(!recompile) {
+                val event = watchservice.take()
+                for (changed in event.pollEvents()) {
+                    val changedPath = changed.context() as Path
+                    if(allImportedFiles.any { it.fileName == changedPath.fileName }) {
+                        println("  change detected: $changedPath")
+                        recompile = true
+                    }
+                }
+                event.reset()
             }
 
-            if(writeAssembly) {
-                val zeropage = C64Zeropage(compilerOptions)
-                intermediate.allocateZeropage(zeropage)
-                val assembly = AsmGen(compilerOptions, intermediate, heap, zeropage).compileToAssembly()
-                assembly.assemble(compilerOptions)
-                programname = assembly.name
+            println("\u001b[H\u001b[2J")      // clear the screen
+        }
+
+    } else {
+        for(filepathRaw in moduleFiles) {
+            val filepath = pathFrom(filepathRaw).normalize()
+            val compilationResult: CompilationResult
+            try {
+                compilationResult = compileProgram(filepath, dontOptimize!=true, dontWriteAssembly!=true, slowCodegenWarnings==true, compilationTarget, outputPath)
+                if(!compilationResult.success)
+                    exitProcess(1)
+            } catch (x: ParsingFailedError) {
+                exitProcess(1)
+            } catch (x: AstException) {
+                exitProcess(1)
+            }
+
+            if (startEmulator==true) {
+                if (compilationResult.programName.isEmpty())
+                    println("\nCan't start emulator because no program was assembled.")
+                else {
+                    compilationResult.compTarget.machine.launchEmulator(compilationResult.programName)
+                }
             }
         }
-        println("\nTotal compilation+assemble time: ${totalTime / 1000.0} sec.")
-
-    } catch (px: ParsingFailedError) {
-        System.err.print("\u001b[91m")  // bright red
-        System.err.println(px.message)
-        System.err.print("\u001b[0m")  // reset
-        exitProcess(1)
-    } catch (x: Exception) {
-        print("\u001b[91m")  // bright red
-        println("\n* internal error *")
-        print("\u001b[0m")  // reset
-        System.out.flush()
-        throw x
-    } catch (x: NotImplementedError) {
-        print("\u001b[91m")  // bright red
-        println("\n* internal error: missing feature/code *")
-        print("\u001b[0m")  // reset
-        System.out.flush()
-        throw x
-    }
-
-    if(emulatorToStart.isNotEmpty()) {
-        println("\nStarting C-64 emulator $emulatorToStart...")
-        val cmdline = listOf(emulatorToStart, "-silent", "-moncommands", "$programname.vice-mon-list",
-                "-autostartprgmode", "1", "-autostart-warp", "-autostart", programname+".prg")
-        val process = ProcessBuilder(cmdline).inheritIO().start()
-        process.waitFor()
     }
 }
-
-fun determineCompilationOptions(moduleAst: Module): CompilationOptions {
-    val options = moduleAst.statements.filter { it is Directive && it.directive == "%option" }.flatMap { (it as Directive).args }.toSet()
-    val outputType = (moduleAst.statements.singleOrNull { it is Directive && it.directive == "%output" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    val launcherType = (moduleAst.statements.singleOrNull { it is Directive && it.directive == "%launcher" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    moduleAst.loadAddress = (moduleAst.statements.singleOrNull { it is Directive && it.directive == "%address" }
-            as? Directive)?.args?.single()?.int ?: 0
-    val zpoption: String? = (moduleAst.statements.singleOrNull { it is Directive && it.directive == "%zeropage" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    val floatsEnabled = options.any { it.name == "enable_floats" }
-    val zpType: ZeropageType =
-            if (zpoption == null)
-                if(floatsEnabled) ZeropageType.FLOATSAFE else ZeropageType.KERNALSAFE
-            else
-                try {
-                    ZeropageType.valueOf(zpoption)
-                } catch (x: IllegalArgumentException) {
-                    ZeropageType.KERNALSAFE
-                    // error will be printed by the astchecker
-                }
-    val zpReserved = moduleAst.statements
-            .asSequence()
-            .filter { it is Directive && it.directive == "%zpreserved" }
-            .map { (it as Directive).args }
-            .map { it[0].int!!..it[1].int!! }
-            .toList()
-
-    return CompilationOptions(
-            if (outputType == null) OutputType.PRG else OutputType.valueOf(outputType),
-            if (launcherType == null) LauncherType.BASIC else LauncherType.valueOf(launcherType),
-            zpType, zpReserved, floatsEnabled
-    )
-}
-
-private fun usage() {
-    System.err.println("Missing argument(s):")
-    System.err.println("    [-emu]       auto-start the 'x64' C-64 emulator after successful compilation")
-    System.err.println("    [-emu2]      auto-start the 'x64sc' C-64 emulator after successful compilation")
-    System.err.println("    [-writevm]   write intermediate vm code to a file as well")
-    System.err.println("    [-noasm]     don't create assembly code")
-    System.err.println("    [-vm]        launch the prog8 virtual machine instead of the compiler")
-    System.err.println("    modulefile   main module file to compile")
-    exitProcess(1)
-}

compiler/src/prog8/StackVmMain.kt

@@ -1,46 +0,0 @@
-package prog8
-
-import prog8.stackvm.*
-import java.awt.EventQueue
-import javax.swing.Timer
-import kotlin.system.exitProcess
-
-fun main(args: Array<String>) {
-    stackVmMain(args)
-}
-
-fun stackVmMain(args: Array<String>) {
-    printSoftwareHeader("StackVM")
-
-    if(args.size != 1) {
-        System.err.println("requires one argument: name of stackvm sourcecode file")
-        exitProcess(1)
-    }
-
-    val program = Program.load(args.first())
-    val vm = StackVm(traceOutputFile = null)
-    val dialog = ScreenDialog()
-    vm.load(program, dialog.canvas)
-    EventQueue.invokeLater {
-        dialog.pack()
-        dialog.isVisible = true
-        dialog.start()
-
-        val programTimer = Timer(10) { a ->
-            try {
-                vm.step()
-            } catch(bp: VmBreakpointException) {
-                println("Breakpoint: execution halted. Press enter to resume.")
-                readLine()
-            } catch (tx: VmTerminationException) {
-                println("Execution halted: ${tx.message}")
-                (a.source as Timer).stop()
-            }
-        }
-
-        val irqTimer = Timer(1000/60) { a -> vm.irq(a.`when`) }
-
-        programTimer.start()
-        irqTimer.start()
-    }
-}

compiler/src/prog8/ast/AstIdentifiersChecker.kt

@@ -1,239 +0,0 @@
-package prog8.ast
-
-import prog8.compiler.HeapValues
-import prog8.functions.BuiltinFunctions
-
-/**
- * Checks the validity of all identifiers (no conflicts)
- * Also builds a list of all (scoped) symbol definitions
- * Also makes sure that subroutine's parameters also become local variable decls in the subroutine's scope.
- * Finally, it also makes sure the datatype of all Var decls and sub Return values is set correctly.
- */
-
-fun Module.checkIdentifiers(heap: HeapValues): MutableMap<String, IStatement> {
-    val checker = AstIdentifiersChecker(heap)
-    this.process(checker)
-
-    // add any anonymous variables for heap values that are used, and replace literalvalue by identifierref
-    for (variable in checker.anonymousVariablesFromHeap) {
-        val scope = variable.first.definingScope()
-        scope.statements.add(variable.second)
-        val parent = variable.first.parent
-        when {
-            parent is Assignment && parent.value === variable.first -> {
-                val idref = IdentifierReference(listOf("auto_heap_value_${variable.first.heapId}"), variable.first.position)
-                idref.linkParents(parent)
-                parent.value = idref
-            }
-            parent is IFunctionCall -> {
-                val parameterPos = parent.arglist.indexOf(variable.first)
-                val idref = IdentifierReference(listOf("auto_heap_value_${variable.first.heapId}"), variable.first.position)
-                idref.linkParents(parent)
-                parent.arglist[parameterPos] = idref
-            }
-            else -> TODO("replace literalvalue by identifierref: $variable  (in $parent)")
-        }
-    }
-
-    printErrors(checker.result(), name)
-    return checker.symbols
-}
-
-
-private class AstIdentifiersChecker(val heap: HeapValues) : IAstProcessor {
-    private val checkResult: MutableList<AstException> = mutableListOf()
-
-    var symbols: MutableMap<String, IStatement> = mutableMapOf()
-        private set
-
-    fun result(): List<AstException> {
-        return checkResult
-    }
-
-    private fun nameError(name: String, position: Position, existing: IStatement) {
-        checkResult.add(NameError("name conflict '$name', first defined in ${existing.position.file} line ${existing.position.line}", position))
-    }
-
-    override fun process(block: Block): IStatement {
-        val scopedName = block.scopedname
-        val existing = symbols[scopedName]
-        if(existing!=null) {
-            nameError(block.name, block.position, existing)
-        } else {
-            symbols[scopedName] = block
-        }
-        return super.process(block)
-    }
-
-    override fun process(functionCall: FunctionCall): IExpression {
-        if(functionCall.target.nameInSource.size==1 && functionCall.target.nameInSource[0]=="lsb") {
-            // lsb(...) is just an alias for type cast to ubyte, so replace with "... as ubyte"
-            val typecast = TypecastExpression(functionCall.arglist.single(), DataType.UBYTE, functionCall.position)
-            typecast.linkParents(functionCall.parent)
-            return super.process(typecast)
-        }
-        return super.process(functionCall)
-    }
-
-    override fun process(decl: VarDecl): IStatement {
-        // first, check if there are datatype errors on the vardecl
-        decl.datatypeErrors.forEach { checkResult.add(it) }
-
-        // now check the identifier
-        if(decl.name in BuiltinFunctions)
-            // the builtin functions can't be redefined
-            checkResult.add(NameError("builtin function cannot be redefined", decl.position))
-
-        val scopedName = decl.scopedname
-        val existing = symbols[scopedName]
-        if(existing!=null) {
-            nameError(decl.name, decl.position, existing)
-        } else {
-            symbols[scopedName] = decl
-        }
-        return super.process(decl)
-    }
-
-    override fun process(subroutine: Subroutine): IStatement {
-        if(subroutine.name in BuiltinFunctions) {
-            // the builtin functions can't be redefined
-            checkResult.add(NameError("builtin function cannot be redefined", subroutine.position))
-        } else {
-            if (subroutine.parameters.any { it.name in BuiltinFunctions })
-                checkResult.add(NameError("builtin function name cannot be used as parameter", subroutine.position))
-
-            val scopedName = subroutine.scopedname
-            val existing = symbols[scopedName]
-            if (existing != null) {
-                nameError(subroutine.name, subroutine.position, existing)
-            } else {
-                symbols[scopedName] = subroutine
-            }
-
-            // check that there are no local variables that redefine the subroutine's parameters
-            val allDefinedNames = subroutine.allLabelsAndVariables()
-            val paramNames = subroutine.parameters.map { it.name }.toSet()
-            val paramsToCheck = paramNames.intersect(allDefinedNames)
-            for(name in paramsToCheck) {
-                val thing = subroutine.getLabelOrVariable(name)!!
-                if(thing.position != subroutine.position)
-                    nameError(name, thing.position, subroutine)
-            }
-
-            // inject subroutine params as local variables (if they're not there yet) (for non-kernel subroutines and non-asm parameters)
-            // NOTE:
-            // - numeric types BYTE and WORD and FLOAT are passed by value;
-            // - strings, arrays, matrices are passed by reference (their 16-bit address is passed as an uword parameter)
-            // - do NOT do this is the statement can be transformed into an asm subroutine later!
-            if(subroutine.asmAddress==null && !subroutine.canBeAsmSubroutine) {
-                if(subroutine.asmParameterRegisters.isEmpty()) {
-                    subroutine.parameters
-                            .filter { it.name !in allDefinedNames }
-                            .forEach {
-                                val vardecl = VarDecl(VarDeclType.VAR, it.type, false, null, it.name, null, subroutine.position)
-                                vardecl.linkParents(subroutine)
-                                subroutine.statements.add(0, vardecl)
-                            }
-                }
-            }
-        }
-        return super.process(subroutine)
-    }
-
-    override fun process(label: Label): IStatement {
-        if(label.name in BuiltinFunctions) {
-            // the builtin functions can't be redefined
-            checkResult.add(NameError("builtin function cannot be redefined", label.position))
-        } else {
-            val scopedName = label.scopedname
-            val existing = symbols[scopedName]
-            if (existing != null) {
-                nameError(label.name, label.position, existing)
-            } else {
-                symbols[scopedName] = label
-            }
-        }
-        return super.process(label)
-    }
-
-    override fun process(forLoop: ForLoop): IStatement {
-        // If the for loop has a decltype, it means to declare the loopvar inside the loop body
-        // rather than reusing an already declared loopvar from an outer scope.
-        // For loops that loop over an interable variable (instead of a range of numbers) get an
-        // additional interation count variable in their scope.
-        if(forLoop.loopRegister!=null) {
-            if(forLoop.decltype!=null)
-                checkResult.add(SyntaxError("register loop variables cannot be explicitly declared with a datatype", forLoop.position))
-            if(forLoop.loopRegister == Register.X)
-                printWarning("writing to the X register is dangerous, because it's used as an internal pointer", forLoop.position)
-        } else if(forLoop.loopVar!=null) {
-            val varName = forLoop.loopVar.nameInSource.last()
-            if(forLoop.decltype!=null) {
-                val existing = if(forLoop.body.isEmpty()) null else forLoop.body.lookup(forLoop.loopVar.nameInSource, forLoop.body.statements.first())
-                if(existing==null) {
-                    // create the local scoped for loop variable itself
-                    val vardecl = VarDecl(VarDeclType.VAR, forLoop.decltype, true, null, varName, null, forLoop.loopVar.position)
-                    vardecl.linkParents(forLoop.body)
-                    forLoop.body.statements.add(0, vardecl)
-                    forLoop.loopVar.parent = forLoop.body   // loopvar 'is defined in the body'
-                }
-
-            }
-
-            if(forLoop.iterable !is RangeExpr) {
-                val existing = if(forLoop.body.isEmpty()) null else forLoop.body.lookup(listOf(ForLoop.iteratorLoopcounterVarname), forLoop.body.statements.first())
-                if(existing==null) {
-                    // create loop iteration counter variable (without value, to avoid an assignment)
-                    val vardecl = VarDecl(VarDeclType.VAR, DataType.UBYTE, true, null, ForLoop.iteratorLoopcounterVarname, null, forLoop.loopVar.position)
-                    vardecl.linkParents(forLoop.body)
-                    forLoop.body.statements.add(0, vardecl)
-                    forLoop.loopVar.parent = forLoop.body   // loopvar 'is defined in the body'
-                }
-            }
-        }
-        return super.process(forLoop)
-    }
-
-    override fun process(assignTarget: AssignTarget): AssignTarget {
-        if(assignTarget.register==Register.X)
-            printWarning("writing to the X register is dangerous, because it's used as an internal pointer", assignTarget.position)
-        return super.process(assignTarget)
-    }
-
-    override fun process(returnStmt: Return): IStatement {
-        if(returnStmt.values.isNotEmpty()) {
-            // possibly adjust any literal values returned, into the desired returning data type
-            val subroutine = returnStmt.definingSubroutine()!!
-            if(subroutine.returntypes.size!=returnStmt.values.size)
-                return returnStmt  // mismatch in number of return values, error will be printed later.
-            val newValues = mutableListOf<IExpression>()
-            for(returnvalue in returnStmt.values.zip(subroutine.returntypes)) {
-                val lval = returnvalue.first as? LiteralValue
-                if(lval!=null) {
-                    val adjusted = lval.intoDatatype(returnvalue.second)
-                    if(adjusted!=null && adjusted !== lval)
-                        newValues.add(adjusted)
-                    else
-                        newValues.add(lval)
-                }
-                else
-                    newValues.add(returnvalue.first)
-            }
-            returnStmt.values = newValues
-        }
-        return super.process(returnStmt)
-    }
-
-
-    internal val anonymousVariablesFromHeap = mutableSetOf<Pair<LiteralValue, VarDecl>>()
-
-    override fun process(literalValue: LiteralValue): LiteralValue {
-        if(literalValue.heapId!=null && literalValue.parent !is VarDecl) {
-            // a literal value that's not declared as a variable, which refers to something on the heap.
-            // we need to introduce an auto-generated variable for this to be able to refer to the value!
-            val variable = VarDecl(VarDeclType.VAR, literalValue.type, false, null, "auto_heap_value_${literalValue.heapId}", literalValue, literalValue.position)
-            anonymousVariablesFromHeap.add(Pair(literalValue, variable))
-        }
-        return super.process(literalValue)
-    }
-}

compiler/src/prog8/ast/AstRecursionChecker.kt

@@ -1,120 +0,0 @@
-package prog8.ast
-
-/**
- * Checks for the occurrence of recursive subroutine calls
- */
-
-fun Module.checkRecursion(namespace: INameScope) {
-    val checker = AstRecursionChecker(namespace)
-    this.process(checker)
-    printErrors(checker.result(), name)
-}
-
-
-private class DirectedGraph<VT> {
-    private val graph = mutableMapOf<VT, MutableSet<VT>>()
-    private var uniqueVertices = mutableSetOf<VT>()
-    val numVertices : Int
-        get() = uniqueVertices.size
-
-    fun add(from: VT, to: VT) {
-        var targets = graph[from]
-        if(targets==null) {
-            targets = mutableSetOf()
-            graph[from] = targets
-        }
-        targets.add(to)
-        uniqueVertices.add(from)
-        uniqueVertices.add(to)
-    }
-
-    fun print() {
-        println("#vertices: $numVertices")
-        graph.forEach { from, to ->
-            println("$from   CALLS:")
-            to.forEach { println("   $it") }
-        }
-        val cycle = checkForCycle()
-        if(cycle.isNotEmpty()) {
-            println("CYCLIC!  $cycle")
-        }
-    }
-
-    fun checkForCycle(): MutableList<VT> {
-        val visited = uniqueVertices.associate { it to false }.toMutableMap()
-        val recStack = uniqueVertices.associate { it to false }.toMutableMap()
-        val cycle = mutableListOf<VT>()
-        for(node in uniqueVertices) {
-            if(isCyclicUntil(node, visited, recStack, cycle))
-                return cycle
-        }
-        return mutableListOf()
-    }
-
-    private fun isCyclicUntil(node: VT,
-                              visited: MutableMap<VT, Boolean>,
-                              recStack: MutableMap<VT, Boolean>,
-                              cycleNodes: MutableList<VT>): Boolean {
-
-        if(recStack[node]==true) return true
-        if(visited[node]==true) return false
-
-        // mark current node as visited and add to recursion stack
-        visited[node] = true
-        recStack[node] = true
-
-        // recurse for all neighbours
-        val neighbors = graph[node]
-        if(neighbors!=null) {
-            for (neighbour in neighbors) {
-                if (isCyclicUntil(neighbour, visited, recStack, cycleNodes)) {
-                    cycleNodes.add(node)
-                    return true
-                }
-            }
-        }
-
-        // pop node from recursion stack
-        recStack[node] = false
-        return false
-    }
-}
-
-
-private class AstRecursionChecker(private val namespace: INameScope) : IAstProcessor {
-    private val callGraph = DirectedGraph<INameScope>()
-
-    fun result(): List<AstException> {
-        val cycle = callGraph.checkForCycle()
-        if(cycle.isEmpty())
-            return emptyList()
-        val chain = cycle.joinToString(" <-- ") { "${it.name} at ${it.position}" }
-        return listOf(AstException("Program contains recursive subroutine calls, this is not supported. Recursive chain:\n (a subroutine call in) $chain"))
-    }
-
-    override fun process(functionCallStatement: FunctionCallStatement): IStatement {
-        val scope = functionCallStatement.definingScope()
-        val targetStatement = functionCallStatement.target.targetStatement(namespace)
-        if(targetStatement!=null) {
-            val targetScope = when (targetStatement) {
-                is Subroutine -> targetStatement
-                else -> targetStatement.definingScope()
-            }
-            callGraph.add(scope, targetScope)
-        }
-        return super.process(functionCallStatement)
-    }
-
-    override fun process(functionCall: FunctionCall): IExpression {
-        val scope = functionCall.definingScope()
-        val targetStatement = functionCall.target.targetStatement(namespace)
-        if(targetStatement!=null) {
-            val targetScope = when (targetStatement) {
-                is Subroutine -> targetStatement
-                else -> targetStatement.definingScope()
-            }
-            callGraph.add(scope, targetScope)
-        }
-        return super.process(functionCall)
-    }
-}

compiler/src/prog8/ast/ImportedAstChecker.kt

@@ -1,43 +0,0 @@
-package prog8.ast
-
-
-/**
- * Checks that are specific for imported modules.
- */
-
-fun Module.checkImportedValid() {
-    val checker = ImportedAstChecker()
-    this.linkParents()
-    this.process(checker)
-    printErrors(checker.result(), name)
-}
-
-
-private class ImportedAstChecker : IAstProcessor {
-    private val checkResult: MutableList<SyntaxError> = mutableListOf()
-
-    fun result(): List<SyntaxError> {
-        return checkResult
-    }
-
-    /**
-     * Module check: most global directives don't apply for imported modules
-     */
-    override fun process(module: Module) {
-        super.process(module)
-        val newStatements : MutableList<IStatement> = mutableListOf()
-
-        val moduleLevelDirectives = listOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address")
-        for (sourceStmt in module.statements) {
-            val stmt = sourceStmt.process(this)
-            if(stmt is Directive && stmt.parent is Module) {
-                if(stmt.directive in moduleLevelDirectives) {
-                    printWarning("ignoring module directive because it was imported", stmt.position, stmt.directive)
-                    continue
-                }
-            }
-            newStatements.add(stmt)
-        }
-        module.statements = newStatements
-    }
-}

compiler/src/prog8/ast/StmtReorderer.kt

@@ -1,255 +0,0 @@
-package prog8.ast
-
-import prog8.compiler.HeapValues
-
-fun Module.reorderStatements(namespace: INameScope, heap: HeapValues) {
-    val initvalueCreator = VarInitValueCreator()
-    this.process(initvalueCreator)
-
-    val checker = StatementReorderer(namespace, heap)
-    this.process(checker)
-}
-
-const val initvarsSubName="prog8_init_vars"    // the name of the subroutine that should be called for every block to initialize its variables
-
-
-private class StatementReorderer(private val namespace: INameScope, private val heap: HeapValues): IAstProcessor {
-    // Reorders the statements in a way the compiler needs.
-    // - 'main' block must be the very first statement UNLESS it has an address set.
-    // - blocks are ordered by address, where blocks without address are put at the end.
-    // - in every scope:
-    //      -- the directives '%output', '%launcher', '%zeropage', '%zpreserved', '%address' and '%option' will come first.
-    //      -- all vardecls then follow.
-    //      -- the remaining statements then follow in their original order.
-    //
-    // - the 'start' subroutine in the 'main' block will be moved to the top immediately following the directives.
-    // - all other subroutines will be moved to the end of their block.
-
-    private val directivesToMove = setOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address", "%option")
-
-    override fun process(module: Module) {
-        super.process(module)
-
-        val (blocks, other) = module.statements.partition { it is Block }
-        module.statements = other.asSequence().plus(blocks.sortedBy { (it as Block).address ?: Int.MAX_VALUE }).toMutableList()
-
-        // make sure user-defined blocks come BEFORE library blocks, and move the "main" block to the top of everything
-        val nonLibraryBlocks = module.statements.withIndex()
-                .filter { it.value is Block && !(it.value as Block).isInLibrary }
-                .map { it.index to it.value }
-                .reversed()
-        for(nonLibBlock in nonLibraryBlocks)
-            module.statements.removeAt(nonLibBlock.first)
-        for(nonLibBlock in nonLibraryBlocks)
-            module.statements.add(0, nonLibBlock.second)
-        val mainBlock = module.statements.single { it is Block && it.name=="main" }
-        if((mainBlock as Block).address==null) {
-            module.statements.remove(mainBlock)
-            module.statements.add(0, mainBlock)
-        }
-
-        val varDecls = module.statements.filterIsInstance<VarDecl>()
-        module.statements.removeAll(varDecls)
-        module.statements.addAll(0, varDecls)
-
-        val directives = module.statements.filter {it is Directive && it.directive in directivesToMove}
-        module.statements.removeAll(directives)
-        module.statements.addAll(0, directives)
-
-        sortConstantAssignments(module.statements)
-    }
-
-    override fun process(block: Block): IStatement {
-
-        val subroutines = block.statements.filterIsInstance<Subroutine>()
-        var numSubroutinesAtEnd = 0
-        // move all subroutines to the end of the block
-        for (subroutine in subroutines) {
-            if(subroutine.name!="start" || block.name!="main") {
-                block.statements.remove(subroutine)
-                block.statements.add(subroutine)
-            }
-            numSubroutinesAtEnd++
-        }
-        // move the "start" subroutine to the top
-        if(block.name=="main") {
-            block.statements.singleOrNull { it is Subroutine && it.name == "start" } ?.let {
-                block.statements.remove(it)
-                block.statements.add(0, it)
-                numSubroutinesAtEnd--
-            }
-        }
-
-        // make sure there is a 'return' in front of the first subroutine
-        // (if it isn't the first statement in the block itself, and isn't the program's entrypoint)
-        if(numSubroutinesAtEnd>0 && block.statements.size > (numSubroutinesAtEnd+1)) {
-            val firstSub = block.statements[block.statements.size - numSubroutinesAtEnd] as Subroutine
-            if(firstSub.name != "start" && block.name != "main") {
-                val stmtBeforeFirstSub = block.statements[block.statements.size - numSubroutinesAtEnd - 1]
-                if (stmtBeforeFirstSub !is Return
-                        && stmtBeforeFirstSub !is Jump
-                        && stmtBeforeFirstSub !is Subroutine
-                        && stmtBeforeFirstSub !is BuiltinFunctionStatementPlaceholder) {
-                    val ret = Return(emptyList(), stmtBeforeFirstSub.position)
-                    ret.linkParents(block)
-                    block.statements.add(block.statements.size - numSubroutinesAtEnd, ret)
-                }
-            }
-        }
-
-        val varDecls = block.statements.filter { it is VarDecl }
-        block.statements.removeAll(varDecls)
-        block.statements.addAll(0, varDecls)
-        val directives = block.statements.filter {it is Directive && it.directive in directivesToMove}
-        block.statements.removeAll(directives)
-        block.statements.addAll(0, directives)
-
-        sortConstantAssignments(block.statements)
-
-        val varInits = block.statements.withIndex().filter { it.value is VariableInitializationAssignment }
-        if(varInits.isNotEmpty()) {
-            val statements = varInits.map{it.value}.toMutableList()
-            val varInitSub = Subroutine(initvarsSubName, emptyList(), emptyList(), emptyList(), emptyList(),
-                    emptySet(), null, false, statements, block.position)
-            varInitSub.linkParents(block)
-            block.statements.add(varInitSub)
-
-            // remove the varinits from the block's statements
-            for(index in varInits.map{it.index}.reversed())
-                block.statements.removeAt(index)
-        }
-
-        return super.process(block)
-    }
-
-    override fun process(subroutine: Subroutine): IStatement {
-        super.process(subroutine)
-
-        sortConstantAssignments(subroutine.statements)
-
-        val varDecls = subroutine.statements.filterIsInstance<VarDecl>()
-        subroutine.statements.removeAll(varDecls)
-        subroutine.statements.addAll(0, varDecls)
-        val directives = subroutine.statements.filter {it is Directive && it.directive in directivesToMove}
-        subroutine.statements.removeAll(directives)
-        subroutine.statements.addAll(0, directives)
-
-        if(subroutine.returntypes.isEmpty()) {
-            // add the implicit return statement at the end (if it's not there yet), but only if it's not a kernel routine.
-            // and if an assembly block doesn't contain a rts/rti
-            if(subroutine.asmAddress==null && subroutine.amountOfRtsInAsm()==0) {
-                if (subroutine.statements.lastOrNull {it !is VarDecl} !is Return) {
-                    val returnStmt = Return(emptyList(), subroutine.position)
-                    returnStmt.linkParents(subroutine)
-                    subroutine.statements.add(returnStmt)
-                }
-            }
-        }
-
-        return subroutine
-    }
-
-    override fun process(scope: AnonymousScope): AnonymousScope {
-        scope.statements = scope.statements.map { it.process(this)}.toMutableList()
-        sortConstantAssignments(scope.statements)
-        return scope
-    }
-
-    private fun sortConstantAssignments(statements: MutableList<IStatement>) {
-        // sort assignments by datatype and value, so multiple initializations with the same value can be optimized (to load the value just once)
-        val result = mutableListOf<IStatement>()
-        val stmtIter = statements.iterator()
-        for(stmt in stmtIter) {
-            if(stmt is Assignment) {
-                val constval = stmt.value.constValue(namespace, heap)
-                if(constval!=null) {
-                    val (sorted, trailing) = sortConstantAssignmentSequence(stmt, stmtIter)
-                    result.addAll(sorted)
-                    if(trailing!=null)
-                        result.add(trailing)
-                }
-                else
-                    result.add(stmt)
-            }
-            else
-                result.add(stmt)
-        }
-        statements.clear()
-        statements.addAll(result)
-    }
-
-    private fun sortConstantAssignmentSequence(first: Assignment, stmtIter: MutableIterator<IStatement>): Pair<List<Assignment>, IStatement?> {
-        val sequence= mutableListOf(first)
-        var trailing: IStatement? = null
-        while(stmtIter.hasNext()) {
-            val next = stmtIter.next()
-            if(next is Assignment) {
-                val constValue = next.value.constValue(namespace, heap)
-                if(constValue==null) {
-                    trailing = next
-                    break
-                }
-                sequence.add(next)
-            }
-            else {
-                trailing=next
-                break
-            }
-        }
-        val sorted = sequence.sortedWith(compareBy({it.value.resultingDatatype(namespace, heap)}, {it.singleTarget?.shortString(true)}))
-        return Pair(sorted, trailing)
-    }
-
-}
-
-
-private class VarInitValueCreator: IAstProcessor {
-    // Replace the var decl with an assignment and add a new vardecl with the default constant value.
-    // This makes sure the variables get reset to the intended value on a next run of the program.
-    // Variable decls without a value don't get this treatment, which means they retain the last
-    // value they had when restarting the program.
-    // This is done in a separate step because it interferes with the namespace lookup of symbols
-    // in other ast processors.
-
-    private val vardeclsToAdd = mutableMapOf<INameScope, MutableList<VarDecl>>()
-
-    override fun process(module: Module) {
-        super.process(module)
-
-        // add any new vardecls to the various scopes
-        for(decl in vardeclsToAdd)
-            for(d in decl.value) {
-                d.linkParents(decl.key as Node)
-                decl.key.statements.add(0, d)
-            }
-    }
-
-    override fun process(decl: VarDecl): IStatement {
-        super.process(decl)
-        if(decl.type!=VarDeclType.VAR || decl.value==null)
-            return decl
-
-        if(decl.datatype in NumericDatatypes) {
-            val scope = decl.definingScope()
-            if(scope !in vardeclsToAdd)
-                vardeclsToAdd[scope] = mutableListOf()
-            vardeclsToAdd[scope]!!.add(decl.asDefaultValueDecl(null))
-            val declvalue = decl.value!!
-            val value =
-                    if(declvalue is LiteralValue) {
-                        val converted = declvalue.intoDatatype(decl.datatype)
-                        converted ?: declvalue
-                    }
-                    else
-                        declvalue
-            return VariableInitializationAssignment(
-                    AssignTarget(null, IdentifierReference(decl.scopedname.split("."), decl.position), null, null, decl.position),
-                    null,
-                    value,
-                    decl.position
-            )
-        }
-        return decl
-    }
-
-}

compiler/src/prog8/compiler/AssemblyError.kt

@@ -0,0 +1,3 @@
+package prog8.compiler
+
+internal class AssemblyError(msg: String) : RuntimeException(msg)

compiler/src/prog8/compiler/BeforeAsmGenerationAstChanger.kt

@@ -0,0 +1,299 @@
+package prog8.compiler
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.target.ICompilationTarget
+
+
+internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: IErrorReporter, private val compTarget: ICompilationTarget) : AstWalker() {
+
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        subroutineVariables.add(decl.name to decl)
+        if (decl.value == null && !decl.autogeneratedDontRemove && decl.type == VarDeclType.VAR && decl.datatype in NumericDatatypes) {
+            // a numeric vardecl without an initial value is initialized with zero,
+            // unless there's already an assignment below, that initializes the value
+            if(decl.allowInitializeWithZero)
+            {
+                val nextAssign = decl.definingScope().nextSibling(decl) as? Assignment
+                if (nextAssign != null && nextAssign.target.isSameAs(IdentifierReference(listOf(decl.name), Position.DUMMY)))
+                    decl.value = null
+                else
+                    decl.value = decl.zeroElementValue()
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        // Try to replace A = B <operator> Something  by A= B, A = A <operator> Something
+        // this triggers the more efficent augmented assignment code generation more often.
+        // But it can only be done if the target variable IS NOT OCCURRING AS AN OPERAND ITSELF.
+        if(!assignment.isAugmentable
+                && assignment.target.identifier != null
+                && compTarget.isInRegularRAM(assignment.target, program)) {
+            val binExpr = assignment.value as? BinaryExpression
+            if (binExpr != null && binExpr.operator !in comparisonOperators) {
+                if (binExpr.left !is BinaryExpression) {
+                    if (binExpr.right.referencesIdentifier(*assignment.target.identifier!!.nameInSource.toTypedArray())) {
+                        // the right part of the expression contains the target variable itself.
+                        // we can't 'split' it trivially because the variable will be changed halfway through.
+                        if(binExpr.operator in associativeOperators) {
+                            // A = <something-without-A>  <associativeoperator>  <otherthing-with-A>
+                            // use the other part of the expression to split.
+                            val assignRight = Assignment(assignment.target, binExpr.right, assignment.position)
+                            return listOf(
+                                    IAstModification.InsertBefore(assignment, assignRight, assignment.definingScope()),
+                                    IAstModification.ReplaceNode(binExpr.right, binExpr.left, binExpr),
+                                    IAstModification.ReplaceNode(binExpr.left, assignment.target.toExpression(), binExpr))
+                        }
+                    } else {
+                        val assignLeft = Assignment(assignment.target, binExpr.left, assignment.position)
+                        return listOf(
+                                IAstModification.InsertBefore(assignment, assignLeft, assignment.definingScope()),
+                                IAstModification.ReplaceNode(binExpr.left, assignment.target.toExpression(), binExpr))
+                    }
+                }
+            }
+        }
+        return noModifications
+    }
+
+    private val subroutineVariables = mutableListOf<Pair<String, VarDecl>>()
+
+    override fun before(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        subroutineVariables.clear()
+        return noModifications
+    }
+
+    override fun after(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
+        val decls = scope.statements.filterIsInstance<VarDecl>()
+        subroutineVariables.addAll(decls.map { it.name to it })
+
+        val sub = scope.definingSubroutine()
+        if (sub != null) {
+            // move vardecls of the scope into the upper scope. Make sure the position remains the same!
+            val numericVarsWithValue = decls.filter { it.value != null && it.datatype in NumericDatatypes }
+            val replaceVardecls =numericVarsWithValue.map {
+                val initValue = it.value!!  // assume here that value has always been set by now
+                it.value = null     // make sure no value init assignment for this vardecl will be created later (would be superfluous)
+                val target = AssignTarget(IdentifierReference(listOf(it.name), it.position), null, null, it.position)
+                val assign = Assignment(target, initValue, it.position)
+                initValue.parent = assign
+                IAstModification.ReplaceNode(it, assign, scope)
+            }
+            val moveVardeclsUp = decls.map { IAstModification.InsertFirst(it, sub) }
+            return replaceVardecls + moveVardeclsUp
+        }
+        return noModifications
+    }
+
+    override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        val firstDeclarations = mutableMapOf<String, VarDecl>()
+        for(decl in subroutineVariables) {
+            val existing = firstDeclarations[decl.first]
+            if(existing!=null && existing !== decl.second) {
+                errors.err("variable ${decl.first} already defined in subroutine ${subroutine.name} at ${existing.position}", decl.second.position)
+            } else {
+                firstDeclarations[decl.first] = decl.second
+            }
+        }
+
+
+        // add the implicit return statement at the end (if it's not there yet), but only if it's not a kernal routine.
+        // and if an assembly block doesn't contain a rts/rti, and some other situations.
+        val mods = mutableListOf<IAstModification>()
+        val returnStmt = Return(null, subroutine.position)
+        if (subroutine.asmAddress == null
+                && !subroutine.inline
+                && subroutine.statements.isNotEmpty()
+                && subroutine.amountOfRtsInAsm() == 0
+                && subroutine.statements.lastOrNull { it !is VarDecl } !is Return
+                && subroutine.statements.last() !is Subroutine) {
+            mods += IAstModification.InsertLast(returnStmt, subroutine)
+        }
+
+        // precede a subroutine with a return to avoid falling through into the subroutine from code above it
+        val outerScope = subroutine.definingScope()
+        val outerStatements = outerScope.statements
+        val subroutineStmtIdx = outerStatements.indexOf(subroutine)
+        if (subroutineStmtIdx > 0
+                && outerStatements[subroutineStmtIdx - 1] !is Jump
+                && outerStatements[subroutineStmtIdx - 1] !is Subroutine
+                && outerStatements[subroutineStmtIdx - 1] !is Return
+                && outerScope !is Block) {
+            mods += IAstModification.InsertAfter(outerStatements[subroutineStmtIdx - 1], returnStmt, outerScope)
+        }
+        return mods
+    }
+
+    override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
+        // see if we can remove superfluous typecasts (outside of expressions)
+        // such as casting byte<->ubyte,  word<->uword
+        // Also the special typecast of a reference type (str, array) to an UWORD will be changed into address-of.
+        val sourceDt = typecast.expression.inferType(program).typeOrElse(DataType.STRUCT)
+        if (typecast.type in ByteDatatypes && sourceDt in ByteDatatypes
+                || typecast.type in WordDatatypes && sourceDt in WordDatatypes) {
+            if(typecast.parent !is Expression) {
+                return listOf(IAstModification.ReplaceNode(typecast, typecast.expression, parent))
+            }
+        }
+
+
+        // Note: for various reasons (most importantly, code simplicity), the code generator assumes/requires
+        // that the types of assignment values and their target are the same,
+        // and that the types of both operands of a binaryexpression node are the same.
+        // So, it is not easily possible to remove the typecasts that are there to make these conditions true.
+        // The only place for now where we can do this is for:
+        //    asmsub register pair parameter.
+
+        if(sourceDt in PassByReferenceDatatypes) {
+            if(typecast.type==DataType.UWORD) {
+                if(typecast.expression is IdentifierReference) {
+                    return listOf(IAstModification.ReplaceNode(
+                            typecast,
+                            AddressOf(typecast.expression as IdentifierReference, typecast.position),
+                            parent
+                    ))
+                } else if(typecast.expression is IFunctionCall) {
+                    return listOf(IAstModification.ReplaceNode(
+                            typecast,
+                            typecast.expression,
+                            parent
+                    ))
+                }
+            } else {
+                errors.err("cannot cast pass-by-reference value to type ${typecast.type} (only to UWORD)", typecast.position)
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(ifStatement: IfStatement, parent: Node): Iterable<IAstModification> {
+        val binExpr = ifStatement.condition as? BinaryExpression
+        if(binExpr==null || binExpr.operator !in comparisonOperators) {
+            // if x  ->  if x!=0,    if x+5  ->  if x+5 != 0
+            val booleanExpr = BinaryExpression(ifStatement.condition, "!=", NumericLiteralValue.optimalInteger(0, ifStatement.condition.position), ifStatement.condition.position)
+            return listOf(IAstModification.ReplaceNode(ifStatement.condition, booleanExpr, ifStatement))
+        }
+        return noModifications
+    }
+
+    override fun after(untilLoop: UntilLoop, parent: Node): Iterable<IAstModification> {
+        val binExpr = untilLoop.condition as? BinaryExpression
+        if(binExpr==null || binExpr.operator !in comparisonOperators) {
+            // until x  ->  until x!=0,    until x+5  ->  until x+5 != 0
+            val booleanExpr = BinaryExpression(untilLoop.condition, "!=", NumericLiteralValue.optimalInteger(0, untilLoop.condition.position), untilLoop.condition.position)
+            return listOf(IAstModification.ReplaceNode(untilLoop.condition, booleanExpr, untilLoop))
+        }
+        return noModifications
+    }
+
+    override fun after(whileLoop: WhileLoop, parent: Node): Iterable<IAstModification> {
+        val binExpr = whileLoop.condition as? BinaryExpression
+        if(binExpr==null || binExpr.operator !in comparisonOperators) {
+            // while x  ->  while x!=0,    while x+5  ->  while x+5 != 0
+            val booleanExpr = BinaryExpression(whileLoop.condition, "!=", NumericLiteralValue.optimalInteger(0, whileLoop.condition.position), whileLoop.condition.position)
+            return listOf(IAstModification.ReplaceNode(whileLoop.condition, booleanExpr, whileLoop))
+        }
+        return noModifications
+    }
+
+    override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        if(functionCallStatement.target.nameInSource==listOf("cmp")) {
+            // if the datatype of the arguments of cmp() are different, cast the byte one to word.
+            val arg1 = functionCallStatement.args[0]
+            val arg2 = functionCallStatement.args[1]
+            val dt1 = arg1.inferType(program).typeOrElse(DataType.STRUCT)
+            val dt2 = arg2.inferType(program).typeOrElse(DataType.STRUCT)
+            if(dt1 in ByteDatatypes) {
+                if(dt2 in ByteDatatypes)
+                    return noModifications
+                val cast1 = TypecastExpression(arg1, if(dt1==DataType.UBYTE) DataType.UWORD else DataType.WORD, true, functionCallStatement.position)
+                return listOf(IAstModification.ReplaceNode(arg1, cast1, functionCallStatement))
+            } else {
+                if(dt2 in WordDatatypes)
+                    return noModifications
+                val cast2 = TypecastExpression(arg2, if(dt2==DataType.UBYTE) DataType.UWORD else DataType.WORD, true, functionCallStatement.position)
+                return listOf(IAstModification.ReplaceNode(arg2, cast2, functionCallStatement))
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
+
+        val containingStatement = getContainingStatement(arrayIndexedExpression)
+        if(getComplexArrayIndexedExpressions(containingStatement).size > 1) {
+            errors.err("it's not possible to use more than one complex array indexing expression in a single statement; break it up via a temporary variable for instance", containingStatement.position)
+            return noModifications
+        }
+
+
+        val index = arrayIndexedExpression.indexer.indexExpr
+        if(index !is NumericLiteralValue && index !is IdentifierReference) {
+            // replace complex indexing expression with a temp variable to hold the computed index first
+            return getAutoIndexerVarFor(arrayIndexedExpression)
+        }
+
+        return noModifications
+    }
+
+    private fun getComplexArrayIndexedExpressions(stmt: Statement): List<ArrayIndexedExpression> {
+
+        class Searcher : IAstVisitor {
+            val complexArrayIndexedExpressions = mutableListOf<ArrayIndexedExpression>()
+            override fun visit(arrayIndexedExpression: ArrayIndexedExpression) {
+                val ix = arrayIndexedExpression.indexer.indexExpr
+                if(ix !is NumericLiteralValue && ix !is IdentifierReference)
+                    complexArrayIndexedExpressions.add(arrayIndexedExpression)
+            }
+
+            override fun visit(branchStatement: BranchStatement) {}
+
+            override fun visit(forLoop: ForLoop) {}
+
+            override fun visit(ifStatement: IfStatement) {
+                ifStatement.condition.accept(this)
+            }
+
+            override fun visit(untilLoop: UntilLoop) {
+                untilLoop.condition.accept(this)
+            }
+        }
+
+        val searcher = Searcher()
+        stmt.accept(searcher)
+        return searcher.complexArrayIndexedExpressions
+    }
+
+    private fun getContainingStatement(expression: Expression): Statement {
+        var node: Node = expression
+        while(node !is Statement)
+            node = node.parent
+
+        return node
+    }
+
+    private fun getAutoIndexerVarFor(expr: ArrayIndexedExpression): MutableList<IAstModification> {
+        val modifications = mutableListOf<IAstModification>()
+        val statement = expr.containingStatement()
+        // replace the indexer with just the variable (simply use a cx16 virtual register r9, that we HOPE is not used for other things in the expression...)
+        // assign the indexing expression to the helper variable, but only if that hasn't been done already
+        val target = AssignTarget(IdentifierReference(listOf("cx16", "r9"), expr.indexer.position), null, null, expr.indexer.position)
+        val assign = Assignment(target, expr.indexer.indexExpr, expr.indexer.position)
+        modifications.add(IAstModification.InsertBefore(statement, assign, statement.definingScope()))
+        modifications.add(IAstModification.ReplaceNode(expr.indexer.indexExpr, target.identifier!!.copy(), expr.indexer))
+        return modifications
+    }
+
+}

compiler/src/prog8/compiler/ErrorReporting.kt

@@ -0,0 +1,57 @@
+package prog8.compiler
+
+import prog8.ast.base.Position
+import prog8.parser.ParsingFailedError
+
+
+interface IErrorReporter {
+    fun err(msg: String, position: Position)
+    fun warn(msg: String, position: Position)
+    fun isEmpty(): Boolean
+    fun report()
+}
+
+
+internal class ErrorReporter: IErrorReporter {
+    private enum class MessageSeverity {
+        WARNING,
+        ERROR
+    }
+    private class CompilerMessage(val severity: MessageSeverity, val message: String, val position: Position)
+
+    private val messages = mutableListOf<CompilerMessage>()
+    private val alreadyReportedMessages = mutableSetOf<String>()
+
+    override fun err(msg: String, position: Position) {
+        messages.add(CompilerMessage(MessageSeverity.ERROR, msg, position))
+    }
+    override fun warn(msg: String, position: Position) {
+        messages.add(CompilerMessage(MessageSeverity.WARNING, msg, position))
+    }
+
+    override fun report() {
+        var numErrors = 0
+        var numWarnings = 0
+        messages.forEach {
+            when(it.severity) {
+                MessageSeverity.ERROR -> System.err.print("\u001b[91m")  // bright red
+                MessageSeverity.WARNING -> System.err.print("\u001b[93m")  // bright yellow
+            }
+            val msg = "${it.position.toClickableStr()} ${it.severity} ${it.message}".trim()
+            if(msg !in alreadyReportedMessages) {
+                System.err.println(msg)
+                alreadyReportedMessages.add(msg)
+                when(it.severity) {
+                    MessageSeverity.WARNING -> numWarnings++
+                    MessageSeverity.ERROR -> numErrors++
+                }
+            }
+            System.err.print("\u001b[0m")  // reset color
+        }
+        messages.clear()
+        if(numErrors>0)
+            throw ParsingFailedError("There are $numErrors errors and $numWarnings warnings.")
+    }
+
+    override fun isEmpty() = messages.isEmpty()
+}

compiler/src/prog8/compiler/Zeropage.kt

@@ -1,6 +1,6 @@
 package prog8.compiler
 
-import prog8.ast.*
+import prog8.ast.base.*
 
 
 class ZeropageDepletedError(message: String) : Exception(message)
@@ -8,26 +8,35 @@ class ZeropageDepletedError(message: String) : Exception(message)
 
 abstract class Zeropage(protected val options: CompilationOptions) {
 
+    abstract val SCRATCH_B1 : Int      // temp storage for a single byte
+    abstract val SCRATCH_REG : Int     // temp storage for a register
+    abstract val SCRATCH_W1 : Int      // temp storage 1 for a word  $fb+$fc
+    abstract val SCRATCH_W2 : Int      // temp storage 2 for a word  $fb+$fc
+
+
     private val allocations = mutableMapOf<Int, Pair<String, DataType>>()
     val free = mutableListOf<Int>()     // subclasses must set this to the appropriate free locations.
 
     val allowedDatatypes = NumericDatatypes
 
-    fun available() = free.size
+    fun available() = if(options.zeropage==ZeropageType.DONTUSE) 0 else free.size
 
-    fun allocate(scopedname: String, datatype: DataType, position: Position?): Int {
-        assert(scopedname.isEmpty() || !allocations.values.any { it.first==scopedname } ) {"same scopedname can't be allocated twice"}
+    fun allocate(scopedname: String, datatype: DataType, position: Position?, errors: IErrorReporter): Int {
+        assert(scopedname.isEmpty() || !allocations.values.any { it.first==scopedname } ) {"scopedname can't be allocated twice"}
+
+        if(options.zeropage==ZeropageType.DONTUSE)
+            throw CompilerException("zero page usage has been disabled")
 
         val size =
                 when (datatype) {
-                    DataType.UBYTE, DataType.BYTE -> 1
-                    DataType.UWORD, DataType.WORD -> 2
+                    in ByteDatatypes -> 1
+                    in WordDatatypes -> 2
                     DataType.FLOAT -> {
                         if (options.floats) {
                             if(position!=null)
-                                printWarning("allocated a large value (float) in zeropage", position)
+                                errors.warn("allocated a large value (float) in zeropage", position)
                             else
-                                printWarning("$scopedname: allocated a large value (float) in zeropage")
+                                errors.warn("$scopedname: allocated a large value (float) in zeropage", position ?: Position.DUMMY)
                             5
                         } else throw CompilerException("floating point option not enabled")
                     }
@@ -36,13 +45,13 @@ abstract class Zeropage(protected val options: CompilationOptions) {
 
         if(free.size > 0) {
             if(size==1) {
-                for(candidate in free.min()!! .. free.max()!!+1) {
+                for(candidate in free.minOrNull()!! .. free.maxOrNull()!!+1) {
                     if(loneByte(candidate))
                         return makeAllocation(candidate, 1, datatype, scopedname)
                 }
                 return makeAllocation(free[0], 1, datatype, scopedname)
             }
-            for(candidate in free.min()!! .. free.max()!!+1) {
+            for(candidate in free.minOrNull()!! .. free.maxOrNull()!!+1) {
                 if (sequentialFree(candidate, size))
                     return makeAllocation(candidate, size, datatype, scopedname)
             }
@@ -55,18 +64,10 @@ abstract class Zeropage(protected val options: CompilationOptions) {
 
     private fun makeAllocation(address: Int, size: Int, datatype: DataType, name: String?): Int {
         free.removeAll(address until address+size)
-        allocations[address] = Pair(name ?: "<unnamed>", datatype)
+        allocations[address] = (name ?: "<unnamed>") to datatype
         return address
     }
 
     private fun loneByte(address: Int) = address in free && address-1 !in free && address+1 !in free
     private fun sequentialFree(address: Int, size: Int) = free.containsAll((address until address+size).toList())
-
-    enum class ExitProgramStrategy {
-        CLEAN_EXIT,
-        SYSTEM_RESET
-    }
-
-    abstract val exitProgramStrategy: ExitProgramStrategy
-
 }

compiler/src/prog8/compiler/astprocessing/AstExtensions.kt

@@ -0,0 +1,97 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.Program
+import prog8.ast.base.FatalAstException
+import prog8.ast.statements.Directive
+import prog8.compiler.BeforeAsmGenerationAstChanger
+import prog8.compiler.CompilationOptions
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
+
+
+internal fun Program.checkValid(compilerOptions: CompilationOptions, errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val checker = AstChecker(this, compilerOptions, errors, compTarget)
+    checker.visit(this)
+}
+
+internal fun Program.processAstBeforeAsmGeneration(errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val fixer = BeforeAsmGenerationAstChanger(this, errors, compTarget)
+    fixer.visit(this)
+    fixer.applyModifications()
+}
+
+internal fun Program.reorderStatements(errors: IErrorReporter) {
+    val reorder = StatementReorderer(this, errors)
+    reorder.visit(this)
+    reorder.applyModifications()
+}
+
+internal fun Program.addTypecasts(errors: IErrorReporter) {
+    val caster = TypecastsAdder(this, errors)
+    caster.visit(this)
+    caster.applyModifications()
+}
+
+internal fun Program.verifyFunctionArgTypes() {
+    val fixer = VerifyFunctionArgTypes(this)
+    fixer.visit(this)
+}
+
+internal fun Program.checkIdentifiers(errors: IErrorReporter, compTarget: ICompilationTarget) {
+
+    val checker2 = AstIdentifiersChecker(this, errors, compTarget)
+    checker2.visit(this)
+
+    if(errors.isEmpty()) {
+        val transforms = AstVariousTransforms(this)
+        transforms.visit(this)
+        transforms.applyModifications()
+        val lit2decl = LiteralsToAutoVars(this)
+        lit2decl.visit(this)
+        lit2decl.applyModifications()
+    }
+
+    if (modules.map { it.name }.toSet().size != modules.size) {
+        throw FatalAstException("modules should all be unique")
+    }
+}
+
+internal fun Program.variousCleanups() {
+    val process = VariousCleanups()
+    process.visit(this)
+    process.applyModifications()
+}
+
+internal fun Program.moveMainAndStartToFirst() {
+    // the module containing the program entrypoint is moved to the first in the sequence.
+    // the "main" block containing the entrypoint is moved to the top in there,
+    // and finally the entrypoint subroutine "start" itself is moved to the top in that block.
+
+    val directives = modules[0].statements.filterIsInstance<Directive>()
+    val start = this.entrypoint()
+    if(start!=null) {
+        val mod = start.definingModule()
+        val block = start.definingBlock()
+        if(!modules.remove(mod))
+            throw FatalAstException("module wrong")
+        modules.add(0, mod)
+        mod.remove(block)
+        var afterDirective = mod.statements.indexOfFirst { it !is Directive }
+        if(afterDirective<0)
+            mod.statements.add(block)
+        else
+            mod.statements.add(afterDirective, block)
+        block.remove(start)
+        afterDirective = block.statements.indexOfFirst { it !is Directive }
+        if(afterDirective<0)
+            block.statements.add(start)
+        else
+            block.statements.add(afterDirective, start)
+
+        // overwrite the directives in the module containing the entrypoint
+        for(directive in directives) {
+            modules[0].statements.removeAll { it is Directive && it.directive == directive.directive }
+            modules[0].statements.add(0, directive)
+        }
+    }
+}

compiler/src/prog8/compiler/astprocessing/AstIdentifiersChecker.kt

@@ -0,0 +1,182 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.Module
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.base.NumericDatatypes
+import prog8.ast.base.Position
+import prog8.ast.expressions.ArrayLiteralValue
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.expressions.StringLiteralValue
+import prog8.ast.statements.*
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+import prog8.compiler.target.ICompilationTarget
+
+internal class AstIdentifiersChecker(private val program: Program, private val errors: IErrorReporter, private val compTarget: ICompilationTarget) : IAstVisitor {
+    private var blocks = mutableMapOf<String, Block>()
+
+    private fun nameError(name: String, position: Position, existing: Statement) {
+        errors.err("name conflict '$name', also defined in ${existing.position.file} line ${existing.position.line}", position)
+    }
+
+    override fun visit(module: Module) {
+        blocks.clear()  // blocks may be redefined within a different module
+
+        super.visit(module)
+    }
+
+    override fun visit(block: Block) {
+        if(block.name in compTarget.machine.opcodeNames)
+            errors.err("can't use a cpu opcode name as a symbol: '${block.name}'", block.position)
+
+        val existing = blocks[block.name]
+        if(existing!=null)
+            nameError(block.name, block.position, existing)
+        else
+            blocks[block.name] = block
+
+        if(!block.isInLibrary) {
+            val libraries = program.modules.filter { it.isLibraryModule }
+            val libraryBlockNames = libraries.flatMap { it.statements.filterIsInstance<Block>().map { b -> b.name } }
+            if(block.name in libraryBlockNames)
+                errors.err("block is already defined in an included library module", block.position)
+        }
+
+        super.visit(block)
+    }
+
+    override fun visit(directive: Directive) {
+        if(directive.directive=="%target") {
+            val compatibleTarget = directive.args.single().name
+            if (compatibleTarget != compTarget.name)
+                errors.err("module's compilation target ($compatibleTarget) differs from active target (${compTarget.name})", directive.position)
+        }
+
+        super.visit(directive)
+    }
+
+    override fun visit(decl: VarDecl) {
+        decl.datatypeErrors.forEach { errors.err(it.message, it.position) }
+
+        if(decl.name in BuiltinFunctions)
+            errors.err("builtin function cannot be redefined", decl.position)
+
+        if(decl.name in compTarget.machine.opcodeNames)
+            errors.err("can't use a cpu opcode name as a symbol: '${decl.name}'", decl.position)
+
+        if(decl.datatype==DataType.STRUCT) {
+            if (decl.structHasBeenFlattened)
+                return super.visit(decl)    // don't do this multiple times
+
+            if (decl.struct == null) {
+                errors.err("undefined struct type", decl.position)
+                return super.visit(decl)
+            }
+
+            if (decl.struct!!.statements.any { (it as VarDecl).datatype !in NumericDatatypes })
+                return super.visit(decl)     // a non-numeric member, not supported. proper error is given by AstChecker later
+
+            if (decl.value is NumericLiteralValue) {
+                errors.err("you cannot initialize a struct using a single value", decl.position)
+                return super.visit(decl)
+            }
+
+            if (decl.value != null && decl.value !is ArrayLiteralValue) {
+                errors.err("initializing a struct requires array literal value", decl.value?.position ?: decl.position)
+                return super.visit(decl)
+            }
+        }
+
+        val existing = program.namespace.lookup(listOf(decl.name), decl)
+        if (existing != null && existing !== decl)
+            nameError(decl.name, decl.position, existing)
+
+        if(decl.definingBlock().name==decl.name)
+            nameError(decl.name, decl.position, decl.definingBlock())
+        if(decl.definingSubroutine()?.name==decl.name)
+            nameError(decl.name, decl.position, decl.definingSubroutine()!!)
+
+        super.visit(decl)
+    }
+
+    override fun visit(subroutine: Subroutine) {
+        if(subroutine.name in compTarget.machine.opcodeNames) {
+            errors.err("can't use a cpu opcode name as a symbol: '${subroutine.name}'", subroutine.position)
+        } else if(subroutine.name in BuiltinFunctions) {
+            // the builtin functions can't be redefined
+            errors.err("builtin function cannot be redefined", subroutine.position)
+        } else {
+            // already reported elsewhere:
+            // if (subroutine.parameters.any { it.name in BuiltinFunctions })
+            //    checkResult.add(NameError("builtin function name cannot be used as parameter", subroutine.position))
+
+            val existing = program.namespace.lookup(listOf(subroutine.name), subroutine)
+            if (existing != null && existing !== subroutine)
+                nameError(subroutine.name, subroutine.position, existing)
+
+            // check that there are no local variables, labels, or other subs that redefine the subroutine's parameters
+            val symbolsInSub = subroutine.allDefinedSymbols()
+            val namesInSub = symbolsInSub.map{ it.first }.toSet()
+            val paramNames = subroutine.parameters.map { it.name }.toSet()
+            val paramsToCheck = paramNames.intersect(namesInSub)
+            for(name in paramsToCheck) {
+                val labelOrVar = subroutine.getLabelOrVariable(name)
+                if(labelOrVar!=null && labelOrVar.position != subroutine.position)
+                    nameError(name, labelOrVar.position, subroutine)
+                val sub = subroutine.statements.firstOrNull { it is Subroutine && it.name==name}
+                if(sub!=null)
+                    nameError(name, subroutine.position, sub)
+                val block = program.allBlocks().firstOrNull { it.name==name }
+                if(block!=null)
+                    nameError(name, subroutine.position, block)
+            }
+
+            if(subroutine.isAsmSubroutine && subroutine.statements.any{it !is InlineAssembly}) {
+                errors.err("asmsub can only contain inline assembly (%asm)", subroutine.position)
+            }
+        }
+
+        super.visit(subroutine)
+    }
+
+    override fun visit(label: Label) {
+        if(label.name in compTarget.machine.opcodeNames)
+            errors.err("can't use a cpu opcode name as a symbol: '${label.name}'", label.position)
+
+        if(label.name in BuiltinFunctions) {
+            // the builtin functions can't be redefined
+            errors.err("builtin function cannot be redefined", label.position)
+        } else {
+            val existing = label.definingSubroutine()?.getAllLabels(label.name) ?: emptyList()
+            for(el in existing) {
+                if(el === label || el.name != label.name)
+                    continue
+                else {
+                    nameError(label.name, label.position, el)
+                    break
+                }
+            }
+        }
+
+        super.visit(label)
+    }
+
+    override fun visit(string: StringLiteralValue) {
+        if (string.value.length > 255)
+            errors.err("string literal length max is 255", string.position)
+
+        super.visit(string)
+    }
+
+    override fun visit(structDecl: StructDecl) {
+        for(member in structDecl.statements){
+            val decl = member as? VarDecl
+            if(decl!=null && decl.datatype !in NumericDatatypes)
+                errors.err("structs can only contain numerical types", decl.position)
+        }
+
+        super.visit(structDecl)
+    }
+}

compiler/src/prog8/compiler/astprocessing/AstVariousTransforms.kt

@@ -0,0 +1,111 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.StringLiteralValue
+import prog8.ast.statements.AnonymousScope
+import prog8.ast.statements.ParameterVarDecl
+import prog8.ast.statements.Subroutine
+import prog8.ast.statements.VarDecl
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+
+
+internal class AstVariousTransforms(private val program: Program) : AstWalker() {
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun before(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        // is it a struct variable? then define all its struct members as mangled names,
+        //    and include the original decl as well.
+        if(decl.datatype==DataType.STRUCT && !decl.structHasBeenFlattened) {
+            val decls = decl.flattenStructMembers()
+            decls.add(decl)
+            val result = AnonymousScope(decls, decl.position)
+            return listOf(IAstModification.ReplaceNode(
+                    decl, result, parent
+            ))
+        }
+
+        return noModifications
+    }
+
+    override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        // For non-kernal subroutines and non-asm parameters:
+        // inject subroutine params as local variables (if they're not there yet).
+        val symbolsInSub = subroutine.allDefinedSymbols()
+        val namesInSub = symbolsInSub.map{ it.first }.toSet()
+        if(subroutine.asmAddress==null) {
+            if(subroutine.asmParameterRegisters.isEmpty() && subroutine.parameters.isNotEmpty()) {
+                val vars = subroutine.statements.filterIsInstance<VarDecl>().map { it.name }.toSet()
+                if(!vars.containsAll(subroutine.parameters.map{it.name})) {
+                    return subroutine.parameters
+                            .filter { it.name !in namesInSub }
+                            .map {
+                                val vardecl = ParameterVarDecl(it.name, it.type, subroutine.position)
+                                IAstModification.InsertFirst(vardecl, subroutine)
+                            }
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        val leftStr = expr.left as? StringLiteralValue
+        val rightStr = expr.right as? StringLiteralValue
+        if(expr.operator == "+") {
+            val concatenatedString = concatString(expr)
+            if(concatenatedString!=null)
+                return listOf(IAstModification.ReplaceNode(expr, concatenatedString, parent))
+        }
+        else if(expr.operator == "*") {
+            if (leftStr!=null) {
+                val amount = expr.right.constValue(program)
+                if(amount!=null) {
+                    val string = leftStr.value.repeat(amount.number.toInt())
+                    val strval = StringLiteralValue(string, leftStr.altEncoding, expr.position)
+                    return listOf(IAstModification.ReplaceNode(expr, strval, parent))
+                }
+            }
+            else if (rightStr!=null) {
+                val amount = expr.right.constValue(program)
+                if(amount!=null) {
+                    val string = rightStr.value.repeat(amount.number.toInt())
+                    val strval = StringLiteralValue(string, rightStr.altEncoding, expr.position)
+                    return listOf(IAstModification.ReplaceNode(expr, strval, parent))
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun concatString(expr: BinaryExpression): StringLiteralValue? {
+        val rightStrval = expr.right as? StringLiteralValue
+        val leftStrval = expr.left as? StringLiteralValue
+        return when {
+            expr.operator!="+" -> null
+            expr.left is BinaryExpression && rightStrval!=null -> {
+                val subStrVal = concatString(expr.left as BinaryExpression)
+                if(subStrVal==null)
+                    null
+                else
+                    StringLiteralValue("${subStrVal.value}${rightStrval.value}", subStrVal.altEncoding, rightStrval.position)
+            }
+            expr.right is BinaryExpression && leftStrval!=null -> {
+                val subStrVal = concatString(expr.right as BinaryExpression)
+                if(subStrVal==null)
+                    null
+                else
+                    StringLiteralValue("${leftStrval.value}${subStrVal.value}", subStrVal.altEncoding, leftStrval.position)
+            }
+            leftStrval!=null && rightStrval!=null -> {
+                StringLiteralValue("${leftStrval.value}${rightStrval.value}", leftStrval.altEncoding, leftStrval.position)
+            }
+            else -> null
+        }
+    }
+}

compiler/src/prog8/compiler/astprocessing/LiteralsToAutoVars.kt

@@ -0,0 +1,55 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.expressions.ArrayLiteralValue
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.StringLiteralValue
+import prog8.ast.statements.VarDecl
+import prog8.ast.statements.WhenChoice
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+
+
+internal class LiteralsToAutoVars(private val program: Program) : AstWalker() {
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun after(string: StringLiteralValue, parent: Node): Iterable<IAstModification> {
+        if(string.parent !is VarDecl && string.parent !is WhenChoice) {
+            // replace the literal string by a identifier reference to the interned string
+            val scopedName = program.internString(string)
+            val identifier = IdentifierReference(scopedName, string.position)
+            return listOf(IAstModification.ReplaceNode(string, identifier, parent))
+        }
+        return noModifications
+    }
+
+    override fun after(array: ArrayLiteralValue, parent: Node): Iterable<IAstModification> {
+        val vardecl = array.parent as? VarDecl
+        if(vardecl!=null) {
+            // adjust the datatype of the array (to an educated guess)
+            val arrayDt = array.type
+            if(!arrayDt.istype(vardecl.datatype)) {
+                val cast = array.cast(vardecl.datatype)
+                if (cast != null && cast !== array)
+                    return listOf(IAstModification.ReplaceNode(vardecl.value!!, cast, vardecl))
+            }
+        } else {
+            val arrayDt = array.guessDatatype(program)
+            if(arrayDt.isKnown) {
+                // this array literal is part of an expression, turn it into an identifier reference
+                val litval2 = array.cast(arrayDt.typeOrElse(DataType.STRUCT))
+                if(litval2!=null) {
+                    val vardecl2 = VarDecl.createAuto(litval2)
+                    val identifier = IdentifierReference(listOf(vardecl2.name), vardecl2.position)
+                    return listOf(
+                            IAstModification.ReplaceNode(array, identifier, parent),
+                            IAstModification.InsertFirst(vardecl2, array.definingScope())
+                    )
+                }
+            }
+        }
+        return noModifications
+    }
+}

compiler/src/prog8/compiler/astprocessing/ReflectionAstWalker.kt

@@ -0,0 +1,71 @@
+package prog8.compiler.astprocessing
+
+
+/*
+This is here for reference only, reflection based ast walking is very slow
+when compared to the more verbose visitor pattern interfaces.
+Too bad, because the code is very small
+*/
+
+
+//import prog8.ast.NoAstWalk
+//import prog8.ast.Node
+//import prog8.ast.Program
+//import prog8.ast.base.Position
+//import prog8.ast.expressions.BinaryExpression
+//import prog8.ast.expressions.NumericLiteralValue
+//import kotlin.reflect.KClass
+//import kotlin.reflect.KVisibility
+//import kotlin.reflect.full.declaredMemberProperties
+//import kotlin.reflect.full.isSubtypeOf
+//import kotlin.reflect.full.starProjectedType
+//
+//
+//class ReflectionAstWalker {
+//    private val nodeType = Node::class.starProjectedType
+//    private val collectionType = Collection::class.starProjectedType
+//
+//
+//    fun walk(node: Node, nesting: Int) {
+//        val nodetype: KClass<out Node> = node::class
+//        val indent = "  ".repeat(nesting)
+//        //println("$indent VISITING ${nodetype.simpleName}")
+//        val visibleAstMembers = nodetype.declaredMemberProperties.filter {
+//            it.visibility!=KVisibility.PRIVATE && !it.isLateinit &&
+//                    !(it.annotations.any{a->a is NoAstWalk})
+//        }
+//        for(prop in visibleAstMembers) {
+//            if(prop.returnType.isSubtypeOf(nodeType)) {
+//                // println("$indent +PROP: ${prop.name}")
+//                walk(prop.call(node) as Node, nesting + 1)
+//            }
+//            else if(prop.returnType.isSubtypeOf(collectionType)) {
+//                val elementType = prop.returnType.arguments.single().type
+//                if(elementType!=null && elementType.isSubtypeOf(nodeType)) {
+//                    val nodes = prop.call(node) as Collection<Node>
+//                    nodes.forEach { walk(it, nesting+1) }
+//                }
+//            }
+//        }
+//    }
+//    fun walk(program: Program) {
+//        for(module in program.modules) {
+//            println("---MODULE $module---")
+//            walk(module, 0)
+//        }
+//    }
+//}
+//
+//
+//fun main() {
+//    val ast = BinaryExpression(
+//            NumericLiteralValue.optimalInteger(100, Position.DUMMY),
+//            "+",
+//            NumericLiteralValue.optimalInteger(200, Position.DUMMY),
+//            Position.DUMMY
+//    )
+//
+//    val walker = ReflectionAstWalker()
+//    walker.walk(ast,0)
+//
+//}

compiler/src/prog8/compiler/astprocessing/StatementReorderer.kt

@@ -0,0 +1,399 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Module
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+
+
+internal class StatementReorderer(val program: Program, val errors: IErrorReporter) : AstWalker() {
+    // Reorders the statements in a way the compiler needs.
+    // - 'main' block must be the very first statement UNLESS it has an address set.
+    // - library blocks are put last.
+    // - blocks are ordered by address, where blocks without address are placed last.
+    // - in every block and module, most directives and vardecls are moved to the top. (not in subroutines!)
+    // - the 'start' subroutine is moved to the top.
+    // - (syntax desugaring) a vardecl with a non-const initializer value is split into a regular vardecl and an assignment statement.
+    // - (syntax desugaring) struct value assignment is expanded into several struct member assignments.
+    // - in-place assignments are reordered a bit so that they are mostly of the form A = A <operator> <rest>
+    // - sorts the choices in when statement.
+    // - insert AddressOf (&) expression where required (string params to a UWORD function param etc).
+
+    private val noModifications = emptyList<IAstModification>()
+    private val directivesToMove = setOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address", "%option")
+
+    override fun after(module: Module, parent: Node): Iterable<IAstModification> {
+        val (blocks, other) = module.statements.partition { it is Block }
+        module.statements = other.asSequence().plus(blocks.sortedBy { (it as Block).address ?: Int.MAX_VALUE }).toMutableList()
+
+        val mainBlock = module.statements.filterIsInstance<Block>().firstOrNull { it.name=="main" }
+        if(mainBlock!=null && mainBlock.address==null) {
+            module.statements.remove(mainBlock)
+            module.statements.add(0, mainBlock)
+        }
+
+        reorderVardeclsAndDirectives(module.statements)
+        return noModifications
+    }
+
+    private fun reorderVardeclsAndDirectives(statements: MutableList<Statement>) {
+        val varDecls = statements.filterIsInstance<VarDecl>()
+        statements.removeAll(varDecls)
+        statements.addAll(0, varDecls)
+
+        val directives = statements.filterIsInstance<Directive>().filter {it.directive in directivesToMove}
+        statements.removeAll(directives)
+        statements.addAll(0, directives)
+    }
+
+    override fun before(block: Block, parent: Node): Iterable<IAstModification> {
+        parent as Module
+        if(block.isInLibrary) {
+            return listOf(
+                    IAstModification.Remove(block, parent),
+                    IAstModification.InsertLast(block, parent)
+            )
+        }
+
+        reorderVardeclsAndDirectives(block.statements)
+        return noModifications
+    }
+
+    override fun before(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        if(subroutine.name=="start" && parent is Block) {
+            if(parent.statements.filterIsInstance<Subroutine>().first().name!="start") {
+                return listOf(
+                        IAstModification.Remove(subroutine, parent),
+                        IAstModification.InsertFirst(subroutine, parent)
+                )
+            }
+        }
+
+        val subs = subroutine.statements.filterIsInstance<Subroutine>()
+        if(subs.isNotEmpty()) {
+            // all subroutines defined within this subroutine are moved to the end
+            return subs.map { IAstModification.Remove(it, subroutine) } +
+                    subs.map { IAstModification.InsertLast(it, subroutine) }
+        }
+
+        return noModifications
+    }
+
+    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
+
+        val arrayVar = arrayIndexedExpression.arrayvar.targetVarDecl(program)
+        if(arrayVar!=null && arrayVar.datatype ==  DataType.UWORD) {
+            // rewrite   pointervar[index]  into  @(pointervar+index)
+            val indexer = arrayIndexedExpression.indexer
+            val add = BinaryExpression(arrayIndexedExpression.arrayvar, "+", indexer.indexExpr, arrayIndexedExpression.position)
+            return if(parent is AssignTarget) {
+                // we're part of the target of an assignment, we have to actually change the assign target itself
+                val memwrite = DirectMemoryWrite(add, arrayIndexedExpression.position)
+                val newtarget = AssignTarget(null, null, memwrite, arrayIndexedExpression.position)
+                listOf(IAstModification.ReplaceNode(parent, newtarget, parent.parent))
+            } else {
+                val memread = DirectMemoryRead(add, arrayIndexedExpression.position)
+                listOf(IAstModification.ReplaceNode(arrayIndexedExpression, memread, parent))
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        // when using a simple bit shift and assigning it to a variable of a different type,
+        // try to make the bit shifting 'wide enough' to fall into the variable's type.
+        // with this, for instance, uword x = 1 << 10  will result in 1024 rather than 0 (the ubyte result).
+        if(expr.operator=="<<" || expr.operator==">>") {
+            val leftDt = expr.left.inferType(program)
+            when (parent) {
+                is Assignment -> {
+                    val targetDt = parent.target.inferType(program)
+                    if(leftDt != targetDt) {
+                        val cast = TypecastExpression(expr.left, targetDt.typeOrElse(DataType.STRUCT), true, parent.position)
+                        return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                    }
+                }
+                is VarDecl -> {
+                    if(!leftDt.istype(parent.datatype)) {
+                        val cast = TypecastExpression(expr.left, parent.datatype, true, parent.position)
+                        return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                    }
+                }
+                is IFunctionCall -> {
+                    val argnum = parent.args.indexOf(expr)
+                    when (val callee = parent.target.targetStatement(program)) {
+                        is Subroutine -> {
+                            val paramType = callee.parameters[argnum].type
+                            if(leftDt isAssignableTo paramType) {
+                                val cast = TypecastExpression(expr.left, paramType, true, parent.position)
+                                return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                            }
+                        }
+                        is BuiltinFunctionStatementPlaceholder -> {
+                            val func = BuiltinFunctions.getValue(callee.name)
+                            val paramTypes = func.parameters[argnum].possibleDatatypes
+                            for(type in paramTypes) {
+                                if(leftDt isAssignableTo type) {
+                                    val cast = TypecastExpression(expr.left, type, true, parent.position)
+                                    return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                                }
+                            }
+                        }
+                        else -> throw FatalAstException("weird callee")
+                    }
+                }
+                else -> return noModifications
+            }
+        }
+        else if(expr.operator in logicalOperators) {
+            // make sure that logical expressions like "var and other-logical-expression
+            // is rewritten as "var!=0 and other-logical-expression", to avoid bitwise boolean and
+            // generating the wrong results later
+
+            fun wrapped(expr: Expression): Expression =
+                BinaryExpression(expr, "!=", NumericLiteralValue(DataType.UBYTE, 0, expr.position), expr.position)
+
+            fun isLogicalExpr(expr: Expression?): Boolean {
+                if(expr is BinaryExpression && expr.operator in (logicalOperators + comparisonOperators))
+                    return true
+                if(expr is PrefixExpression && expr.operator in logicalOperators)
+                    return true
+                return false
+            }
+
+            return if(isLogicalExpr(expr.left)) {
+                if(isLogicalExpr(expr.right))
+                    noModifications
+                else
+                    listOf(IAstModification.ReplaceNode(expr.right, wrapped(expr.right), expr))
+            } else {
+                if(isLogicalExpr(expr.right))
+                    listOf(IAstModification.ReplaceNode(expr.left, wrapped(expr.left), expr))
+                else {
+                    listOf(
+                        IAstModification.ReplaceNode(expr.left, wrapped(expr.left), expr),
+                        IAstModification.ReplaceNode(expr.right, wrapped(expr.right), expr)
+                    )
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(whenStatement: WhenStatement, parent: Node): Iterable<IAstModification> {
+        val choices = whenStatement.choiceValues(program).sortedBy {
+            it.first?.first() ?: Int.MAX_VALUE
+        }
+        whenStatement.choices.clear()
+        choices.mapTo(whenStatement.choices) { it.second }
+        return noModifications
+    }
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        val declValue = decl.value
+        if(declValue!=null && decl.type== VarDeclType.VAR && decl.datatype in NumericDatatypes) {
+            val declConstValue = declValue.constValue(program)
+            if(declConstValue==null) {
+                // move the vardecl (without value) to the scope and replace this with a regular assignment
+                // Unless we're dealing with a floating point variable because that will actually make things less efficient at the moment (because floats are mostly calcualated via the stack)
+                if(decl.datatype!=DataType.FLOAT) {
+                    decl.value = null
+                    decl.allowInitializeWithZero = false
+                    val target = AssignTarget(IdentifierReference(listOf(decl.name), decl.position), null, null, decl.position)
+                    val assign = Assignment(target, declValue, decl.position)
+                    return listOf(
+                            IAstModification.ReplaceNode(decl, assign, parent),
+                            IAstModification.InsertFirst(decl, decl.definingScope())
+                    )
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun before(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        val valueType = assignment.value.inferType(program)
+        val targetType = assignment.target.inferType(program)
+
+        if(targetType.istype(DataType.STRUCT) && (valueType.istype(DataType.STRUCT) || valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes )) {
+            if (assignment.value is ArrayLiteralValue) {
+                errors.err("cannot assign non-const array value, use separate assignment per field", assignment.position)
+            } else {
+                return copyStructValue(assignment)
+            }
+        }
+
+        if(targetType.typeOrElse(DataType.STRUCT) in ArrayDatatypes && valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes ) {
+            if (assignment.value is ArrayLiteralValue) {
+                errors.err("cannot assign non-const array value, use separate assignment per element", assignment.position)
+            } else {
+                return copyArrayValue(assignment)
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        // rewrite in-place assignment expressions a bit so that the assignment target usually is the leftmost operand
+        val binExpr = assignment.value as? BinaryExpression
+        if(binExpr!=null) {
+            if(binExpr.left isSameAs assignment.target) {
+                // A = A <operator> 5, unchanged
+                return noModifications
+            }
+
+            if(binExpr.operator in associativeOperators) {
+                if (binExpr.right isSameAs assignment.target) {
+                    // A = v <associative-operator> A  ==>  A = A <associative-operator> v
+                    return listOf(IAstModification.SwapOperands(binExpr))
+                }
+
+                val leftBinExpr = binExpr.left as? BinaryExpression
+                if(leftBinExpr?.operator == binExpr.operator) {
+                    return if(leftBinExpr.left isSameAs assignment.target) {
+                        // A = (A <associative-operator> x) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(leftBinExpr.right, binExpr.operator, binExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(leftBinExpr.left, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    } else {
+                        // A = (x <associative-operator> A) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(leftBinExpr.left, binExpr.operator, binExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(leftBinExpr.right, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    }
+                }
+                val rightBinExpr = binExpr.right as? BinaryExpression
+                if(rightBinExpr?.operator == binExpr.operator) {
+                    return if(rightBinExpr.left isSameAs assignment.target) {
+                        // A = x <associative-operator> (A <same-operator> y) ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(rightBinExpr.left, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    } else {
+                        // A = x <associative-operator> (y <same-operator> A) ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.left, binExpr.position)
+                        val newValue = BinaryExpression(rightBinExpr.right, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    }
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun copyArrayValue(assign: Assignment): List<IAstModification> {
+        val identifier = assign.target.identifier!!
+        val targetVar = identifier.targetVarDecl(program)!!
+
+        if(targetVar.arraysize==null)
+            errors.err("array has no defined size", assign.position)
+
+        if(assign.value !is IdentifierReference) {
+            errors.err("invalid array value to assign to other array", assign.value.position)
+            return noModifications
+        }
+        val sourceIdent = assign.value as IdentifierReference
+        val sourceVar = sourceIdent.targetVarDecl(program)!!
+        if(!sourceVar.isArray) {
+            errors.err("value must be an array", sourceIdent.position)
+        } else {
+            if (sourceVar.arraysize!!.constIndex() != targetVar.arraysize!!.constIndex())
+                errors.err("element count mismatch", assign.position)
+            if (sourceVar.datatype != targetVar.datatype)
+                errors.err("element type mismatch", assign.position)
+        }
+
+        if(!errors.isEmpty())
+            return noModifications
+
+        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), assign.position),
+            mutableListOf(
+                AddressOf(sourceIdent, assign.position),
+                AddressOf(identifier, assign.position),
+                NumericLiteralValue.optimalInteger(targetVar.arraysize!!.constIndex()!!, assign.position)
+            ),
+            true,
+            assign.position
+        )
+        return listOf(IAstModification.ReplaceNode(assign, memcopy, assign.parent))
+    }
+
+    private fun copyStructValue(structAssignment: Assignment): List<IAstModification> {
+        val identifier = structAssignment.target.identifier!!
+        val targetVar = identifier.targetVarDecl(program)!!
+        val struct = targetVar.struct!!
+        when (structAssignment.value) {
+            is IdentifierReference -> {
+                val sourceVar = (structAssignment.value as IdentifierReference).targetVarDecl(program)!!
+                val memsize = struct.memsize(program.memsizer)
+                when {
+                    sourceVar.struct!=null -> {
+                        // struct memberwise copy
+                        val sourceStruct = sourceVar.struct!!
+                        if(sourceStruct!==targetVar.struct) {
+                            errors.err("struct type mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(struct.statements.size!=sourceStruct.statements.size) {
+                            errors.err("struct element count mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(memsize!=sourceStruct.memsize(program.memsizer)) {
+                            errors.err("memory size mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), structAssignment.position),
+                            mutableListOf(
+                                AddressOf(structAssignment.value as IdentifierReference, structAssignment.position),
+                                AddressOf(identifier, structAssignment.position),
+                                NumericLiteralValue.optimalInteger(memsize, structAssignment.position)
+                            ),
+                            true,
+                            structAssignment.position
+                        )
+                        return listOf(IAstModification.ReplaceNode(structAssignment, memcopy, structAssignment.parent))
+                    }
+                    sourceVar.isArray -> {
+                        val array = sourceVar.value as ArrayLiteralValue
+                        if(struct.statements.size!=array.value.size) {
+                            errors.err("struct element count mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(memsize!=array.memsize(program.memsizer)) {
+                            errors.err("memory size mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), structAssignment.position),
+                            mutableListOf(
+                                AddressOf(structAssignment.value as IdentifierReference, structAssignment.position),
+                                AddressOf(identifier, structAssignment.position),
+                                NumericLiteralValue.optimalInteger(memsize, structAssignment.position)
+                            ),
+                            true,
+                            structAssignment.position
+                        )
+                        return listOf(IAstModification.ReplaceNode(structAssignment, memcopy, structAssignment.parent))
+                    }
+                    else -> {
+                        throw FatalAstException("can only assign arrays or structs to structs")
+                    }
+                }
+            }
+            is ArrayLiteralValue -> {
+                throw IllegalArgumentException("not going to do a structLv assignment here")
+            }
+            else -> throw FatalAstException("strange struct value")
+        }
+    }
+
+}

compiler/src/prog8/compiler/astprocessing/TypecastsAdder.kt

@@ -0,0 +1,234 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+
+
+class TypecastsAdder(val program: Program, val errors: IErrorReporter) : AstWalker() {
+    /*
+     * Make sure any value assignments get the proper type casts if needed to cast them into the target variable's type.
+     * (this includes function call arguments)
+     */
+
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        val declValue = decl.value
+        if(decl.type==VarDeclType.VAR && declValue!=null && decl.struct==null) {
+            val valueDt = declValue.inferType(program)
+            if(!valueDt.istype(decl.datatype)) {
+
+                // don't add a typecast on an array initializer value
+                if(valueDt.typeOrElse(DataType.STRUCT) in IntegerDatatypes && decl.datatype in ArrayDatatypes)
+                    return noModifications
+
+                return listOf(IAstModification.ReplaceNode(
+                        declValue,
+                        TypecastExpression(declValue, decl.datatype, true, declValue.position),
+                        decl
+                ))
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        val leftDt = expr.left.inferType(program)
+        val rightDt = expr.right.inferType(program)
+        if(leftDt.isKnown && rightDt.isKnown && leftDt!=rightDt) {
+            // determine common datatype and add typecast as required to make left and right equal types
+            val (commonDt, toFix) = BinaryExpression.commonDatatype(leftDt.typeOrElse(DataType.STRUCT), rightDt.typeOrElse(DataType.STRUCT), expr.left, expr.right)
+            if(toFix!=null) {
+                return when {
+                    toFix===expr.left -> listOf(IAstModification.ReplaceNode(
+                            expr.left, TypecastExpression(expr.left, commonDt, true, expr.left.position), expr))
+                    toFix===expr.right -> listOf(IAstModification.ReplaceNode(
+                            expr.right, TypecastExpression(expr.right, commonDt, true, expr.right.position), expr))
+                    else -> throw FatalAstException("confused binary expression side")
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        // see if a typecast is needed to convert the value's type into the proper target type
+        val valueItype = assignment.value.inferType(program)
+        val targetItype = assignment.target.inferType(program)
+        if(targetItype.isKnown && valueItype.isKnown) {
+            val targettype = targetItype.typeOrElse(DataType.STRUCT)
+            val valuetype = valueItype.typeOrElse(DataType.STRUCT)
+            if (valuetype != targettype) {
+                if (valuetype isAssignableTo targettype) {
+                    if(valuetype in IterableDatatypes && targettype==DataType.UWORD)
+                        // special case, don't typecast STR/arrays to UWORD, we support those assignments "directly"
+                        return noModifications
+                    return listOf(IAstModification.ReplaceNode(
+                            assignment.value,
+                            TypecastExpression(assignment.value, targettype, true, assignment.value.position),
+                            assignment))
+                } else {
+                    fun castLiteral(cvalue: NumericLiteralValue): List<IAstModification.ReplaceNode> {
+                        val cast = cvalue.cast(targettype)
+                        return if(cast.isValid)
+                            listOf(IAstModification.ReplaceNode(cvalue, cast.valueOrZero(), cvalue.parent))
+                        else
+                            emptyList()
+                    }
+                    val cvalue = assignment.value.constValue(program)
+                    if(cvalue!=null) {
+                        val number = cvalue.number.toDouble()
+                        // more complex comparisons if the type is different, but the constant value is compatible
+                        if (valuetype == DataType.BYTE && targettype == DataType.UBYTE) {
+                            if(number>0)
+                                return castLiteral(cvalue)
+                        } else if (valuetype == DataType.WORD && targettype == DataType.UWORD) {
+                            if(number>0)
+                                return castLiteral(cvalue)
+                        } else if (valuetype == DataType.UBYTE && targettype == DataType.BYTE) {
+                            if(number<0x80)
+                                return castLiteral(cvalue)
+                        } else if (valuetype == DataType.UWORD && targettype == DataType.WORD) {
+                            if(number<0x8000)
+                                return castLiteral(cvalue)
+                        }
+                    }
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        return afterFunctionCallArgs(functionCallStatement)
+    }
+
+    override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        return afterFunctionCallArgs(functionCall)
+    }
+
+    private fun afterFunctionCallArgs(call: IFunctionCall): Iterable<IAstModification> {
+        // see if a typecast is needed to convert the arguments into the required parameter's type
+        val modifications = mutableListOf<IAstModification>()
+
+        when(val sub = call.target.targetStatement(program)) {
+            is Subroutine -> {
+                sub.parameters.zip(call.args).forEachIndexed { index, pair ->
+                    val argItype = pair.second.inferType(program)
+                    if(argItype.isKnown) {
+                        val argtype = argItype.typeOrElse(DataType.STRUCT)
+                        val requiredType = pair.first.type
+                        if (requiredType != argtype) {
+                            if (argtype isAssignableTo requiredType) {
+                                modifications += IAstModification.ReplaceNode(
+                                        call.args[index],
+                                        TypecastExpression(pair.second, requiredType, true, pair.second.position),
+                                        call as Node)
+                            } else if(requiredType == DataType.UWORD && argtype in PassByReferenceDatatypes) {
+                                // we allow STR/ARRAY values in place of UWORD parameters. Take their address instead.
+                                if(pair.second is IdentifierReference) {
+                                    modifications += IAstModification.ReplaceNode(
+                                            call.args[index],
+                                            AddressOf(pair.second as IdentifierReference, pair.second.position),
+                                            call as Node)
+                                }
+                            } else if(pair.second is NumericLiteralValue) {
+                                val cast = (pair.second as NumericLiteralValue).cast(requiredType)
+                                if(cast.isValid)
+                                    modifications += IAstModification.ReplaceNode(
+                                            call.args[index],
+                                            cast.valueOrZero(),
+                                            call as Node)
+                            }
+                        }
+                    }
+                }
+            }
+            is BuiltinFunctionStatementPlaceholder -> {
+                val func = BuiltinFunctions.getValue(sub.name)
+                func.parameters.zip(call.args).forEachIndexed { index, pair ->
+                    val argItype = pair.second.inferType(program)
+                    if (argItype.isKnown) {
+                        val argtype = argItype.typeOrElse(DataType.STRUCT)
+                        if (pair.first.possibleDatatypes.all { argtype != it }) {
+                            for (possibleType in pair.first.possibleDatatypes) {
+                                if (argtype isAssignableTo possibleType) {
+                                    modifications += IAstModification.ReplaceNode(
+                                            call.args[index],
+                                            TypecastExpression(pair.second, possibleType, true, pair.second.position),
+                                            call as Node)
+                                    break
+                                }
+                            }
+                        }
+                    }
+                }
+            }
+            else -> { }
+        }
+
+        return modifications
+    }
+
+    override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
+        // warn about any implicit type casts to Float, because that may not be intended
+        if(typecast.implicit && typecast.type in setOf(DataType.FLOAT, DataType.ARRAY_F)) {
+            errors.warn("integer implicitly converted to float. Suggestion: use float literals, add an explicit cast, or revert to integer arithmetic", typecast.position)
+        }
+        return noModifications
+    }
+
+    override fun after(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
+        // make sure the memory address is an uword
+        val dt = memread.addressExpression.inferType(program)
+        if(dt.isKnown && dt.typeOrElse(DataType.UWORD)!=DataType.UWORD) {
+            val typecast = (memread.addressExpression as? NumericLiteralValue)?.cast(DataType.UWORD)?.valueOrZero()
+                    ?: TypecastExpression(memread.addressExpression, DataType.UWORD, true, memread.addressExpression.position)
+            return listOf(IAstModification.ReplaceNode(memread.addressExpression, typecast, memread))
+        }
+        return noModifications
+    }
+
+    override fun after(memwrite: DirectMemoryWrite, parent: Node): Iterable<IAstModification> {
+        // make sure the memory address is an uword
+        val dt = memwrite.addressExpression.inferType(program)
+        if(dt.isKnown && dt.typeOrElse(DataType.UWORD)!=DataType.UWORD) {
+            val typecast = (memwrite.addressExpression as? NumericLiteralValue)?.cast(DataType.UWORD)?.valueOrZero()
+                    ?: TypecastExpression(memwrite.addressExpression, DataType.UWORD, true, memwrite.addressExpression.position)
+            return listOf(IAstModification.ReplaceNode(memwrite.addressExpression, typecast, memwrite))
+        }
+        return noModifications
+    }
+
+    override fun after(returnStmt: Return, parent: Node): Iterable<IAstModification> {
+        // add a typecast to the return type if it doesn't match the subroutine's signature
+        val returnValue = returnStmt.value
+        if(returnValue!=null) {
+            val subroutine = returnStmt.definingSubroutine()!!
+            if(subroutine.returntypes.size==1) {
+                val subReturnType = subroutine.returntypes.first()
+                if (returnValue.inferType(program).istype(subReturnType))
+                    return noModifications
+                if (returnValue is NumericLiteralValue) {
+                    val cast = returnValue.cast(subroutine.returntypes.single())
+                    if(cast.isValid)
+                        returnStmt.value = cast.valueOrZero()
+                } else {
+                    return listOf(IAstModification.ReplaceNode(
+                            returnValue,
+                            TypecastExpression(returnValue, subReturnType, true, returnValue.position),
+                            returnStmt))
+                }
+            }
+        }
+        return noModifications
+    }
+}

compiler/src/prog8/compiler/astprocessing/VariousCleanups.kt

@@ -0,0 +1,73 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.INameScope
+import prog8.ast.Node
+import prog8.ast.base.Position
+import prog8.ast.expressions.DirectMemoryRead
+import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.expressions.TypecastExpression
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+
+
+internal class VariousCleanups: AstWalker() {
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun before(nopStatement: NopStatement, parent: Node): Iterable<IAstModification> {
+        return listOf(IAstModification.Remove(nopStatement, parent as INameScope))
+    }
+
+    override fun before(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
+        return if(parent is INameScope)
+            listOf(ScopeFlatten(scope, parent as INameScope))
+        else
+            noModifications
+    }
+
+    class ScopeFlatten(val scope: AnonymousScope, val into: INameScope) : IAstModification {
+        override fun perform() {
+            val idx = into.statements.indexOf(scope)
+            if(idx>=0) {
+                into.statements.addAll(idx+1, scope.statements)
+                into.statements.remove(scope)
+            }
+        }
+    }
+
+    override fun before(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
+        if(typecast.expression is NumericLiteralValue) {
+            val value = (typecast.expression as NumericLiteralValue).cast(typecast.type)
+            if(value.isValid)
+                return listOf(IAstModification.ReplaceNode(typecast, value.valueOrZero(), parent))
+        }
+
+        return noModifications
+    }
+
+    override fun before(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        return before(functionCallStatement as IFunctionCall, parent, functionCallStatement.position)
+    }
+
+    override fun before(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        return before(functionCall as IFunctionCall, parent, functionCall.position)
+    }
+
+    private fun before(functionCall: IFunctionCall, parent: Node, position: Position): Iterable<IAstModification> {
+        if(functionCall.target.nameInSource==listOf("peek")) {
+            // peek(a) is synonymous with @(a)
+            val memread = DirectMemoryRead(functionCall.args.single(), position)
+            return listOf(IAstModification.ReplaceNode(functionCall as Node, memread, parent))
+        }
+        if(functionCall.target.nameInSource==listOf("poke")) {
+            // poke(a, v) is synonymous with @(a) = v
+            val tgt = AssignTarget(null, null, DirectMemoryWrite(functionCall.args[0], position), position)
+            val assign = Assignment(tgt, functionCall.args[1], position)
+            return listOf(IAstModification.ReplaceNode(functionCall as Node, assign, parent))
+        }
+        return noModifications
+    }
+
+}

compiler/src/prog8/compiler/astprocessing/VerifyFunctionArgTypes.kt

@@ -0,0 +1,95 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.expressions.Expression
+import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.TypecastExpression
+import prog8.ast.statements.*
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.CompilerException
+import prog8.compiler.functions.BuiltinFunctions
+
+class VerifyFunctionArgTypes(val program: Program) : IAstVisitor {
+
+    override fun visit(functionCall: FunctionCall) {
+        val error = checkTypes(functionCall as IFunctionCall, program)
+        if(error!=null)
+            throw CompilerException(error)
+    }
+
+    override fun visit(functionCallStatement: FunctionCallStatement) {
+        val error = checkTypes(functionCallStatement as IFunctionCall, program)
+        if (error!=null)
+            throw CompilerException(error)
+    }
+
+    companion object {
+
+        private fun argTypeCompatible(argDt: DataType, paramDt: DataType): Boolean {
+            if(argDt==paramDt)
+                return true
+
+            // there are some exceptions that are considered compatible, such as STR <> UWORD
+            if(argDt==DataType.STR && paramDt==DataType.UWORD ||
+                    argDt==DataType.UWORD && paramDt==DataType.STR)
+                return true
+
+            return false
+        }
+
+        fun checkTypes(call: IFunctionCall, program: Program): String? {
+            val argITypes = call.args.map { it.inferType(program) }
+            val firstUnknownDt = argITypes.indexOfFirst { it.isUnknown }
+            if(firstUnknownDt>=0)
+                return "argument ${firstUnknownDt+1} invalid argument type"
+            val argtypes = argITypes.map { it.typeOrElse(DataType.STRUCT) }
+            val target = call.target.targetStatement(program)
+            if (target is Subroutine) {
+                if(call.args.size != target.parameters.size)
+                    return "invalid number of arguments"
+                val paramtypes = target.parameters.map { it.type }
+                val mismatch = argtypes.zip(paramtypes).indexOfFirst { !argTypeCompatible(it.first, it.second) }
+                if(mismatch>=0) {
+                    val actual = argtypes[mismatch].toString()
+                    val expected = paramtypes[mismatch].toString()
+                    return "argument ${mismatch + 1} type mismatch, was: $actual expected: $expected"
+                }
+                if(target.isAsmSubroutine) {
+                    if(target.asmReturnvaluesRegisters.size>1) {
+                        // multiple return values will NOT work inside an expression.
+                        // they MIGHT work in a regular assignment or just a function call statement.
+                        val parent = if(call is Statement) call.parent else if(call is Expression) call.parent else null
+                        if (call !is FunctionCallStatement) {
+                            val checkParent =
+                                if(parent is TypecastExpression)
+                                    parent.parent
+                                else
+                                    parent
+                            if (checkParent !is Assignment && checkParent !is VarDecl) {
+                                return "can't use subroutine call that returns multiple return values here"
+                            }
+                        }
+                    }
+                }
+            }
+            else if (target is BuiltinFunctionStatementPlaceholder) {
+                val func = BuiltinFunctions.getValue(target.name)
+                if(call.args.size != func.parameters.size)
+                    return "invalid number of arguments"
+                val paramtypes = func.parameters.map { it.possibleDatatypes }
+                argtypes.zip(paramtypes).forEachIndexed { index, pair ->
+                    val anyCompatible = pair.second.any { argTypeCompatible(pair.first, it) }
+                    if (!anyCompatible) {
+                        val actual = pair.first.toString()
+                        val expected = pair.second.toString()
+                        return "argument ${index + 1} type mismatch, was: $actual expected: $expected"
+                    }
+                }
+            }
+
+            return null
+        }
+    }
+}

compiler/src/prog8/compiler/functions/BuiltinFunctions.kt

@@ -0,0 +1,489 @@
+package prog8.compiler.functions
+
+import prog8.ast.IMemSizer
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.StructDecl
+import prog8.ast.statements.VarDecl
+import prog8.compiler.CompilerException
+import kotlin.math.*
+
+
+class FParam(val name: String, val possibleDatatypes: Set<DataType>)
+
+
+typealias ConstExpressionCaller = (args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer) -> NumericLiteralValue
+
+
+class ReturnConvention(val dt: DataType, val reg: RegisterOrPair?, val floatFac1: Boolean)
+class ParamConvention(val dt: DataType, val reg: RegisterOrPair?, val variable: Boolean)
+class CallConvention(val params: List<ParamConvention>, val returns: ReturnConvention) {
+    override fun toString(): String {
+        val paramConvs =  params.mapIndexed { index, it ->
+            when {
+                it.reg!=null -> "$index:${it.reg}"
+                it.variable -> "$index:variable"
+                else -> "$index:???"
+            }
+        }
+        val returnConv =
+                when {
+                    returns.reg!=null -> returns.reg.toString()
+                    returns.floatFac1 -> "floatFAC1"
+                    else -> "<no returnvalue>"
+                }
+        return "CallConvention[" + paramConvs.joinToString() + " ; returns: $returnConv]"
+    }
+}
+
+
+class FSignature(val name: String,
+                 val pure: Boolean,      // does it have side effects?
+                 val parameters: List<FParam>,
+                 val known_returntype: DataType?,     // specify return type if fixed, otherwise null if it depends on the arguments
+                 val constExpressionFunc: ConstExpressionCaller? = null) {
+
+    fun callConvention(actualParamTypes: List<DataType>): CallConvention {
+        val returns = when(known_returntype) {
+            DataType.UBYTE, DataType.BYTE -> ReturnConvention(known_returntype, RegisterOrPair.A, false)
+            DataType.UWORD, DataType.WORD -> ReturnConvention(known_returntype, RegisterOrPair.AY, false)
+            DataType.FLOAT -> ReturnConvention(known_returntype, null, true)
+            in PassByReferenceDatatypes -> ReturnConvention(known_returntype!!, RegisterOrPair.AY, false)
+            else -> {
+                val paramType = actualParamTypes.first()
+                if(pure)
+                    // return type depends on arg type
+                    when(paramType) {
+                        DataType.UBYTE, DataType.BYTE -> ReturnConvention(paramType, RegisterOrPair.A, false)
+                        DataType.UWORD, DataType.WORD -> ReturnConvention(paramType, RegisterOrPair.AY, false)
+                        DataType.FLOAT -> ReturnConvention(paramType, null, true)
+                        in PassByReferenceDatatypes -> ReturnConvention(paramType, RegisterOrPair.AY, false)
+                        else -> ReturnConvention(paramType, null, false)
+                    }
+                else {
+                    ReturnConvention(paramType, null, false)
+                }
+            }
+        }
+        return when {
+            actualParamTypes.isEmpty() -> CallConvention(emptyList(), returns)
+            actualParamTypes.size==1 -> {
+                // one parameter? via register/registerpair
+                val paramConv = when(val paramType = actualParamTypes[0]) {
+                    DataType.UBYTE, DataType.BYTE -> ParamConvention(paramType, RegisterOrPair.A, false)
+                    DataType.UWORD, DataType.WORD -> ParamConvention(paramType, RegisterOrPair.AY, false)
+                    DataType.FLOAT -> ParamConvention(paramType, RegisterOrPair.AY, false)
+                    in PassByReferenceDatatypes -> ParamConvention(paramType, RegisterOrPair.AY, false)
+                    else -> ParamConvention(paramType, null, false)
+                }
+                CallConvention(listOf(paramConv), returns)
+            }
+            else -> {
+                // multiple parameters? via variables
+                val paramConvs = actualParamTypes.map { ParamConvention(it, null, true) }
+                CallConvention(paramConvs, returns)
+            }
+        }
+    }
+}
+
+@Suppress("UNUSED_ANONYMOUS_PARAMETER")
+private val functionSignatures: List<FSignature> = listOf(
+        // this set of function have no return value and operate in-place:
+    FSignature("rol"         , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
+    FSignature("ror"         , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
+    FSignature("rol2"        , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
+    FSignature("ror2"        , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
+    FSignature("sort"        , false, listOf(FParam("array", ArrayDatatypes)), null),
+    FSignature("reverse"     , false, listOf(FParam("array", ArrayDatatypes)), null),
+    FSignature("cmp"         , false, listOf(FParam("value1", IntegerDatatypes), FParam("value2", NumericDatatypes)), null),
+        // these few have a return value depending on the argument(s):
+    FSignature("max"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinMax) },    // type depends on args
+    FSignature("min"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinMin) },    // type depends on args
+    FSignature("sum"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinSum) },      // type depends on args
+    FSignature("abs"         , true, listOf(FParam("value", NumericDatatypes)), null, ::builtinAbs),      // type depends on argument
+    FSignature("len"         , true, listOf(FParam("values", IterableDatatypes)), null, ::builtinLen),    // type is UBYTE or UWORD depending on actual length
+    FSignature("sizeof"      , true, listOf(FParam("object", DataType.values().toSet())), DataType.UBYTE, ::builtinSizeof),
+    FSignature("offsetof"    , true, listOf(FParam("object", DataType.values().toSet())), DataType.UBYTE, ::builtinOffsetof),
+        // normal functions follow:
+    FSignature("sgn"         , true, listOf(FParam("value", NumericDatatypes)), DataType.BYTE, ::builtinSgn ),
+    FSignature("sin"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::sin) },
+    FSignature("sin8"        , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.BYTE, ::builtinSin8 ),
+    FSignature("sin8u"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UBYTE, ::builtinSin8u ),
+    FSignature("sin16"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.WORD, ::builtinSin16 ),
+    FSignature("sin16u"      , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinSin16u ),
+    FSignature("cos"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::cos) },
+    FSignature("cos8"        , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.BYTE, ::builtinCos8 ),
+    FSignature("cos8u"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UBYTE, ::builtinCos8u ),
+    FSignature("cos16"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.WORD, ::builtinCos16 ),
+    FSignature("cos16u"      , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinCos16u ),
+    FSignature("tan"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::tan) },
+    FSignature("atan"        , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::atan) },
+    FSignature("ln"          , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::log) },
+    FSignature("log2"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, ::log2) },
+    FSignature("sqrt16"      , true, listOf(FParam("value", setOf(DataType.UWORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { sqrt(it.toDouble()).toInt() } },
+    FSignature("sqrt"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::sqrt) },
+    FSignature("rad"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::toRadians) },
+    FSignature("deg"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::toDegrees) },
+    FSignature("round"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::round) },
+    FSignature("floor"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::floor) },
+    FSignature("ceil"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::ceil) },
+    FSignature("any"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinAny) },
+    FSignature("all"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinAll) },
+    FSignature("lsb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { x: Int -> x and 255 } },
+    FSignature("msb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { x: Int -> x ushr 8 and 255} },
+    FSignature("mkword"      , true, listOf(FParam("msb", setOf(DataType.UBYTE)), FParam("lsb", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinMkword),
+    FSignature("peek"        , true, listOf(FParam("address", setOf(DataType.UWORD))), DataType.UBYTE),
+    FSignature("peekw"       , true, listOf(FParam("address", setOf(DataType.UWORD))), DataType.UWORD),
+    FSignature("poke"        , false, listOf(FParam("address", setOf(DataType.UWORD)), FParam("value", setOf(DataType.UBYTE))), null),
+    FSignature("pokew"       , false, listOf(FParam("address", setOf(DataType.UWORD)), FParam("value", setOf(DataType.UWORD))), null),
+    FSignature("fastrnd8"    , false, emptyList(), DataType.UBYTE),
+    FSignature("rnd"         , false, emptyList(), DataType.UBYTE),
+    FSignature("rndw"        , false, emptyList(), DataType.UWORD),
+    FSignature("rndf"        , false, emptyList(), DataType.FLOAT),
+    FSignature("memory"      , true, listOf(FParam("name", setOf(DataType.STR)), FParam("size", setOf(DataType.UWORD))), DataType.UWORD),
+    FSignature("swap"        , false, listOf(FParam("first", NumericDatatypes), FParam("second", NumericDatatypes)), null),
+
+)
+
+val BuiltinFunctions = functionSignatures.associateBy { it.name }
+
+
+fun builtinMax(array: List<Number>): Number = array.maxByOrNull { it.toDouble() }!!
+
+fun builtinMin(array: List<Number>): Number = array.minByOrNull { it.toDouble() }!!
+
+fun builtinSum(array: List<Number>): Number = array.sumByDouble { it.toDouble() }
+
+fun builtinAny(array: List<Number>): Number = if(array.any { it.toDouble()!=0.0 }) 1 else 0
+
+fun builtinAll(array: List<Number>): Number = if(array.all { it.toDouble()!=0.0 }) 1 else 0
+
+
+fun builtinFunctionReturnType(function: String, args: List<Expression>, program: Program): InferredTypes.InferredType {
+
+    fun datatypeFromIterableArg(arglist: Expression): DataType {
+        if(arglist is ArrayLiteralValue) {
+            val dt = arglist.value.map {it.inferType(program).typeOrElse(DataType.STRUCT)}.toSet()
+            if(dt.any { it !in NumericDatatypes }) {
+                throw FatalAstException("fuction $function only accepts array of numeric values")
+            }
+            if(DataType.FLOAT in dt) return DataType.FLOAT
+            if(DataType.UWORD in dt) return DataType.UWORD
+            if(DataType.WORD in dt) return DataType.WORD
+            if(DataType.BYTE in dt) return DataType.BYTE
+            return DataType.UBYTE
+        }
+        if(arglist is IdentifierReference) {
+            val idt = arglist.inferType(program)
+            if(!idt.isKnown)
+                throw FatalAstException("couldn't determine type of iterable $arglist")
+            return when(val dt = idt.typeOrElse(DataType.STRUCT)) {
+                DataType.STR, in NumericDatatypes -> dt
+                in ArrayDatatypes -> ArrayElementTypes.getValue(dt)
+                else -> throw FatalAstException("function '$function' requires one argument which is an iterable")
+            }
+        }
+        throw FatalAstException("function '$function' requires one argument which is an iterable")
+    }
+
+    val func = BuiltinFunctions.getValue(function)
+    if(func.known_returntype!=null)
+        return InferredTypes.knownFor(func.known_returntype)
+
+    // function has return values, but the return type depends on the arguments
+    return when (function) {
+        "abs" -> {
+            val dt = args.single().inferType(program)
+            return if(dt.typeOrElse(DataType.STRUCT) in NumericDatatypes)
+                dt
+            else
+                InferredTypes.InferredType.unknown()
+        }
+        "max", "min" -> {
+            when(val dt = datatypeFromIterableArg(args.single())) {
+                DataType.STR -> InferredTypes.knownFor(DataType.UBYTE)
+                in NumericDatatypes -> InferredTypes.knownFor(dt)
+                in ArrayDatatypes -> InferredTypes.knownFor(ArrayElementTypes.getValue(dt))
+                else -> InferredTypes.unknown()
+            }
+        }
+        "sum" -> {
+            when(datatypeFromIterableArg(args.single())) {
+                DataType.UBYTE, DataType.UWORD -> InferredTypes.knownFor(DataType.UWORD)
+                DataType.BYTE, DataType.WORD -> InferredTypes.knownFor(DataType.WORD)
+                DataType.FLOAT -> InferredTypes.knownFor(DataType.FLOAT)
+                DataType.ARRAY_UB, DataType.ARRAY_UW -> InferredTypes.knownFor(DataType.UWORD)
+                DataType.ARRAY_B, DataType.ARRAY_W -> InferredTypes.knownFor(DataType.WORD)
+                DataType.ARRAY_F -> InferredTypes.knownFor(DataType.FLOAT)
+                DataType.STR -> InferredTypes.knownFor(DataType.UWORD)
+                else -> InferredTypes.unknown()
+            }
+        }
+        "len" -> {
+            // a length can be >255 so in that case, the result is an UWORD instead of an UBYTE
+            // but to avoid a lot of code duplication we simply assume UWORD in all cases for now
+            return InferredTypes.knownFor(DataType.UWORD)
+        }
+        else -> return InferredTypes.unknown()
+    }
+}
+
+
+class NotConstArgumentException: AstException("not a const argument to a built-in function")
+class CannotEvaluateException(func:String, msg: String): FatalAstException("cannot evaluate built-in function $func: $msg")
+
+
+private fun oneDoubleArg(args: List<Expression>, position: Position, program: Program, function: (arg: Double)->Number): NumericLiteralValue {
+    if(args.size!=1)
+        throw SyntaxError("built-in function requires one floating point argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val float = constval.number.toDouble()
+    return numericLiteral(function(float), args[0].position)
+}
+
+private fun oneDoubleArgOutputWord(args: List<Expression>, position: Position, program: Program, function: (arg: Double)->Number): NumericLiteralValue {
+    if(args.size!=1)
+        throw SyntaxError("built-in function requires one floating point argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val float = constval.number.toDouble()
+    return NumericLiteralValue(DataType.WORD, function(float).toInt(), args[0].position)
+}
+
+private fun oneIntArgOutputInt(args: List<Expression>, position: Position, program: Program, function: (arg: Int)->Number): NumericLiteralValue {
+    if(args.size!=1)
+        throw SyntaxError("built-in function requires one integer argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    if(constval.type != DataType.UBYTE && constval.type!= DataType.UWORD)
+        throw SyntaxError("built-in function requires one integer argument", position)
+
+    val integer = constval.number.toInt()
+    return numericLiteral(function(integer).toInt(), args[0].position)
+}
+
+private fun collectionArg(args: List<Expression>, position: Position, program: Program, function: (arg: List<Number>)->Number): NumericLiteralValue {
+    if(args.size!=1)
+        throw SyntaxError("builtin function requires one non-scalar argument", position)
+
+    val array= args[0] as? ArrayLiteralValue ?: throw NotConstArgumentException()
+    val constElements = array.value.map{it.constValue(program)?.number}
+    if(constElements.contains(null))
+        throw NotConstArgumentException()
+
+    return NumericLiteralValue.optimalNumeric(function(constElements.mapNotNull { it }), args[0].position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinAbs(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    // 1 arg, type = float or int, result type= isSameAs as argument type
+    if(args.size!=1)
+        throw SyntaxError("abs requires one numeric argument", position)
+
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    return when (constval.type) {
+        in IntegerDatatypes -> numericLiteral(abs(constval.number.toInt()), args[0].position)
+        DataType.FLOAT -> numericLiteral(abs(constval.number.toDouble()), args[0].position)
+        else -> throw SyntaxError("abs requires one numeric argument", position)
+    }
+}
+
+private fun builtinOffsetof(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    // 1 arg, type = anything, result type = ubyte
+    if(args.size!=1)
+        throw SyntaxError("offsetof requires one argument", position)
+    val idref = args[0] as? IdentifierReference
+        ?: throw SyntaxError("offsetof argument should be an identifier", position)
+
+    val vardecl = idref.targetVarDecl(program)!!
+    val struct = vardecl.struct
+    if (struct == null || vardecl.datatype == DataType.STRUCT)
+        throw SyntaxError("offsetof can only be used on struct members", position)
+
+    val membername = idref.nameInSource.last()
+    var offset = 0
+    for(member in struct.statements) {
+        if((member as VarDecl).name == membername)
+            return NumericLiteralValue(DataType.UBYTE, offset, position)
+        offset += memsizer.memorySize(member.datatype)
+    }
+    throw SyntaxError("undefined struct member", position)
+}
+
+private fun builtinSizeof(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    // 1 arg, type = anything, result type = ubyte
+    if(args.size!=1)
+        throw SyntaxError("sizeof requires one argument", position)
+    if(args[0] !is IdentifierReference)
+        throw SyntaxError("sizeof argument should be an identifier", position)
+
+    val dt = args[0].inferType(program)
+    if(dt.isKnown) {
+        val target = (args[0] as IdentifierReference).targetStatement(program)
+                ?: throw CannotEvaluateException("sizeof", "no target")
+
+        fun structSize(target: StructDecl) = NumericLiteralValue(DataType.UBYTE, target.memsize(memsizer), position)
+
+        return when {
+            dt.typeOrElse(DataType.STRUCT) in ArrayDatatypes -> {
+                val length = (target as VarDecl).arraysize!!.constIndex() ?: throw CannotEvaluateException("sizeof", "unknown array size")
+                val elementDt = ArrayElementTypes.getValue(dt.typeOrElse(DataType.STRUCT))
+                numericLiteral(memsizer.memorySize(elementDt) * length, position)
+            }
+            dt.istype(DataType.STRUCT) -> {
+                when (target) {
+                    is VarDecl -> structSize(target.struct!!)
+                    is StructDecl -> structSize(target)
+                    else -> throw CompilerException("weird struct type $target")
+                }
+            }
+            dt.istype(DataType.STR) -> throw SyntaxError("sizeof str is undefined, did you mean len?", position)
+            else -> NumericLiteralValue(DataType.UBYTE, memsizer.memorySize(dt.typeOrElse(DataType.STRUCT)), position)
+        }
+    } else {
+        throw SyntaxError("sizeof invalid argument type", position)
+    }
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinLen(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    // note: in some cases the length is > 255 and then we have to return a UWORD type instead of a UBYTE.
+    if(args.size!=1)
+        throw SyntaxError("len requires one argument", position)
+
+    val directMemVar = ((args[0] as? DirectMemoryRead)?.addressExpression as? IdentifierReference)?.targetVarDecl(program)
+    var arraySize = directMemVar?.arraysize?.constIndex()
+    if(arraySize != null)
+        return NumericLiteralValue.optimalInteger(arraySize, position)
+    if(args[0] is ArrayLiteralValue)
+        return NumericLiteralValue.optimalInteger((args[0] as ArrayLiteralValue).value.size, position)
+    if(args[0] !is IdentifierReference)
+        throw SyntaxError("len argument should be an identifier", position)
+    val target = (args[0] as IdentifierReference).targetVarDecl(program)
+            ?: throw CannotEvaluateException("len", "no target vardecl")
+
+    return when(target.datatype) {
+        DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W, DataType.ARRAY_F -> {
+            arraySize = target.arraysize?.constIndex()
+            if(arraySize==null)
+                throw CannotEvaluateException("len", "arraysize unknown")
+            NumericLiteralValue.optimalInteger(arraySize, args[0].position)
+        }
+        DataType.STR -> {
+            val refLv = target.value as? StringLiteralValue ?: throw CannotEvaluateException("len", "stringsize unknown")
+            NumericLiteralValue.optimalInteger(refLv.value.length, args[0].position)
+        }
+        DataType.STRUCT -> throw SyntaxError("cannot use len on struct, did you mean sizeof?", args[0].position)
+        in NumericDatatypes -> throw SyntaxError("cannot use len on numeric value, did you mean sizeof?", args[0].position)
+        else -> throw CompilerException("weird datatype")
+    }
+}
+
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinMkword(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 2)
+        throw SyntaxError("mkword requires msb and lsb arguments", position)
+    val constMsb = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val constLsb = args[1].constValue(program) ?: throw NotConstArgumentException()
+    val result = (constMsb.number.toInt() shl 8) or constLsb.number.toInt()
+    return NumericLiteralValue(DataType.UWORD, result, position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin8(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("sin8 requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.BYTE, (127.0 * sin(rad)).toInt().toShort(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin8u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("sin8u requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.UBYTE, (128.0 + 127.5 * sin(rad)).toInt().toShort(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos8(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("cos8 requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.BYTE, (127.0 * cos(rad)).toInt().toShort(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos8u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("cos8u requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.UBYTE, (128.0 + 127.5 * cos(rad)).toInt().toShort(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin16(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("sin16 requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.WORD, (32767.0 * sin(rad)).toInt(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin16u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("sin16u requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.UWORD, (32768.0 + 32767.5 * sin(rad)).toInt(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos16(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("cos16 requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.WORD, (32767.0 * cos(rad)).toInt(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos16u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("cos16u requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    val rad = constval.number.toDouble() /256.0 * 2.0 * PI
+    return NumericLiteralValue(DataType.UWORD, (32768.0 + 32767.5 * cos(rad)).toInt(), position)
+}
+
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSgn(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    if (args.size != 1)
+        throw SyntaxError("sgn requires one argument", position)
+    val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
+    return NumericLiteralValue(DataType.BYTE, constval.number.toDouble().sign.toInt().toShort(), position)
+}
+
+private fun numericLiteral(value: Number, position: Position): NumericLiteralValue {
+    val floatNum=value.toDouble()
+    val tweakedValue: Number =
+            if(floatNum== floor(floatNum) && (floatNum>=-32768 && floatNum<=65535))
+                floatNum.toInt()  // we have an integer disguised as a float.
+            else
+                floatNum
+
+    return when(tweakedValue) {
+        is Int -> NumericLiteralValue.optimalInteger(value.toInt(), position)
+        is Short -> NumericLiteralValue.optimalInteger(value.toInt(), position)
+        is Byte -> NumericLiteralValue(DataType.UBYTE, value.toShort(), position)
+        is Double -> NumericLiteralValue(DataType.FLOAT, value.toDouble(), position)
+        is Float -> NumericLiteralValue(DataType.FLOAT, value.toDouble(), position)
+        else -> throw FatalAstException("invalid number type ${value::class}")
+    }
+}

compiler/src/prog8/compiler/intermediate/Instruction.kt

@@ -1,41 +0,0 @@
-package prog8.compiler.intermediate
-
-import prog8.stackvm.Syscall
-
-open class Instruction(val opcode: Opcode,
-                       val arg: Value? = null,
-                       val arg2: Value? = null,
-                       val callLabel: String? = null,
-                       val callLabel2: String? = null)
-{
-    lateinit var next: Instruction
-    var nextAlt: Instruction? = null
-
-    override fun toString(): String {
-        val argStr = arg?.toString() ?: ""
-        val result =
-                when {
-                    opcode==Opcode.LINE -> "_line  $callLabel"
-                    opcode==Opcode.INLINE_ASSEMBLY -> "inline_assembly"
-                    opcode==Opcode.SYSCALL -> {
-                        val syscall = Syscall.values().find { it.callNr==arg!!.numericValue() }
-                        "syscall  $syscall"
-                    }
-                    opcode in opcodesWithVarArgument -> {
-                        // opcodes that manipulate a variable
-                        "${opcode.toString().toLowerCase()}  ${callLabel?:""}  ${callLabel2?:""}".trimEnd()
-                    }
-                    callLabel==null -> "${opcode.toString().toLowerCase()}  $argStr"
-                    else -> "${opcode.toString().toLowerCase()}  $callLabel  $argStr"
-                }
-                .trimEnd()
-
-        return "    $result"
-    }
-}
-
-class LabelInstr(val name: String, val asmProc: Boolean) : Instruction(Opcode.NOP, null, null) {
-    override fun toString(): String {
-        return "\n$name:"
-    }
-}

compiler/src/prog8/compiler/intermediate/IntermediateProgram.kt

@@ -1,504 +0,0 @@
-package prog8.compiler.intermediate
-
-import prog8.ast.*
-import prog8.compiler.CompilerException
-import prog8.compiler.HeapValues
-import prog8.compiler.Zeropage
-import prog8.compiler.ZeropageDepletedError
-import java.io.PrintStream
-import java.nio.file.Path
-
-
-class IntermediateProgram(val name: String, var loadAddress: Int, val heap: HeapValues, val importedFrom: Path) {
-
-    class ProgramBlock(val scopedname: String,
-                       val shortname: String,
-                       var address: Int?,
-                       val instructions: MutableList<Instruction> = mutableListOf(),
-                       val variables: MutableMap<String, Value> = mutableMapOf(),
-                       val memoryPointers: MutableMap<String, Pair<Int, DataType>> = mutableMapOf(),
-                       val labels: MutableMap<String, Instruction> = mutableMapOf(),
-                       val force_output: Boolean)
-    {
-        val numVariables: Int
-            get() { return variables.size }
-        val numInstructions: Int
-            get() { return instructions.filter { it.opcode!= Opcode.LINE }.size }
-        val variablesMarkedForZeropage: MutableSet<String> = mutableSetOf()
-    }
-
-    val allocatedZeropageVariables = mutableMapOf<String, Pair<Int, DataType>>()
-    val blocks = mutableListOf<ProgramBlock>()
-    val memory = mutableMapOf<Int, List<Value>>()
-    private lateinit var currentBlock: ProgramBlock
-
-    val numVariables: Int
-        get() = blocks.sumBy { it.numVariables }
-    val numInstructions: Int
-        get() = blocks.sumBy { it.numInstructions }
-
-    fun allocateZeropage(zeropage: Zeropage) {
-        // allocates all @zp marked variables on the zeropage (for all blocks, as long as there is space in the ZP)
-        var notAllocated = 0
-        for(block in blocks) {
-            val zpVariables = block.variables.filter { it.key in block.variablesMarkedForZeropage }
-            if (zpVariables.isNotEmpty()) {
-                for (variable in zpVariables) {
-                    try {
-                        val address = zeropage.allocate(variable.key, variable.value.type, null)
-                        allocatedZeropageVariables[variable.key] = Pair(address, variable.value.type)
-                    } catch (x: ZeropageDepletedError) {
-                        printWarning(x.toString() + " variable ${variable.key} type ${variable.value.type}")
-                        notAllocated++
-                    }
-                }
-            }
-        }
-        if(notAllocated>0)
-            printWarning("$notAllocated variables marked for Zeropage could not be allocated there")
-    }
-
-    fun optimize() {
-        println("Optimizing stackVM code...")
-        // remove nops (that are not a label)
-        for (blk in blocks) {
-            blk.instructions.removeIf { it.opcode== Opcode.NOP && it !is LabelInstr }
-        }
-
-        optimizeDataConversionAndUselessDiscards()
-        optimizeVariableCopying()
-        optimizeMultipleSequentialLineInstrs()
-        optimizeCallReturnIntoJump()
-        optimizeConditionalBranches()
-        // todo: add more optimizations to stackvm code
-
-        optimizeRemoveNops()    //  must be done as the last step
-        optimizeMultipleSequentialLineInstrs()      // once more
-        optimizeRemoveNops()    // once more
-    }
-
-    private fun optimizeConditionalBranches() {
-        // conditional branches that consume the value on the stack
-        // sometimes these are just constant values, so we can statically determine the branch
-        // or, they are preceded by a NOT instruction so we can simply remove that and flip the branch condition
-        val pushvalue = setOf(Opcode.PUSH_BYTE, Opcode.PUSH_WORD)
-        val notvalue = setOf(Opcode.NOT_BYTE, Opcode.NOT_WORD)
-        val branchOpcodes = setOf(Opcode.JZ, Opcode.JNZ, Opcode.JZW, Opcode.JNZW)
-        for(blk in blocks) {
-            val instructionsToReplace = mutableMapOf<Int, Instruction>()
-            blk.instructions.asSequence().withIndex().filter {it.value.opcode!=Opcode.LINE}.windowed(2).toList().forEach {
-                if (it[1].value.opcode in branchOpcodes) {
-                    if (it[0].value.opcode in pushvalue) {
-                        val value = it[0].value.arg!!.asBooleanValue
-                        instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                        val replacement: Instruction =
-                                if (value) {
-                                    when (it[1].value.opcode) {
-                                        Opcode.JNZ -> Instruction(Opcode.JUMP, callLabel = it[1].value.callLabel)
-                                        Opcode.JNZW -> Instruction(Opcode.JUMP, callLabel = it[1].value.callLabel)
-                                        else -> Instruction(Opcode.NOP)
-                                    }
-                                } else {
-                                    when (it[1].value.opcode) {
-                                        Opcode.JZ -> Instruction(Opcode.JUMP, callLabel = it[1].value.callLabel)
-                                        Opcode.JZW -> Instruction(Opcode.JUMP, callLabel = it[1].value.callLabel)
-                                        else -> Instruction(Opcode.NOP)
-                                    }
-                                }
-                        instructionsToReplace[it[1].index] = replacement
-                    }
-                    else if (it[0].value.opcode in notvalue) {
-                        instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                        val replacement: Instruction =
-                                when (it[1].value.opcode) {
-                                    Opcode.JZ -> Instruction(Opcode.JNZ, callLabel = it[1].value.callLabel)
-                                    Opcode.JZW -> Instruction(Opcode.JNZW, callLabel = it[1].value.callLabel)
-                                    Opcode.JNZ -> Instruction(Opcode.JZ, callLabel = it[1].value.callLabel)
-                                    Opcode.JNZW -> Instruction(Opcode.JZW, callLabel = it[1].value.callLabel)
-                                    else -> Instruction(Opcode.NOP)
-                                }
-                        instructionsToReplace[it[1].index] = replacement
-                    }
-                }
-            }
-
-            for (rins in instructionsToReplace) {
-                blk.instructions[rins.key] = rins.value
-            }
-        }
-    }
-
-    private fun optimizeRemoveNops() {
-        // remove nops (that are not a label)
-        for (blk in blocks)
-            blk.instructions.removeIf { it.opcode== Opcode.NOP && it !is LabelInstr }
-    }
-
-    private fun optimizeCallReturnIntoJump() {
-        // replaces call X followed by return, by jump X
-        for(blk in blocks) {
-            val instructionsToReplace = mutableMapOf<Int, Instruction>()
-
-            blk.instructions.asSequence().withIndex().filter {it.value.opcode!=Opcode.LINE}.windowed(2).toList().forEach {
-                if(it[0].value.opcode==Opcode.CALL && it[1].value.opcode==Opcode.RETURN) {
-                    instructionsToReplace[it[1].index] = Instruction(Opcode.JUMP, callLabel = it[0].value.callLabel)
-                    instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                }
-            }
-
-            for (rins in instructionsToReplace) {
-                blk.instructions[rins.key] = rins.value
-            }
-        }
-    }
-
-    private fun optimizeMultipleSequentialLineInstrs() {
-        for(blk in blocks) {
-            val instructionsToReplace = mutableMapOf<Int, Instruction>()
-
-            blk.instructions.asSequence().withIndex().windowed(2).toList().forEach {
-                if (it[0].value.opcode == Opcode.LINE && it[1].value.opcode == Opcode.LINE)
-                    instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-            }
-
-            for (rins in instructionsToReplace) {
-                blk.instructions[rins.key] = rins.value
-            }
-        }
-    }
-
-    private fun optimizeVariableCopying() {
-        for(blk in blocks) {
-
-            val instructionsToReplace = mutableMapOf<Int, Instruction>()
-
-            blk.instructions.asSequence().withIndex().windowed(2).toList().forEach {
-                when (it[0].value.opcode) {
-                    Opcode.PUSH_VAR_BYTE ->
-                        if (it[1].value.opcode == Opcode.POP_VAR_BYTE) {
-                            if (it[0].value.callLabel == it[1].value.callLabel) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    Opcode.PUSH_VAR_WORD ->
-                        if (it[1].value.opcode == Opcode.POP_VAR_WORD) {
-                            if (it[0].value.callLabel == it[1].value.callLabel) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    Opcode.PUSH_VAR_FLOAT ->
-                        if (it[1].value.opcode == Opcode.POP_VAR_FLOAT) {
-                            if (it[0].value.callLabel == it[1].value.callLabel) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    Opcode.PUSH_MEM_B, Opcode.PUSH_MEM_UB ->
-                        if(it[1].value.opcode == Opcode.POP_MEM_BYTE) {
-                            if(it[0].value.arg == it[1].value.arg) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    Opcode.PUSH_MEM_W, Opcode.PUSH_MEM_UW ->
-                        if(it[1].value.opcode == Opcode.POP_MEM_WORD) {
-                            if(it[0].value.arg == it[1].value.arg) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    Opcode.PUSH_MEM_FLOAT ->
-                        if(it[1].value.opcode == Opcode.POP_MEM_FLOAT) {
-                            if(it[0].value.arg == it[1].value.arg) {
-                                instructionsToReplace[it[0].index] = Instruction(Opcode.NOP)
-                                instructionsToReplace[it[1].index] = Instruction(Opcode.NOP)
-                            }
-                        }
-                    else -> {}
-                }
-            }
-
-            for (rins in instructionsToReplace) {
-                blk.instructions[rins.key] = rins.value
-            }
-        }
-    }
-
-    private fun optimizeDataConversionAndUselessDiscards() {
-        // - push value followed by a data type conversion -> push the value in the correct type and remove the conversion
-        // - push something followed by a discard -> remove both
-        val instructionsToReplace = mutableMapOf<Int, Instruction>()
-
-        fun optimizeDiscardAfterPush(index0: Int, index1: Int, ins1: Instruction) {
-            if (ins1.opcode == Opcode.DISCARD_FLOAT || ins1.opcode == Opcode.DISCARD_WORD || ins1.opcode == Opcode.DISCARD_BYTE) {
-                instructionsToReplace[index0] = Instruction(Opcode.NOP)
-                instructionsToReplace[index1] = Instruction(Opcode.NOP)
-            }
-        }
-
-        fun optimizeFloatConversion(index0: Int, index1: Int, ins1: Instruction) {
-            when (ins1.opcode) {
-                Opcode.DISCARD_FLOAT -> {
-                    instructionsToReplace[index0] = Instruction(Opcode.NOP)
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.DISCARD_BYTE, Opcode.DISCARD_WORD -> throw CompilerException("invalid discard type following a float")
-                else -> throw CompilerException("invalid conversion opcode ${ins1.opcode} following a float")
-            }
-        }
-
-        fun optimizeWordConversion(index0: Int, ins0: Instruction, index1: Int, ins1: Instruction) {
-            when (ins1.opcode) {
-                Opcode.CAST_UW_TO_B, Opcode.CAST_W_TO_B -> {
-                    val ins = Instruction(Opcode.PUSH_BYTE, ins0.arg!!.cast(DataType.BYTE))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_W_TO_UB, Opcode.CAST_UW_TO_UB -> {
-                    val ins = Instruction(Opcode.PUSH_BYTE, Value(DataType.UBYTE, ins0.arg!!.integerValue() and 255))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.MSB -> {
-                    val ins = Instruction(Opcode.PUSH_BYTE, Value(DataType.UBYTE, ins0.arg!!.integerValue() ushr 8 and 255))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_W_TO_F, Opcode.CAST_UW_TO_F -> {
-                    val ins = Instruction(Opcode.PUSH_FLOAT, Value(DataType.FLOAT, ins0.arg!!.integerValue().toDouble()))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_UW_TO_W -> {
-                    val cv = ins0.arg!!.cast(DataType.WORD)
-                    instructionsToReplace[index0] = Instruction(Opcode.PUSH_WORD, cv)
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_W_TO_UW -> {
-                    val cv = ins0.arg!!.cast(DataType.UWORD)
-                    instructionsToReplace[index0] = Instruction(Opcode.PUSH_WORD, cv)
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.DISCARD_WORD -> {
-                    instructionsToReplace[index0] = Instruction(Opcode.NOP)
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.DISCARD_BYTE, Opcode.DISCARD_FLOAT -> throw CompilerException("invalid discard type following a byte")
-                else -> throw CompilerException("invalid conversion opcode ${ins1.opcode} following a word")
-            }
-        }
-
-        fun optimizeByteConversion(index0: Int, ins0: Instruction, index1: Int, ins1: Instruction) {
-            when (ins1.opcode) {
-                Opcode.CAST_B_TO_UB, Opcode.CAST_UB_TO_B,
-                Opcode.CAST_W_TO_B, Opcode.CAST_W_TO_UB,
-                Opcode.CAST_UW_TO_B, Opcode.CAST_UW_TO_UB -> instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                Opcode.MSB -> throw CompilerException("msb of a byte")
-                Opcode.CAST_UB_TO_UW -> {
-                    val ins = Instruction(Opcode.PUSH_WORD, Value(DataType.UWORD, ins0.arg!!.integerValue()))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_B_TO_W -> {
-                    val ins = Instruction(Opcode.PUSH_WORD, Value(DataType.WORD, ins0.arg!!.integerValue()))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_B_TO_UW -> {
-                    val ins = Instruction(Opcode.PUSH_WORD, ins0.arg!!.cast(DataType.UWORD))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_UB_TO_W -> {
-                    val ins = Instruction(Opcode.PUSH_WORD, ins0.arg!!.cast(DataType.WORD))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_B_TO_F, Opcode.CAST_UB_TO_F-> {
-                    val ins = Instruction(Opcode.PUSH_FLOAT, Value(DataType.FLOAT, ins0.arg!!.integerValue().toDouble()))
-                    instructionsToReplace[index0] = ins
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.CAST_W_TO_F, Opcode.CAST_UW_TO_F-> throw CompilerException("invalid conversion following a byte")
-                Opcode.DISCARD_BYTE -> {
-                    instructionsToReplace[index0] = Instruction(Opcode.NOP)
-                    instructionsToReplace[index1] = Instruction(Opcode.NOP)
-                }
-                Opcode.DISCARD_WORD, Opcode.DISCARD_FLOAT -> throw CompilerException("invalid discard type following a byte")
-                else -> throw CompilerException("invalid conversion opcode ${ins1.opcode}")
-            }
-        }
-
-        for(blk in blocks) {
-            instructionsToReplace.clear()
-
-            val typeConversionOpcodes = setOf(
-                    Opcode.MSB,
-                    Opcode.MKWORD,
-                    Opcode.CAST_UB_TO_B,
-                    Opcode.CAST_UB_TO_UW,
-                    Opcode.CAST_UB_TO_W,
-                    Opcode.CAST_UB_TO_F,
-                    Opcode.CAST_B_TO_UB,
-                    Opcode.CAST_B_TO_UW,
-                    Opcode.CAST_B_TO_W,
-                    Opcode.CAST_B_TO_F,
-                    Opcode.CAST_UW_TO_UB,
-                    Opcode.CAST_UW_TO_B,
-                    Opcode.CAST_UW_TO_W,
-                    Opcode.CAST_UW_TO_F,
-                    Opcode.CAST_W_TO_UB,
-                    Opcode.CAST_W_TO_B,
-                    Opcode.CAST_W_TO_UW,
-                    Opcode.CAST_W_TO_F,
-                    Opcode.CAST_F_TO_UB,
-                    Opcode.CAST_F_TO_B,
-                    Opcode.CAST_F_TO_UW,
-                    Opcode.CAST_F_TO_W,
-                    Opcode.DISCARD_BYTE,
-                    Opcode.DISCARD_WORD,
-                    Opcode.DISCARD_FLOAT
-            )
-            blk.instructions.asSequence().withIndex().windowed(2).toList().forEach {
-                if (it[1].value.opcode in typeConversionOpcodes) {
-                    when (it[0].value.opcode) {
-                        Opcode.PUSH_BYTE -> optimizeByteConversion(it[0].index, it[0].value, it[1].index, it[1].value)
-                        Opcode.PUSH_WORD -> optimizeWordConversion(it[0].index, it[0].value, it[1].index, it[1].value)
-                        Opcode.PUSH_FLOAT -> optimizeFloatConversion(it[0].index, it[1].index, it[1].value)
-                        Opcode.PUSH_VAR_FLOAT,
-                        Opcode.PUSH_VAR_WORD,
-                        Opcode.PUSH_VAR_BYTE,
-                        Opcode.PUSH_MEM_B, Opcode.PUSH_MEM_UB,
-                        Opcode.PUSH_MEM_W, Opcode.PUSH_MEM_UW,
-                        Opcode.PUSH_MEM_FLOAT -> optimizeDiscardAfterPush(it[0].index, it[1].index, it[1].value)
-                        else -> {
-                        }
-                    }
-                }
-            }
-
-            for (rins in instructionsToReplace) {
-                blk.instructions[rins.key] = rins.value
-            }
-        }
-    }
-
-    fun variable(scopedname: String, decl: VarDecl) {
-        when(decl.type) {
-            VarDeclType.VAR -> {
-                val value = when(decl.datatype) {
-                    DataType.UBYTE, DataType.BYTE, DataType.UWORD, DataType.WORD, DataType.FLOAT -> Value(decl.datatype, (decl.value as LiteralValue).asNumericValue!!)
-                    DataType.STR, DataType.STR_P, DataType.STR_S, DataType.STR_PS -> {
-                        val litval = (decl.value as LiteralValue)
-                        if(litval.heapId==null)
-                            throw CompilerException("string should already be in the heap")
-                        Value(decl.datatype, litval.heapId)
-                    }
-                    DataType.ARRAY_B, DataType.ARRAY_W,
-                    DataType.ARRAY_UB, DataType.ARRAY_UW, DataType.ARRAY_F -> {
-                        val litval = (decl.value as LiteralValue)
-                        if(litval.heapId==null)
-                            throw CompilerException("array should already be in the heap")
-                        Value(decl.datatype, litval.heapId)
-                    }
-                }
-                currentBlock.variables[scopedname] = value
-                if(decl.zeropage)
-                    currentBlock.variablesMarkedForZeropage.add(scopedname)
-            }
-            VarDeclType.MEMORY -> {
-                // note that constants are all folded away, but assembly code may still refer to them
-                val lv = decl.value as LiteralValue
-                if(lv.type!=DataType.UWORD && lv.type!=DataType.UBYTE)
-                    throw CompilerException("expected integer memory address $lv")
-                currentBlock.memoryPointers[scopedname] = Pair(lv.asIntegerValue!!, decl.datatype)
-            }
-            VarDeclType.CONST -> {
-                // note that constants are all folded away, but assembly code may still refer to them (if their integers)
-                // floating point constants are not generated at all!!
-                val lv = decl.value as LiteralValue
-                if(lv.type in IntegerDatatypes)
-                    currentBlock.memoryPointers[scopedname] = Pair(lv.asIntegerValue!!, decl.datatype)
-            }
-        }
-    }
-
-    fun instr(opcode: Opcode, arg: Value? = null, arg2: Value? = null, callLabel: String? = null, callLabel2: String? = null) {
-        currentBlock.instructions.add(Instruction(opcode, arg, arg2, callLabel, callLabel2))
-    }
-
-    fun label(labelname: String, asmProc: Boolean=false) {
-        val instr = LabelInstr(labelname, asmProc)
-        currentBlock.instructions.add(instr)
-        currentBlock.labels[labelname] = instr
-    }
-
-    fun line(position: Position) {
-        currentBlock.instructions.add(Instruction(Opcode.LINE, callLabel = "${position.line} ${position.file}"))
-    }
-
-    fun removeLastInstruction() {
-        currentBlock.instructions.removeAt(currentBlock.instructions.lastIndex)
-    }
-
-    fun memoryPointer(name: String, address: Int, datatype: DataType) {
-        currentBlock.memoryPointers[name] = Pair(address, datatype)
-    }
-
-    fun newBlock(scopedname: String, shortname: String, address: Int?, options: Set<String>) {
-        currentBlock = ProgramBlock(scopedname, shortname, address, force_output="force_output" in options)
-        blocks.add(currentBlock)
-    }
-
-    fun writeCode(out: PrintStream, embeddedLabels: Boolean=true) {
-        out.println("; stackVM program code for '$name'")
-        out.println("%memory")
-        if(memory.isNotEmpty())
-            TODO("add support for writing/reading initial memory values")
-        out.println("%end_memory")
-        out.println("%heap")
-        heap.allEntries().forEach {
-            when {
-                it.value.str!=null ->
-                    out.println("${it.key}  ${it.value.type.toString().toLowerCase()}  \"${escape(it.value.str!!)}\"")
-                it.value.array!=null ->
-                    out.println("${it.key}  ${it.value.type.toString().toLowerCase()}  ${it.value.array!!.toList()}")
-                it.value.doubleArray!=null ->
-                    out.println("${it.key}  ${it.value.type.toString().toLowerCase()}  ${it.value.doubleArray!!.toList()}")
-                else -> throw CompilerException("invalid heap entry $it")
-            }
-        }
-        out.println("%end_heap")
-        for(blk in blocks) {
-            out.println("\n%block ${blk.scopedname} ${blk.address?.toString(16) ?: ""}")
-
-            out.println("%variables")
-            for(variable in blk.variables) {
-                val valuestr = variable.value.toString()
-                out.println("${variable.key}  ${variable.value.type.toString().toLowerCase()}  $valuestr")
-            }
-            out.println("%end_variables")
-            out.println("%memorypointers")
-            for(iconst in blk.memoryPointers) {
-                out.println("${iconst.key}  ${iconst.value.second.toString().toLowerCase()}  uw:${iconst.value.first.toString(16)}")
-            }
-            out.println("%end_memorypointers")
-            out.println("%instructions")
-            val labels = blk.labels.entries.associateBy({it.value}) {it.key}
-            for(instr in blk.instructions) {
-                if(!embeddedLabels) {
-                    val label = labels[instr]
-                    if (label != null)
-                        out.println("$label:")
-                } else {
-                    out.println(instr)
-                }
-            }
-            out.println("%end_instructions")
-
-            out.println("%end_block")
-        }
-    }
-}

compiler/src/prog8/compiler/intermediate/Opcode.kt

@@ -1,292 +0,0 @@
-package prog8.compiler.intermediate
-
-enum class Opcode {
-
-    // pushing values on the (evaluation) stack
-    PUSH_BYTE,       // push byte value
-    PUSH_WORD,       // push word value   (or 'address' of string / array)
-    PUSH_FLOAT,      // push float value
-    PUSH_MEM_B,      // push byte value from memory to stack
-    PUSH_MEM_UB,     // push unsigned byte value from memory to stack
-    PUSH_MEM_W,      // push word value from memory to stack
-    PUSH_MEM_UW,     // push unsigned word value from memory to stack
-    PUSH_MEM_FLOAT,  // push float value from memory to stack
-    PUSH_MEMREAD,    // push memory value from address that's on the stack
-    PUSH_VAR_BYTE,   // push byte variable (ubyte, byte)
-    PUSH_VAR_WORD,   // push word variable (uword, word)
-    PUSH_VAR_FLOAT,  // push float variable
-    PUSH_REGAX_WORD, // push registers A/X as a 16-bit word
-    PUSH_REGAY_WORD, // push registers A/Y as a 16-bit word
-    PUSH_REGXY_WORD, // push registers X/Y as a 16-bit word
-    PUSH_ADDR_HEAPVAR,  // push the address of the variable that's on the heap (string or array)
-
-    // popping values off the (evaluation) stack, possibly storing them in another location
-    DISCARD_BYTE,    // discard top byte value
-    DISCARD_WORD,    // discard top word value
-    DISCARD_FLOAT,   // discard top float value
-    POP_MEM_BYTE,    // pop (u)byte value into destination memory address
-    POP_MEM_WORD,    // pop (u)word value into destination memory address
-    POP_MEM_FLOAT,   // pop float value into destination memory address
-    POP_MEMWRITE,    // pop address and byte stack and write the byte to the memory address
-    POP_VAR_BYTE,    // pop (u)byte value into variable
-    POP_VAR_WORD,    // pop (u)word value into variable
-    POP_VAR_FLOAT,   // pop float value into variable
-    POP_REGAX_WORD,  // pop uword from stack into A/X registers
-    POP_REGAY_WORD,  // pop uword from stack into A/Y registers
-    POP_REGXY_WORD,  // pop uword from stack into X/Y registers
-
-    // numeric arithmetic
-    ADD_UB,
-    ADD_B,
-    ADD_UW,
-    ADD_W,
-    ADD_F,
-    SUB_UB,
-    SUB_B,
-    SUB_UW,
-    SUB_W,
-    SUB_F,
-    MUL_UB,
-    MUL_B,
-    MUL_UW,
-    MUL_W,
-    MUL_F,
-    IDIV_UB,
-    IDIV_B,
-    IDIV_UW,
-    IDIV_W,
-    DIV_F,
-    REMAINDER_UB,   // signed remainder is undefined/unimplemented
-    REMAINDER_UW,   // signed remainder is undefined/unimplemented
-    POW_UB,
-    POW_B,
-    POW_UW,
-    POW_W,
-    POW_F,
-    NEG_B,
-    NEG_W,
-    NEG_F,
-    ABS_B,
-    ABS_W,
-    ABS_F,
-
-    // bit shifts and bitwise arithmetic
-    SHIFTEDL_BYTE,      // shifts stack value rather than in-place mem/var
-    SHIFTEDL_WORD,      // shifts stack value rather than in-place mem/var
-    SHIFTEDR_UBYTE,     // shifts stack value rather than in-place mem/var
-    SHIFTEDR_SBYTE,     // shifts stack value rather than in-place mem/var
-    SHIFTEDR_UWORD,     // shifts stack value rather than in-place mem/var
-    SHIFTEDR_SWORD,     // shifts stack value rather than in-place mem/var
-    SHL_BYTE,
-    SHL_WORD,
-    SHL_MEM_BYTE,
-    SHL_MEM_WORD,
-    SHL_VAR_BYTE,
-    SHL_VAR_WORD,
-    SHR_UBYTE,
-    SHR_SBYTE,
-    SHR_UWORD,
-    SHR_SWORD,
-    SHR_MEM_UBYTE,
-    SHR_MEM_SBYTE,
-    SHR_MEM_UWORD,
-    SHR_MEM_SWORD,
-    SHR_VAR_UBYTE,
-    SHR_VAR_SBYTE,
-    SHR_VAR_UWORD,
-    SHR_VAR_SWORD,
-    ROL_BYTE,
-    ROL_WORD,
-    ROL_MEM_BYTE,
-    ROL_MEM_WORD,
-    ROL_VAR_BYTE,
-    ROL_VAR_WORD,
-    ROR_BYTE,
-    ROR_WORD,
-    ROR_MEM_BYTE,
-    ROR_MEM_WORD,
-    ROR_VAR_BYTE,
-    ROR_VAR_WORD,
-    ROL2_BYTE,
-    ROL2_WORD,
-    ROL2_MEM_BYTE,
-    ROL2_MEM_WORD,
-    ROL2_VAR_BYTE,
-    ROL2_VAR_WORD,
-    ROR2_BYTE,
-    ROR2_WORD,
-    ROR2_MEM_BYTE,
-    ROR2_MEM_WORD,
-    ROR2_VAR_BYTE,
-    ROR2_VAR_WORD,
-    BITAND_BYTE,
-    BITAND_WORD,
-    BITOR_BYTE,
-    BITOR_WORD,
-    BITXOR_BYTE,
-    BITXOR_WORD,
-    INV_BYTE,
-    INV_WORD,
-
-    // numeric type conversions
-    MSB,        // note: lsb is equivalent to  CAST_UW_TO_UB  or CAST_W_TO_UB
-    MKWORD,        // create a word from lsb + msb
-    CAST_UB_TO_B,
-    CAST_UB_TO_UW,
-    CAST_UB_TO_W,
-    CAST_UB_TO_F,
-    CAST_B_TO_UB,
-    CAST_B_TO_UW,
-    CAST_B_TO_W,
-    CAST_B_TO_F,
-    CAST_W_TO_UB,
-    CAST_W_TO_B,
-    CAST_W_TO_UW,
-    CAST_W_TO_F,
-    CAST_UW_TO_UB,
-    CAST_UW_TO_B,
-    CAST_UW_TO_W,
-    CAST_UW_TO_F,
-    CAST_F_TO_UB,
-    CAST_F_TO_B,
-    CAST_F_TO_UW,
-    CAST_F_TO_W,
-
-    // logical operations
-    AND_BYTE,
-    AND_WORD,
-    OR_BYTE,
-    OR_WORD,
-    XOR_BYTE,
-    XOR_WORD,
-    NOT_BYTE,
-    NOT_WORD,
-
-    // increment, decrement
-    INC_VAR_B,
-    INC_VAR_UB,
-    INC_VAR_W,
-    INC_VAR_UW,
-    INC_VAR_F,
-    DEC_VAR_B,
-    DEC_VAR_UB,
-    DEC_VAR_W,
-    DEC_VAR_UW,
-    DEC_VAR_F,
-    INC_MEMORY,             // increment direct address
-    DEC_MEMORY,             // decrement direct address
-    POP_INC_MEMORY,         // increment address from stack
-    POP_DEC_MEMORY,         // decrement address from address
-
-    // comparisons
-    LESS_B,
-    LESS_UB,
-    LESS_W,
-    LESS_UW,
-    LESS_F,
-    GREATER_B,
-    GREATER_UB,
-    GREATER_W,
-    GREATER_UW,
-    GREATER_F,
-    LESSEQ_B,
-    LESSEQ_UB,
-    LESSEQ_W,
-    LESSEQ_UW,
-    LESSEQ_F,
-    GREATEREQ_B,
-    GREATEREQ_UB,
-    GREATEREQ_W,
-    GREATEREQ_UW,
-    GREATEREQ_F,
-    EQUAL_BYTE,
-    EQUAL_WORD,
-    EQUAL_F,
-    NOTEQUAL_BYTE,
-    NOTEQUAL_WORD,
-    NOTEQUAL_F,
-    CMP_B,          // sets processor status flags based on comparison, instead of pushing a result value
-    CMP_UB,         // sets processor status flags based on comparison, instead of pushing a result value
-    CMP_W,          // sets processor status flags based on comparison, instead of pushing a result value
-    CMP_UW,         // sets processor status flags based on comparison, instead of pushing a result value
-
-    // array access and simple manipulations
-    READ_INDEXED_VAR_BYTE,
-    READ_INDEXED_VAR_WORD,
-    READ_INDEXED_VAR_FLOAT,
-    WRITE_INDEXED_VAR_BYTE,
-    WRITE_INDEXED_VAR_WORD,
-    WRITE_INDEXED_VAR_FLOAT,
-    INC_INDEXED_VAR_B,
-    INC_INDEXED_VAR_UB,
-    INC_INDEXED_VAR_W,
-    INC_INDEXED_VAR_UW,
-    INC_INDEXED_VAR_FLOAT,
-    DEC_INDEXED_VAR_B,
-    DEC_INDEXED_VAR_UB,
-    DEC_INDEXED_VAR_W,
-    DEC_INDEXED_VAR_UW,
-    DEC_INDEXED_VAR_FLOAT,
-
-    // branching, without consuming a value from the stack
-    JUMP,
-    BCS,       // branch if carry set
-    BCC,       // branch if carry clear
-    BZ,        // branch if zero flag
-    BNZ,       // branch if not zero flag
-    BNEG,      // branch if negative flag
-    BPOS,      // branch if not negative flag
-    BVS,       // branch if overflow flag
-    BVC,       // branch if not overflow flag
-    // branching, based on value on the stack (which is consumed)
-    JZ,         // branch if value is zero (byte)
-    JNZ,        // branch if value is not zero (byte)
-    JZW,         // branch if value is zero (word)
-    JNZW,        // branch if value is not zero (word)
-
-
-    // subroutines
-    CALL,
-    RETURN,
-    SYSCALL,
-    START_PROCDEF,
-    END_PROCDEF,
-
-    // misc
-    SEC,        // set carry status flag  NOTE: is mostly fake, carry flag is not affected by any numeric operations
-    CLC,        // clear carry status flag  NOTE: is mostly fake, carry flag is not affected by any numeric operations
-    SEI,        // set irq-disable status flag
-    CLI,        // clear irq-disable status flag
-    RSAVE,      // save all internal registers and status flags
-    RSAVEX,     // save just X (the evaluation stack pointer)
-    RRESTORE,   // restore all internal registers and status flags
-    RRESTOREX,  // restore just X (the evaluation stack pointer)
-
-    NOP,        // do nothing
-    BREAKPOINT, // breakpoint
-    TERMINATE,  // end the program
-    LINE,       // track source file line number
-    INLINE_ASSEMBLY         // container to hold inline raw assembly code
-}
-
-val opcodesWithVarArgument = setOf(
-        Opcode.INC_VAR_B, Opcode.INC_VAR_W, Opcode.DEC_VAR_B, Opcode.DEC_VAR_W,
-        Opcode.INC_VAR_UB, Opcode.INC_VAR_UW, Opcode.DEC_VAR_UB, Opcode.DEC_VAR_UW,
-        Opcode.SHR_VAR_SBYTE, Opcode.SHR_VAR_UBYTE, Opcode.SHR_VAR_SWORD, Opcode.SHR_VAR_UWORD,
-        Opcode.SHL_VAR_BYTE, Opcode.SHL_VAR_WORD,
-        Opcode.ROL_VAR_BYTE, Opcode.ROL_VAR_WORD, Opcode.ROR_VAR_BYTE, Opcode.ROR_VAR_WORD,
-        Opcode.ROL2_VAR_BYTE, Opcode.ROL2_VAR_WORD, Opcode.ROR2_VAR_BYTE, Opcode.ROR2_VAR_WORD,
-        Opcode.POP_VAR_BYTE, Opcode.POP_VAR_WORD, Opcode.POP_VAR_FLOAT,
-        Opcode.PUSH_VAR_BYTE, Opcode.PUSH_VAR_WORD, Opcode.PUSH_VAR_FLOAT, Opcode.PUSH_ADDR_HEAPVAR,
-        Opcode.READ_INDEXED_VAR_BYTE, Opcode.READ_INDEXED_VAR_WORD, Opcode.READ_INDEXED_VAR_FLOAT,
-        Opcode.WRITE_INDEXED_VAR_BYTE, Opcode.WRITE_INDEXED_VAR_WORD, Opcode.WRITE_INDEXED_VAR_FLOAT,
-        Opcode.INC_INDEXED_VAR_UB, Opcode.INC_INDEXED_VAR_B, Opcode.INC_INDEXED_VAR_UW,
-        Opcode.INC_INDEXED_VAR_W, Opcode.INC_INDEXED_VAR_FLOAT,
-        Opcode.DEC_INDEXED_VAR_UB, Opcode.DEC_INDEXED_VAR_B, Opcode.DEC_INDEXED_VAR_UW,
-        Opcode.DEC_INDEXED_VAR_W, Opcode.DEC_INDEXED_VAR_FLOAT
-)
-
-val branchOpcodes = setOf(
-        Opcode.BCS, Opcode.BCC, Opcode.BZ, Opcode.BNZ,
-        Opcode.BNEG, Opcode.BPOS, Opcode.BVS, Opcode.BVC
-)

compiler/src/prog8/compiler/intermediate/Value.kt

@@ -1,478 +0,0 @@
-package prog8.compiler.intermediate
-
-import prog8.ast.*
-import java.lang.Exception
-import kotlin.math.abs
-import kotlin.math.pow
-
-
-class ValueException(msg: String?) : Exception(msg)
-
-
-class Value(val type: DataType, numericvalueOrHeapId: Number) {
-    private var byteval: Short? = null
-    private var wordval: Int? = null
-    private var floatval: Double? = null
-    var heapId: Int = -1
-        private set
-    val asBooleanValue: Boolean
-
-    init {
-        when(type) {
-            DataType.UBYTE -> {
-                if(numericvalueOrHeapId.toInt() !in 0..255)
-                    throw ValueException("value out of range: $numericvalueOrHeapId")
-                byteval = numericvalueOrHeapId.toShort()
-                asBooleanValue = byteval != (0.toShort())
-            }
-            DataType.BYTE -> {
-                if(numericvalueOrHeapId.toInt() !in -128..127)
-                    throw ValueException("value out of range: $numericvalueOrHeapId")
-                byteval = numericvalueOrHeapId.toShort()
-                asBooleanValue = byteval != (0.toShort())
-            }
-            DataType.UWORD -> {
-                if(numericvalueOrHeapId.toInt() !in 0..65535)
-                    throw ValueException("value out of range: $numericvalueOrHeapId")
-                wordval = numericvalueOrHeapId.toInt()
-                asBooleanValue = wordval != 0
-            }
-            DataType.WORD -> {
-                if(numericvalueOrHeapId.toInt() !in -32768..32767)
-                    throw ValueException("value out of range: $numericvalueOrHeapId")
-                wordval = numericvalueOrHeapId.toInt()
-                asBooleanValue = wordval != 0
-            }
-            DataType.FLOAT -> {
-                floatval = numericvalueOrHeapId.toDouble()
-                asBooleanValue = floatval != 0.0
-            }
-            else -> {
-                if(numericvalueOrHeapId !is Int || numericvalueOrHeapId<0)
-                    throw ValueException("for non-numeric types, the value should be a integer heapId >= 0")
-                heapId = numericvalueOrHeapId
-                asBooleanValue=true
-            }
-        }
-    }
-
-    override fun toString(): String {
-        return when(type) {
-            DataType.UBYTE -> "ub:%02x".format(byteval)
-            DataType.BYTE -> {
-                if(byteval!!<0)
-                    "b:-%02x".format(abs(byteval!!.toInt()))
-                else
-                    "b:%02x".format(byteval)
-            }
-            DataType.UWORD -> "uw:%04x".format(wordval)
-            DataType.WORD -> {
-                if(wordval!!<0)
-                    "w:-%04x".format(abs(wordval!!))
-                else
-                    "w:%04x".format(wordval)
-            }
-            DataType.FLOAT -> "f:$floatval"
-            else -> "heap:$heapId"
-        }
-    }
-
-    fun numericValue(): Number {
-        return when(type) {
-            DataType.UBYTE, DataType.BYTE -> byteval!!
-            DataType.UWORD, DataType.WORD -> wordval!!
-            DataType.FLOAT -> floatval!!
-            else -> throw ValueException("invalid datatype for numeric value: $type")
-        }
-    }
-
-    fun integerValue(): Int {
-        return when(type) {
-            DataType.UBYTE, DataType.BYTE -> byteval!!.toInt()
-            DataType.UWORD, DataType.WORD -> wordval!!
-            DataType.FLOAT -> throw ValueException("float to integer loss of precision")
-            else -> throw ValueException("invalid datatype for integer value: $type")
-        }
-    }
-
-    override fun hashCode(): Int {
-        val bh = byteval?.hashCode() ?: 0x10001234
-        val wh = wordval?.hashCode() ?: 0x01002345
-        val fh = floatval?.hashCode() ?: 0x00103456
-        return bh xor wh xor fh xor heapId.hashCode() xor type.hashCode()
-    }
-
-    override fun equals(other: Any?): Boolean {
-        if(other==null || other !is Value)
-            return false
-        if(type==other.type)
-            return if (type in IterableDatatypes) heapId==other.heapId else compareTo(other)==0
-        return compareTo(other)==0      // note: datatype doesn't matter
-    }
-
-    operator fun compareTo(other: Value): Int {
-        return if (type in NumericDatatypes && other.type in NumericDatatypes)
-                numericValue().toDouble().compareTo(other.numericValue().toDouble())
-            else throw ValueException("comparison can only be done between two numeric values")
-    }
-
-    private fun arithResult(leftDt: DataType, result: Number, rightDt: DataType, op: String): Value {
-        if(leftDt!=rightDt)
-            throw ValueException("left and right datatypes are not the same")
-        if(result.toDouble() < 0 ) {
-            return when(leftDt) {
-                DataType.UBYTE, DataType.UWORD -> {
-                    // storing a negative number in an unsigned one is done by storing the 2's complement instead
-                    val number = abs(result.toDouble().toInt())
-                    if(leftDt==DataType.UBYTE)
-                        Value(DataType.UBYTE, (number xor 255) + 1)
-                    else
-                        Value(DataType.UBYTE, (number xor 65535) + 1)
-                }
-                DataType.BYTE -> Value(DataType.BYTE, result.toInt())
-                DataType.WORD -> Value(DataType.WORD, result.toInt())
-                DataType.FLOAT -> Value(DataType.FLOAT, result)
-                else -> throw ValueException("$op on non-numeric type")
-            }
-        }
-
-        return when(leftDt) {
-            DataType.UBYTE -> Value(DataType.UBYTE, result.toInt() and 255)
-            DataType.BYTE -> Value(DataType.BYTE, result.toInt())
-            DataType.UWORD -> Value(DataType.UWORD, result.toInt() and 65535)
-            DataType.WORD -> Value(DataType.WORD, result.toInt())
-            DataType.FLOAT -> Value(DataType.FLOAT, result)
-            else -> throw ValueException("$op on non-numeric type")
-        }
-    }
-
-    fun add(other: Value): Value {
-        if(other.type == DataType.FLOAT && (type!= DataType.FLOAT))
-            throw ValueException("floating point loss of precision on type $type")
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        val result = v1.toDouble() + v2.toDouble()
-        return arithResult(type, result, other.type, "add")
-    }
-
-    fun sub(other: Value): Value {
-        if(other.type == DataType.FLOAT && (type!= DataType.FLOAT))
-            throw ValueException("floating point loss of precision on type $type")
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        val result = v1.toDouble() - v2.toDouble()
-        return arithResult(type, result, other.type, "sub")
-    }
-
-    fun mul(other: Value): Value {
-        if(other.type == DataType.FLOAT && (type!= DataType.FLOAT))
-            throw ValueException("floating point loss of precision on type $type")
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        val result = v1.toDouble() * v2.toDouble()
-        return arithResult(type, result, other.type, "mul")
-    }
-
-    fun div(other: Value): Value {
-        if(other.type == DataType.FLOAT && (type!= DataType.FLOAT))
-            throw ValueException("floating point loss of precision on type $type")
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        if(v2.toDouble()==0.0) {
-            when (type) {
-                DataType.UBYTE -> return Value(DataType.UBYTE, 255)
-                DataType.BYTE -> return Value(DataType.BYTE, 127)
-                DataType.UWORD -> return Value(DataType.UWORD, 65535)
-                DataType.WORD -> return Value(DataType.WORD, 32767)
-                else -> {}
-            }
-        }
-        val result = v1.toDouble() / v2.toDouble()
-        // NOTE: integer division returns integer result!
-        return when(type) {
-            DataType.UBYTE -> Value(DataType.UBYTE, result)
-            DataType.BYTE -> Value(DataType.BYTE, result)
-            DataType.UWORD -> Value(DataType.UWORD, result)
-            DataType.WORD -> Value(DataType.WORD, result)
-            DataType.FLOAT -> Value(DataType.FLOAT, result)
-            else -> throw ValueException("div on non-numeric type")
-        }
-    }
-
-    fun remainder(other: Value): Value? {
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        val result = v1.toDouble() % v2.toDouble()
-        return arithResult(type, result, other.type, "remainder")
-    }
-
-    fun pow(other: Value): Value {
-        val v1 = numericValue()
-        val v2 = other.numericValue()
-        val result = v1.toDouble().pow(v2.toDouble())
-        return arithResult(type, result, other.type,"pow")
-    }
-
-    fun shl(): Value {
-        val v = integerValue()
-        return when (type) {
-            DataType.UBYTE -> return Value(type, (v shl 1) and 255)
-            DataType.BYTE -> {
-                if(v<0)
-                    Value(type, -((-v shl 1) and 255))
-                else
-                    Value(type, (v shl 1) and 255)
-            }
-            DataType.UWORD -> return Value(type, (v shl 1) and 65535)
-            DataType.WORD -> {
-                if(v<0)
-                    Value(type, -((-v shl 1) and 65535))
-                else
-                    Value(type, (v shl 1) and 65535)
-            }
-            else -> throw ValueException("invalid type for shl: $type")
-        }
-    }
-
-    fun shr(): Value {
-        val v = integerValue()
-        return when(type){
-            DataType.UBYTE -> Value(type, (v ushr 1) and 255)
-            DataType.BYTE -> Value(type, v shr 1)
-            DataType.UWORD -> Value(type, (v ushr 1) and 65535)
-            DataType.WORD -> Value(type, v shr 1)
-            else -> throw ValueException("invalid type for shr: $type")
-        }
-    }
-
-    fun rol(carry: Boolean): Pair<Value, Boolean> {
-        // 9 or 17 bit rotate left (with carry))
-        return when(type) {
-            DataType.UBYTE -> {
-                val v = byteval!!.toInt()
-                val newCarry = (v and 0x80) != 0
-                val newval = (v and 0x7f shl 1) or (if(carry) 1 else 0)
-                Pair(Value(DataType.UBYTE, newval), newCarry)
-            }
-            DataType.UWORD -> {
-                val v = wordval!!
-                val newCarry = (v and 0x8000) != 0
-                val newval = (v and 0x7fff shl 1) or (if(carry) 1 else 0)
-                Pair(Value(DataType.UWORD, newval), newCarry)
-            }
-            else -> throw ValueException("rol can only work on byte/word")
-        }
-    }
-
-    fun ror(carry: Boolean): Pair<Value, Boolean> {
-        // 9 or 17 bit rotate right (with carry)
-        return when(type) {
-            DataType.UBYTE -> {
-                val v = byteval!!.toInt()
-                val newCarry = v and 1 != 0
-                val newval = (v ushr 1) or (if(carry) 0x80 else 0)
-                Pair(Value(DataType.UBYTE, newval), newCarry)
-            }
-            DataType.UWORD -> {
-                val v = wordval!!
-                val newCarry = v and 1 != 0
-                val newval = (v ushr 1) or (if(carry) 0x8000 else 0)
-                Pair(Value(DataType.UWORD, newval), newCarry)
-            }
-            else -> throw ValueException("ror2 can only work on byte/word")
-        }
-    }
-
-    fun rol2(): Value {
-        // 8 or 16 bit rotate left
-        return when(type) {
-            DataType.UBYTE -> {
-                val v = byteval!!.toInt()
-                val carry = (v and 0x80) ushr 7
-                val newval = (v and 0x7f shl 1) or carry
-                Value(DataType.UBYTE, newval)
-            }
-            DataType.UWORD -> {
-                val v = wordval!!
-                val carry = (v and 0x8000) ushr 15
-                val newval = (v and 0x7fff shl 1) or carry
-                Value(DataType.UWORD, newval)
-            }
-            else -> throw ValueException("rol2 can only work on byte/word")
-        }
-    }
-
-    fun ror2(): Value {
-        // 8 or 16 bit rotate right
-        return when(type) {
-            DataType.UBYTE -> {
-                val v = byteval!!.toInt()
-                val carry = v and 1 shl 7
-                val newval = (v ushr 1) or carry
-                Value(DataType.UBYTE, newval)
-            }
-            DataType.UWORD -> {
-                val v = wordval!!
-                val carry = v and 1 shl 15
-                val newval = (v ushr 1) or carry
-                Value(DataType.UWORD, newval)
-            }
-            else -> throw ValueException("ror2 can only work on byte/word")
-        }
-    }
-
-    fun neg(): Value {
-        return when(type) {
-            DataType.BYTE -> Value(DataType.BYTE, -(byteval!!))
-            DataType.WORD -> Value(DataType.WORD, -(wordval!!))
-            DataType.FLOAT -> Value(DataType.FLOAT, -(floatval)!!)
-            else -> throw ValueException("neg can only work on byte/word/float")
-        }
-    }
-
-    fun abs(): Value {
-        return when(type) {
-            DataType.BYTE -> Value(DataType.BYTE, abs(byteval!!.toInt()))
-            DataType.WORD -> Value(DataType.WORD, abs(wordval!!))
-            DataType.FLOAT -> Value(DataType.FLOAT, abs(floatval!!))
-            else -> throw ValueException("abs can only work on byte/word/float")
-        }
-    }
-
-    fun bitand(other: Value): Value {
-        val v1 = integerValue()
-        val v2 = other.integerValue()
-        val result = v1 and v2
-        return Value(type, result)
-    }
-
-    fun bitor(other: Value): Value {
-        val v1 = integerValue()
-        val v2 = other.integerValue()
-        val result = v1 or v2
-        return Value(type, result)
-    }
-
-    fun bitxor(other: Value): Value {
-        val v1 = integerValue()
-        val v2 = other.integerValue()
-        val result = v1 xor v2
-        return Value(type, result)
-    }
-
-    fun and(other: Value) = Value(DataType.UBYTE, if (this.asBooleanValue && other.asBooleanValue) 1 else 0)
-    fun or(other: Value) = Value(DataType.UBYTE, if (this.asBooleanValue || other.asBooleanValue) 1 else 0)
-    fun xor(other: Value) = Value(DataType.UBYTE, if (this.asBooleanValue xor other.asBooleanValue) 1 else 0)
-    fun not() = Value(DataType.UBYTE, if (this.asBooleanValue) 0 else 1)
-
-    fun inv(): Value {
-        return when(type) {
-            DataType.UBYTE -> Value(DataType.UBYTE, byteval!!.toInt().inv() and 255)
-            DataType.UWORD -> Value(DataType.UWORD, wordval!!.inv() and 65535)
-            else -> throw ValueException("inv can only work on byte/word")
-        }
-    }
-
-    fun inc(): Value {
-        return when(type) {
-            DataType.UBYTE -> Value(DataType.UBYTE, (byteval!! + 1) and 255)
-            DataType.UWORD -> Value(DataType.UWORD, (wordval!! + 1) and 65535)
-            DataType.FLOAT -> Value(DataType.FLOAT, floatval!! + 1)
-            else -> throw ValueException("inc can only work on byte/word/float")
-        }
-    }
-
-    fun dec(): Value {
-        return when(type) {
-            DataType.UBYTE -> Value(DataType.UBYTE, (byteval!! - 1) and 255)
-            DataType.UWORD -> Value(DataType.UWORD, (wordval!! - 1) and 65535)
-            DataType.FLOAT -> Value(DataType.FLOAT, floatval!! - 1)
-            else -> throw ValueException("dec can only work on byte/word/float")
-        }
-    }
-
-    fun msb(): Value {
-        return when(type) {
-            DataType.UBYTE, DataType.BYTE -> Value(DataType.UBYTE, 0)
-            DataType.UWORD, DataType.WORD -> Value(DataType.UBYTE, wordval!! ushr 8 and 255)
-            else -> throw ValueException("msb can only work on (u)byte/(u)word")
-        }
-    }
-
-    fun cast(targetType: DataType): Value {
-        return when (type) {
-            DataType.UBYTE -> {
-                when (targetType) {
-                    DataType.UBYTE -> this
-                    DataType.BYTE -> {
-                        if(byteval!!<=127)
-                            Value(DataType.BYTE, byteval!!)
-                        else
-                            Value(DataType.BYTE, -(256-byteval!!))
-                    }
-                    DataType.UWORD -> Value(DataType.UWORD, numericValue())
-                    DataType.WORD -> Value(DataType.WORD, numericValue())
-                    DataType.FLOAT -> Value(DataType.FLOAT, numericValue())
-                    else -> throw ValueException("invalid type cast from $type to $targetType")
-                }
-            }
-            DataType.BYTE -> {
-                when (targetType) {
-                    DataType.BYTE -> this
-                    DataType.UBYTE -> Value(DataType.UBYTE, integerValue() and 255)
-                    DataType.UWORD -> Value(DataType.UWORD, integerValue() and 65535)
-                    DataType.WORD -> Value(DataType.WORD, integerValue())
-                    DataType.FLOAT -> Value(DataType.FLOAT, numericValue())
-                    else -> throw ValueException("invalid type cast from $type to $targetType")
-                }
-            }
-            DataType.UWORD -> {
-                when (targetType) {
-                    DataType.BYTE, DataType.UBYTE -> Value(DataType.UBYTE, integerValue() and 255)
-                    DataType.UWORD -> this
-                    DataType.WORD -> {
-                        if(integerValue()<=32767)
-                            Value(DataType.WORD, integerValue())
-                        else
-                            Value(DataType.WORD, -(65536-integerValue()))
-                    }
-                    DataType.FLOAT -> Value(DataType.FLOAT, numericValue())
-                    else -> throw ValueException("invalid type cast from $type to $targetType")
-                }
-            }
-            DataType.WORD -> {
-                when (targetType) {
-                    DataType.BYTE, DataType.UBYTE -> Value(DataType.UBYTE, integerValue() and 255)
-                    DataType.UWORD -> Value(DataType.UWORD, integerValue() and 65535)
-                    DataType.WORD -> this
-                    DataType.FLOAT -> Value(DataType.FLOAT, numericValue())
-                    else -> throw ValueException("invalid type cast from $type to $targetType")
-                }
-            }
-            DataType.FLOAT -> {
-                when (targetType) {
-                    DataType.BYTE -> {
-                        val integer=numericValue().toInt()
-                        if(integer in -128..127)
-                            Value(DataType.BYTE, integer)
-                        else
-                            throw ValueException("overflow when casting float to byte: $this")
-                    }
-                    DataType.UBYTE -> Value(DataType.UBYTE, numericValue().toInt() and 255)
-                    DataType.UWORD -> Value(DataType.UWORD, numericValue().toInt() and 65535)
-                    DataType.WORD -> {
-                        val integer=numericValue().toInt()
-                        if(integer in -32768..32767)
-                            Value(DataType.WORD, integer)
-                        else
-                            throw ValueException("overflow when casting float to word: $this")
-                    }
-                    DataType.FLOAT -> this
-                    else -> throw ValueException("invalid type cast from $type to $targetType")
-                }
-            }
-            else -> throw ValueException("invalid type cast from $type to $targetType")
-        }
-    }
-
-}

compiler/src/prog8/compiler/target/IAssemblyGenerator.kt

@@ -0,0 +1,16 @@
+package prog8.compiler.target
+
+import prog8.compiler.CompilationOptions
+
+internal interface IAssemblyGenerator {
+    fun compileToAssembly(): IAssemblyProgram
+}
+
+internal const val generatedLabelPrefix = "_prog8_label_"
+internal const val subroutineFloatEvalResultVar1 = "_prog8_float_eval_result1"
+internal const val subroutineFloatEvalResultVar2 = "_prog8_float_eval_result2"
+
+internal interface IAssemblyProgram {
+    val name: String
+    fun assemble(options: CompilationOptions)
+}

compiler/src/prog8/compiler/target/ICompilationTarget.kt

@@ -0,0 +1,119 @@
+package prog8.compiler.target
+
+import prog8.ast.IMemSizer
+import prog8.ast.IStringEncoding
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.AssignTarget
+import prog8.compiler.CompilationOptions
+import prog8.compiler.IErrorReporter
+import prog8.compiler.Zeropage
+import prog8.compiler.target.c64.C64MachineDefinition
+import prog8.compiler.target.c64.Petscii
+import prog8.compiler.target.cpu6502.codegen.AsmGen
+import prog8.compiler.target.cx16.CX16MachineDefinition
+import java.nio.file.Path
+
+
+interface ICompilationTarget: IStringEncoding, IMemSizer {
+    val name: String
+    val machine: IMachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean): List<Short>
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean): String
+
+    fun isInRegularRAM(target: AssignTarget, program: Program): Boolean {
+        val memAddr = target.memoryAddress
+        val arrayIdx = target.arrayindexed
+        val ident = target.identifier
+        when {
+            memAddr != null -> {
+                return when (memAddr.addressExpression) {
+                    is NumericLiteralValue -> {
+                        machine.isRegularRAMaddress((memAddr.addressExpression as NumericLiteralValue).number.toInt())
+                    }
+                    is IdentifierReference -> {
+                        val decl = (memAddr.addressExpression as IdentifierReference).targetVarDecl(program)
+                        if ((decl?.type == VarDeclType.VAR || decl?.type == VarDeclType.CONST) && decl.value is NumericLiteralValue)
+                            machine.isRegularRAMaddress((decl.value as NumericLiteralValue).number.toInt())
+                        else
+                            false
+                    }
+                    else -> false
+                }
+            }
+            arrayIdx != null -> {
+                val targetStmt = arrayIdx.arrayvar.targetVarDecl(program)
+                return if (targetStmt?.type == VarDeclType.MEMORY) {
+                    val addr = targetStmt.value as? NumericLiteralValue
+                    if (addr != null)
+                        machine.isRegularRAMaddress(addr.number.toInt())
+                    else
+                        false
+                } else true
+            }
+            ident != null -> {
+                val decl = ident.targetVarDecl(program)!!
+                return if (decl.type == VarDeclType.MEMORY && decl.value is NumericLiteralValue)
+                    machine.isRegularRAMaddress((decl.value as NumericLiteralValue).number.toInt())
+                else
+                    true
+            }
+            else -> return true
+        }
+    }
+}
+
+
+internal object C64Target: ICompilationTarget {
+    override val name = "c64"
+    override val machine = C64MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean) =
+            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
+            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            in PassByReferenceDatatypes -> machine.POINTER_MEM_SIZE
+            else -> -9999999
+        }
+    }
+}
+
+internal object Cx16Target: ICompilationTarget {
+    override val name = "cx16"
+    override val machine = CX16MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean) =
+            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
+            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            in PassByReferenceDatatypes -> machine.POINTER_MEM_SIZE
+            else -> -9999999
+        }
+    }
+}
+
+
+internal fun asmGeneratorFor(
+    compTarget: ICompilationTarget,
+    program: Program,
+    errors: IErrorReporter,
+    zp: Zeropage,
+    options: CompilationOptions,
+    outputDir: Path
+): IAssemblyGenerator
+{
+    // at the moment we only have one code generation backend (for 6502 and 65c02)
+    return AsmGen(program, errors, zp, options, compTarget, outputDir)
+}

compiler/src/prog8/compiler/target/IMachineDefinition.kt

@@ -0,0 +1,37 @@
+package prog8.compiler.target
+
+import prog8.compiler.CompilationOptions
+import prog8.compiler.Zeropage
+
+
+interface IMachineFloat {
+    fun toDouble(): Double
+    fun makeFloatFillAsm(): String
+}
+
+enum class CpuType {
+    CPU6502,
+    CPU65c02
+}
+
+interface IMachineDefinition {
+    val FLOAT_MAX_NEGATIVE: Double
+    val FLOAT_MAX_POSITIVE: Double
+    val FLOAT_MEM_SIZE: Int
+    val POINTER_MEM_SIZE: Int
+    val ESTACK_LO: Int
+    val ESTACK_HI: Int
+    val BASIC_LOAD_ADDRESS : Int
+    val RAW_LOAD_ADDRESS : Int
+
+    val opcodeNames: Set<String>
+    var zeropage: Zeropage
+    val cpu: CpuType
+
+    fun initializeZeropage(compilerOptions: CompilationOptions)
+    fun getFloat(num: Number): IMachineFloat
+
+    fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String>
+    fun launchEmulator(programName: String)
+    fun isRegularRAMaddress(address: Int): Boolean
+}

compiler/src/prog8/compiler/target/c64/AsmOptimizer.kt

@@ -1,176 +0,0 @@
-package prog8.compiler.target.c64
-
-import prog8.compiler.toHex
-
-fun optimizeAssembly(lines: MutableList<String>): Int {
-
-    var numberOfOptimizations = 0
-
-    var linesByFour = getLinesBy(lines, 4)
-
-    var removeLines = optimizeUselessStackByteWrites(linesByFour)
-    if(removeLines.isNotEmpty()) {
-        for (i in removeLines.reversed())
-            lines.removeAt(i)
-        linesByFour = getLinesBy(lines, 4)
-        numberOfOptimizations++
-    }
-
-    removeLines = optimizeIncDec(linesByFour)
-    if(removeLines.isNotEmpty()) {
-        for (i in removeLines.reversed())
-            lines.removeAt(i)
-        linesByFour = getLinesBy(lines, 4)
-        numberOfOptimizations++
-    }
-
-    removeLines = optimizeStoreLoadSame(linesByFour)
-    if(removeLines.isNotEmpty()) {
-        for (i in removeLines.reversed())
-            lines.removeAt(i)
-        numberOfOptimizations++
-    }
-
-    var linesByFourteen = getLinesBy(lines, 14)
-    removeLines = optimizeSameAssignments(linesByFourteen)
-    if(removeLines.isNotEmpty()) {
-        for (i in removeLines.reversed())
-            lines.removeAt(i)
-        numberOfOptimizations++
-    }
-
-    // TODO more assembly optimizations?
-
-    return numberOfOptimizations
-}
-
-fun optimizeUselessStackByteWrites(linesByFour: List<List<IndexedValue<String>>>): List<Int> {
-    // sta on stack, dex, inx, lda from stack -> eliminate this useless stack byte write
-    // this is a lot harder for word values because the instruction sequence varies.
-    val removeLines = mutableListOf<Int>()
-    for(lines in linesByFour) {
-        if(lines[0].value.trim()=="sta  ${ESTACK_LO.toHex()},x" &&
-                lines[1].value.trim()=="dex" &&
-                lines[2].value.trim()=="inx" &&
-                lines[3].value.trim()=="lda  ${ESTACK_LO.toHex()},x") {
-            removeLines.add(lines[0].index)
-            removeLines.add(lines[1].index)
-            removeLines.add(lines[2].index)
-            removeLines.add(lines[3].index)
-        }
-    }
-    return removeLines
-}
-
-fun optimizeSameAssignments(linesByFourteen: List<List<IndexedValue<String>>>): List<Int> {
-
-    // optimize sequential assignments of the same value to various targets (bytes, words, floats)
-    // the float one is the one that requires 2*7=14 lines of code to check...
-    // @todo a better place to do this is in the Compiler instead and work on opcodes, and never even create the inefficient asm...
-
-    val removeLines = mutableListOf<Int>()
-    for (pair in linesByFourteen) {
-        val first = pair[0].value.trimStart()
-        val second = pair[1].value.trimStart()
-        val third = pair[2].value.trimStart()
-        val fourth = pair[3].value.trimStart()
-        val fifth = pair[4].value.trimStart()
-        val sixth = pair[5].value.trimStart()
-        val seventh = pair[6].value.trimStart()
-        val eighth = pair[7].value.trimStart()
-
-        if(first.startsWith("lda") && second.startsWith("ldy") && third.startsWith("sta") && fourth.startsWith("sty") &&
-                fifth.startsWith("lda") && sixth.startsWith("ldy") && seventh.startsWith("sta") && eighth.startsWith("sty")) {
-            val firstvalue = first.substring(4)
-            val secondvalue = second.substring(4)
-            val thirdvalue = fifth.substring(4)
-            val fourthvalue = sixth.substring(4)
-            if(firstvalue==thirdvalue && secondvalue==fourthvalue) {
-                // lda/ldy   sta/sty   twice the same word -->  remove second lda/ldy pair (fifth and sixth lines)
-                removeLines.add(pair[4].index)
-                removeLines.add(pair[5].index)
-            }
-        }
-
-        if(first.startsWith("lda") && second.startsWith("sta") && third.startsWith("lda") && fourth.startsWith("sta")) {
-            val firstvalue = first.substring(4)
-            val secondvalue = third.substring(4)
-            if(firstvalue==secondvalue) {
-                // lda value / sta ? / lda same-value / sta ?  -> remove second lda (third line)
-                removeLines.add(pair[2].index)
-            }
-        }
-
-        if(first.startsWith("lda") && second.startsWith("ldy") && third.startsWith("sta") && fourth.startsWith("sty") &&
-                fifth.startsWith("lda") && sixth.startsWith("ldy") && seventh.startsWith("jsr  c64flt.copy_float")) {
-
-            val nineth = pair[8].value.trimStart()
-            val tenth = pair[9].value.trimStart()
-            val eleventh = pair[10].value.trimStart()
-            val twelveth = pair[11].value.trimStart()
-            val thirteenth = pair[12].value.trimStart()
-            val fourteenth = pair[13].value.trimStart()
-
-            if(eighth.startsWith("lda") && nineth.startsWith("ldy") && tenth.startsWith("sta") && eleventh.startsWith("sty") &&
-                    twelveth.startsWith("lda") && thirteenth.startsWith("ldy") && fourteenth.startsWith("jsr  c64flt.copy_float")) {
-
-                if(first.substring(4) == eighth.substring(4) && second.substring(4)==nineth.substring(4)) {
-                    // identical float init
-                    removeLines.add(pair[7].index)
-                    removeLines.add(pair[8].index)
-                    removeLines.add(pair[9].index)
-                    removeLines.add(pair[10].index)
-                }
-            }
-        }
-    }
-    return removeLines
-}
-
-private fun getLinesBy(lines: MutableList<String>, windowSize: Int) =
-// all lines (that aren't empty or comments) in sliding pairs of 2
-        lines.withIndex().filter { it.value.isNotBlank() && !it.value.trimStart().startsWith(';') }.windowed(windowSize, partialWindows = false)
-
-private fun optimizeStoreLoadSame(linesByFour: List<List<IndexedValue<String>>>): List<Int> {
-    // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can be eliminated
-    val removeLines = mutableListOf<Int>()
-    for (pair in linesByFour) {
-        val first = pair[0].value.trimStart()
-        val second = pair[1].value.trimStart()
-
-        if ((first.startsWith("sta ") && second.startsWith("lda ")) ||
-                (first.startsWith("stx ") && second.startsWith("ldx ")) ||
-                (first.startsWith("sty ") && second.startsWith("ldy ")) ||
-                (first.startsWith("lda ") && second.startsWith("lda ")) ||
-                (first.startsWith("ldy ") && second.startsWith("ldy ")) ||
-                (first.startsWith("ldx ") && second.startsWith("ldx ")) ||
-                (first.startsWith("sta ") && second.startsWith("lda ")) ||
-                (first.startsWith("sty ") && second.startsWith("ldy ")) ||
-                (first.startsWith("stx ") && second.startsWith("ldx "))
-        ) {
-            val firstLoc = first.substring(4)
-            val secondLoc = second.substring(4)
-            if (firstLoc == secondLoc) {
-                removeLines.add(pair[1].index)
-            }
-        }
-    }
-    return removeLines
-}
-
-private fun optimizeIncDec(linesByTwo: List<List<IndexedValue<String>>>): List<Int> {
-    // sometimes, iny+dey / inx+dex / dey+iny / dex+inx sequences are generated, these can be eliminated.
-    val removeLines = mutableListOf<Int>()
-    for (pair in linesByTwo) {
-        val first = pair[0].value
-        val second = pair[1].value
-        if ((" iny" in first || "\tiny" in first) && (" dey" in second || "\tdey" in second)
-                || (" inx" in first || "\tinx" in first) && (" dex" in second || "\tdex" in second)
-                || (" dey" in first || "\tdey" in first) && (" iny" in second || "\tiny" in second)
-                || (" dex" in first || "\tdex" in first) && (" inx" in second || "\tinx" in second)) {
-            removeLines.add(pair[0].index)
-            removeLines.add(pair[1].index)
-        }
-    }
-    return removeLines
-}

compiler/src/prog8/compiler/target/c64/AssemblyProgram.kt

@@ -2,55 +2,72 @@ package prog8.compiler.target.c64
 
 import prog8.compiler.CompilationOptions
 import prog8.compiler.OutputType
-import java.io.File
+import prog8.compiler.target.IAssemblyProgram
+import prog8.compiler.target.generatedLabelPrefix
+import java.nio.file.Path
 import kotlin.system.exitProcess
 
-class AssemblyProgram(val name: String) {
-    private val assemblyFile = "$name.asm"
-    private val viceMonListFile = "$name.vice-mon-list"
+class AssemblyProgram(override val name: String, outputDir: Path, private val compTarget: String) : IAssemblyProgram {
+    private val assemblyFile = outputDir.resolve("$name.asm")
+    private val prgFile = outputDir.resolve("$name.prg")
+    private val binFile = outputDir.resolve("$name.bin")
+    private val viceMonListFile = outputDir.resolve("$name.vice-mon-list")
 
-    fun assemble(options: CompilationOptions) {
-        // add "-Wlong-branch"  to see warnings about conversion of branch instructions to jumps
-        val command = mutableListOf("64tass", "--ascii", "--case-sensitive", "--long-branch", "-Wall", "-Wno-strict-bool",
-                "-Werror", "-Wno-error=long-branch", "--dump-labels", "--vice-labels", "-l", viceMonListFile, "--no-monitor")
+    override fun assemble(options: CompilationOptions) {
+        // add "-Wlong-branch"  to see warnings about conversion of branch instructions to jumps (default = do this silently)
+        val command = mutableListOf("64tass", "--ascii", "--case-sensitive", "--long-branch",
+                "-Wall", "-Wno-strict-bool", "-Wno-shadow", // "-Werror",
+                "--dump-labels", "--vice-labels", "-l", viceMonListFile.toString(), "--no-monitor")
 
-        val outFile = when(options.output) {
+        val outFile = when (options.output) {
             OutputType.PRG -> {
                 command.add("--cbm-prg")
-                println("\nCreating C-64 prg.")
-                "$name.prg"
+                println("\nCreating prg for target $compTarget.")
+                prgFile
             }
             OutputType.RAW -> {
                 command.add("--nostart")
-                println("\nCreating raw binary.")
-                "$name.bin"
+                println("\nCreating raw binary for target $compTarget.")
+                binFile
             }
         }
-        command.addAll(listOf("--output", outFile, assemblyFile))
+        command.addAll(listOf("--output", outFile.toString(), assemblyFile.toString()))
 
         val proc = ProcessBuilder(command).inheritIO().start()
         val result = proc.waitFor()
-        if(result!=0) {
+        if (result != 0) {
             System.err.println("assembler failed with returncode $result")
             exitProcess(result)
         }
 
+        removeGeneratedLabelsFromMonlist()
         generateBreakpointList()
     }
 
+    private fun removeGeneratedLabelsFromMonlist() {
+        val pattern = Regex("""al (\w+) \S+${generatedLabelPrefix}.+?""")
+        val lines = viceMonListFile.toFile().readLines()
+        viceMonListFile.toFile().outputStream().bufferedWriter().use {
+            for (line in lines) {
+                if(pattern.matchEntire(line)==null)
+                    it.write(line+"\n")
+            }
+        }
+    }
+
     private fun generateBreakpointList() {
         // builds list of breakpoints, appends to monitor list file
         val breakpoints = mutableListOf<String>()
         val pattern = Regex("""al (\w+) \S+_prog8_breakpoint_\d+.?""")      // gather breakpoints by the source label that's generated for them
-        for(line in File(viceMonListFile).readLines()) {
+        for (line in viceMonListFile.toFile().readLines()) {
             val match = pattern.matchEntire(line)
-            if(match!=null)
-            breakpoints.add("break \$" + match.groupValues[1])
+            if (match != null)
+                breakpoints.add("break \$" + match.groupValues[1])
         }
         val num = breakpoints.size
         breakpoints.add(0, "; vice monitor breakpoint list now follows")
         breakpoints.add(1, "; $num breakpoints have been defined")
         breakpoints.add(2, "del")
-        File(viceMonListFile).appendText(breakpoints.joinToString("\n")+"\n")
+        viceMonListFile.toFile().appendText(breakpoints.joinToString("\n") + "\n")
     }
 }

compiler/src/prog8/compiler/target/c64/C64MachineDefinition.kt

@@ -0,0 +1,207 @@
+package prog8.compiler.target.c64
+
+import prog8.compiler.*
+import prog8.compiler.target.CpuType
+import prog8.compiler.target.IMachineDefinition
+import prog8.compiler.target.IMachineFloat
+import java.io.IOException
+import kotlin.math.absoluteValue
+import kotlin.math.pow
+
+internal object C64MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU6502
+
+    // 5-byte cbm MFLPT format limitations:
+    override val FLOAT_MAX_POSITIVE = 1.7014118345e+38         // bytes: 255,127,255,255,255
+    override val FLOAT_MAX_NEGATIVE = -1.7014118345e+38        // bytes: 255,255,255,255,255
+    override val FLOAT_MEM_SIZE = 5
+    override val POINTER_MEM_SIZE = 2
+    override val BASIC_LOAD_ADDRESS = 0x0801
+    override val RAW_LOAD_ADDRESS = 0xc000
+
+    // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
+    override val ESTACK_LO = 0xce00        //  $ce00-$ceff inclusive
+    override val ESTACK_HI = 0xcf00        //  $ce00-$ceff inclusive
+
+    override lateinit var zeropage: Zeropage
+
+    override fun getFloat(num: Number) = Mflpt5.fromNumber(num)
+
+    override fun importLibs(compilerOptions: CompilationOptions,compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
+    }
+
+    override fun launchEmulator(programName: String) {
+        for(emulator in listOf("x64sc", "x64")) {
+            println("\nStarting C-64 emulator $emulator...")
+            val cmdline = listOf(emulator, "-silent", "-moncommands", "$programName.vice-mon-list",
+                    "-autostartprgmode", "1", "-autostart-warp", "-autostart", programName + ".prg")
+            val processb = ProcessBuilder(cmdline).inheritIO()
+            val process: Process
+            try {
+                process=processb.start()
+            } catch(x: IOException) {
+                continue  // try the next emulator executable
+            }
+            process.waitFor()
+            break
+        }
+    }
+
+    override fun isRegularRAMaddress(address: Int): Boolean = address<0xa000 || address in 0xc000..0xcfff
+
+    override fun initializeZeropage(compilerOptions: CompilationOptions) {
+        zeropage = C64Zeropage(compilerOptions)
+    }
+
+    // 6502 opcodes (including aliases and illegal opcodes), these cannot be used as variable or label names
+    override val opcodeNames = setOf("adc", "ahx", "alr", "anc", "and", "ane", "arr", "asl", "asr", "axs", "bcc", "bcs",
+            "beq", "bge", "bit", "blt", "bmi", "bne", "bpl", "brk", "bvc", "bvs", "clc",
+            "cld", "cli", "clv", "cmp", "cpx", "cpy", "dcm", "dcp", "dec", "dex", "dey",
+            "eor", "gcc", "gcs", "geq", "gge", "glt", "gmi", "gne", "gpl", "gvc", "gvs",
+            "inc", "ins", "inx", "iny", "isb", "isc", "jam", "jmp", "jsr", "lae", "las",
+            "lax", "lda", "lds", "ldx", "ldy", "lsr", "lxa", "nop", "ora", "pha", "php",
+            "pla", "plp", "rla", "rol", "ror", "rra", "rti", "rts", "sax", "sbc", "sbx",
+            "sec", "sed", "sei", "sha", "shl", "shr", "shs", "shx", "shy", "slo", "sre",
+            "sta", "stx", "sty", "tas", "tax", "tay", "tsx", "txa", "txs", "tya", "xaa")
+
+
+    internal class C64Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+        override val SCRATCH_B1 = 0x02      // temp storage for a single byte
+        override val SCRATCH_REG = 0x03     // temp storage for a register, must be B1+1
+        override val SCRATCH_W1 = 0xfb      // temp storage 1 for a word  $fb+$fc
+        override val SCRATCH_W2 = 0xfd      // temp storage 2 for a word  $fb+$fc
+
+
+        init {
+            if (options.floats && options.zeropage !in setOf(ZeropageType.FLOATSAFE, ZeropageType.BASICSAFE, ZeropageType.DONTUSE ))
+                throw CompilerException("when floats are enabled, zero page type should be 'floatsafe' or 'basicsafe' or 'dontuse'")
+
+            if (options.zeropage == ZeropageType.FULL) {
+                free.addAll(0x04..0xf9)
+                free.add(0xff)
+                free.removeAll(listOf(SCRATCH_B1, SCRATCH_REG, SCRATCH_W1, SCRATCH_W1 + 1, SCRATCH_W2, SCRATCH_W2 + 1))
+                free.removeAll(listOf(0xa0, 0xa1, 0xa2, 0x91, 0xc0, 0xc5, 0xcb, 0xf5, 0xf6))        // these are updated by IRQ
+            } else {
+                if (options.zeropage == ZeropageType.KERNALSAFE || options.zeropage == ZeropageType.FLOATSAFE) {
+                    free.addAll(listOf(
+                            0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11,
+                            0x16, 0x17, 0x18, 0x19, 0x1a,
+                            0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 0x21,
+                            0x22, 0x23, 0x24, 0x25,
+                            0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46,
+                            0x47, 0x48, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x51, 0x52, 0x53,
+                            0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60,
+                            0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
+                            0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72,
+                            0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c,
+                            0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a,
+                            0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0xff
+                            // 0x90-0xfa is 'kernal work storage area'
+                    ))
+                }
+
+                if (options.zeropage == ZeropageType.FLOATSAFE) {
+                    // remove the zero page locations used for floating point operations from the free list
+                    free.removeAll(listOf(
+                            0x22, 0x23, 0x24, 0x25,
+                            0x10, 0x11, 0x12, 0x26, 0x27, 0x28, 0x29, 0x2a,
+                            0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60,
+                            0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
+                            0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72,
+                            0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0xff
+                    ))
+                }
+
+                if(options.zeropage!=ZeropageType.DONTUSE) {
+                    // add the free Zp addresses
+                    // these are valid for the C-64 but allow BASIC to keep running fully *as long as you don't use tape I/O*
+                    free.addAll(listOf(0x04, 0x05, 0x06, 0x0a, 0x0e,
+                            0x92, 0x96, 0x9b, 0x9c, 0x9e, 0x9f, 0xa5, 0xa6,
+                            0xb0, 0xb1, 0xbe, 0xbf, 0xf9))
+                } else {
+                    // don't use the zeropage at all
+                    free.clear()
+                }
+            }
+            require(SCRATCH_B1 !in free)
+            require(SCRATCH_REG !in free)
+            require(SCRATCH_W1 !in free)
+            require(SCRATCH_W2 !in free)
+
+            for (reserved in options.zpReserved)
+                reserve(reserved)
+        }
+    }
+
+    internal data class Mflpt5(val b0: Short, val b1: Short, val b2: Short, val b3: Short, val b4: Short): IMachineFloat {
+
+        companion object {
+            val zero = Mflpt5(0, 0, 0, 0, 0)
+            fun fromNumber(num: Number): Mflpt5 {
+                // see https://en.wikipedia.org/wiki/Microsoft_Binary_Format
+                // and https://sourceforge.net/p/acme-crossass/code-0/62/tree/trunk/ACME_Lib/cbm/mflpt.a
+                // and https://en.wikipedia.org/wiki/IEEE_754-1985
+
+                val flt = num.toDouble()
+                if (flt < FLOAT_MAX_NEGATIVE || flt > FLOAT_MAX_POSITIVE)
+                    throw CompilerException("floating point number out of 5-byte mflpt range: $this")
+                if (flt == 0.0)
+                    return zero
+
+                val sign = if (flt < 0.0) 0x80L else 0x00L
+                var exponent = 128 + 32    // 128 is cbm's bias, 32 is this algo's bias
+                var mantissa = flt.absoluteValue
+
+                // if mantissa is too large, shift right and adjust exponent
+                while (mantissa >= 0x100000000) {
+                    mantissa /= 2.0
+                    exponent++
+                }
+                // if mantissa is too small, shift left and adjust exponent
+                while (mantissa < 0x80000000) {
+                    mantissa *= 2.0
+                    exponent--
+                }
+
+                return when {
+                    exponent < 0 -> zero  // underflow, use zero instead
+                    exponent > 255 -> throw CompilerException("floating point overflow: $this")
+                    exponent == 0 -> zero
+                    else -> {
+                        val mantLong = mantissa.toLong()
+                        Mflpt5(
+                                exponent.toShort(),
+                                (mantLong.and(0x7f000000L) ushr 24).or(sign).toShort(),
+                                (mantLong.and(0x00ff0000L) ushr 16).toShort(),
+                                (mantLong.and(0x0000ff00L) ushr 8).toShort(),
+                                (mantLong.and(0x000000ffL)).toShort())
+                    }
+                }
+            }
+        }
+
+        override fun toDouble(): Double {
+            if (this == zero) return 0.0
+            val exp = b0 - 128
+            val sign = (b1.toInt() and 0x80) > 0
+            val number = 0x80000000L.or(b1.toLong() shl 24).or(b2.toLong() shl 16).or(b3.toLong() shl 8).or(b4.toLong())
+            val result = number.toDouble() * (2.0).pow(exp) / 0x100000000
+            return if (sign) -result else result
+        }
+
+        override fun makeFloatFillAsm(): String {
+            val b0 = "$" + b0.toString(16).padStart(2, '0')
+            val b1 = "$" + b1.toString(16).padStart(2, '0')
+            val b2 = "$" + b2.toString(16).padStart(2, '0')
+            val b3 = "$" + b3.toString(16).padStart(2, '0')
+            val b4 = "$" + b4.toString(16).padStart(2, '0')
+            return "$b0, $b1, $b2, $b3, $b4"
+        }
+    }
+}

compiler/src/prog8/compiler/target/c64/Commodore64.kt

@@ -1,186 +0,0 @@
-package prog8.compiler.target.c64
-
-import prog8.compiler.CompilationOptions
-import prog8.compiler.CompilerException
-import prog8.compiler.Zeropage
-import prog8.compiler.ZeropageType
-import java.awt.Color
-import java.awt.image.BufferedImage
-import javax.imageio.ImageIO
-import kotlin.math.absoluteValue
-import kotlin.math.pow
-
-
-// 5-byte cbm MFLPT format limitations:
-const val FLOAT_MAX_POSITIVE = 1.7014118345e+38         // bytes: 255,127,255,255,255
-const val FLOAT_MAX_NEGATIVE = -1.7014118345e+38        // bytes: 255,255,255,255,255
-
-const val BASIC_LOAD_ADDRESS = 0x0801
-const val RAW_LOAD_ADDRESS = 0xc000
-
-// the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
-const val ESTACK_LO = 0xce00        //  $ce00-$ceff inclusive
-const val ESTACK_HI = 0xcf00        //  $cf00-$cfff inclusive
-
-
-class C64Zeropage(options: CompilationOptions) : Zeropage(options) {
-
-    companion object {
-        const val SCRATCH_B1 = 0x02
-        const val SCRATCH_REG = 0x03    // temp storage for a register
-        const val SCRATCH_REG_X = 0xfa    // temp storage for register X (the evaluation stack pointer)
-        const val SCRATCH_W1 = 0xfb     // $fb+$fc
-        const val SCRATCH_W2 = 0xfd     // $fd+$fe
-    }
-
-    override val exitProgramStrategy: ExitProgramStrategy = when(options.zeropage) {
-        ZeropageType.BASICSAFE -> ExitProgramStrategy.CLEAN_EXIT
-        ZeropageType.FLOATSAFE, ZeropageType.KERNALSAFE, ZeropageType.FULL -> ExitProgramStrategy.SYSTEM_RESET
-    }
-
-
-    init {
-        if(options.floats && options.zeropage!=ZeropageType.FLOATSAFE && options.zeropage!=ZeropageType.BASICSAFE)
-            throw CompilerException("when floats are enabled, zero page type should be 'floatsafe' or 'basicsafe'")
-
-        if(options.zeropage == ZeropageType.FULL) {
-            free.addAll(0x04 .. 0xf9)
-            free.add(0xff)
-            free.removeAll(listOf(SCRATCH_B1, SCRATCH_REG, SCRATCH_REG_X, SCRATCH_W1, SCRATCH_W1+1, SCRATCH_W2, SCRATCH_W2+1))
-            free.removeAll(listOf(0xa0, 0xa1, 0xa2, 0x91, 0xc0, 0xc5, 0xcb, 0xf5, 0xf6))        // these are updated by IRQ
-        } else {
-            if(options.zeropage == ZeropageType.KERNALSAFE || options.zeropage == ZeropageType.FLOATSAFE) {
-                free.addAll(listOf(0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10, 0x11,
-                        0x14, 0x15, 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, 0x20, 0x21,
-                        0x22, 0x23, 0x24, 0x25,
-                        0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f, 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46,
-                        0x47, 0x48, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f, 0x51, 0x52, 0x53,
-                        0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60,
-                        0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-                        0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72,
-                        0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c,
-                        0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a,
-                        0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0xff
-                        // 0x90-0xfa is 'kernel work storage area'
-                ))
-            }
-
-            if(options.zeropage == ZeropageType.FLOATSAFE) {
-                // remove the zero page locations used for floating point operations from the free list
-                free.removeAll(listOf(
-                        0x12, 0x26, 0x27, 0x28, 0x29, 0x2a,
-                        0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60,
-                        0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,
-                        0x69, 0x6a, 0x6b, 0x6c, 0x6d, 0x6e, 0x6f, 0x70, 0x71, 0x72,
-                        0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0xf
-                ))
-            }
-
-            // add the other free Zp addresses,
-            // these are valid for the C-64 (when no RS232 I/O is performed) but to keep BASIC running fully:
-            free.addAll(listOf(0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0d, 0x0e,
-                    0x94, 0x95, 0xa7, 0xa8, 0xa9, 0xaa,
-                    0xb5, 0xb6, 0xf7, 0xf8, 0xf9))
-        }
-        assert(SCRATCH_B1 !in free)
-        assert(SCRATCH_REG !in free)
-        assert(SCRATCH_REG_X !in free)
-        assert(SCRATCH_W1 !in free)
-        assert(SCRATCH_W2 !in free)
-
-        for(reserved in options.zpReserved)
-            reserve(reserved)
-    }
-}
-
-
-data class Mflpt5(val b0: Short, val b1: Short, val b2: Short, val b3: Short, val b4: Short) {
-
-    companion object {
-        const val MemorySize = 5
-
-        val zero = Mflpt5(0, 0,0,0,0)
-        fun fromNumber(num: Number): Mflpt5 {
-            // see https://en.wikipedia.org/wiki/Microsoft_Binary_Format
-            // and https://sourceforge.net/p/acme-crossass/code-0/62/tree/trunk/ACME_Lib/cbm/mflpt.a
-            // and https://en.wikipedia.org/wiki/IEEE_754-1985
-
-            val flt = num.toDouble()
-            if(flt < FLOAT_MAX_NEGATIVE || flt > FLOAT_MAX_POSITIVE)
-                throw CompilerException("floating point number out of 5-byte mflpt range: $this")
-            if(flt==0.0)
-                return zero
-
-            val sign = if(flt<0.0) 0x80L else 0x00L
-            var exponent = 128 + 32	// 128 is cbm's bias, 32 is this algo's bias
-            var mantissa = flt.absoluteValue
-
-            // if mantissa is too large, shift right and adjust exponent
-            while(mantissa >= 0x100000000) {
-                mantissa /= 2.0
-                exponent ++
-            }
-            // if mantissa is too small, shift left and adjust exponent
-            while(mantissa < 0x80000000) {
-                mantissa *= 2.0
-                exponent --
-            }
-
-            return when {
-                exponent<0 -> zero  // underflow, use zero instead
-                exponent>255 -> throw CompilerException("floating point overflow: $this")
-                exponent==0 -> zero
-                else -> {
-                    val mantLong = mantissa.toLong()
-                    Mflpt5(
-                            exponent.toShort(),
-                            (mantLong.and(0x7f000000L) ushr 24).or(sign).toShort(),
-                            (mantLong.and(0x00ff0000L) ushr 16).toShort(),
-                            (mantLong.and(0x0000ff00L) ushr 8).toShort(),
-                            (mantLong.and(0x000000ffL)).toShort())
-                }
-            }
-        }
-    }
-
-    fun toDouble(): Double {
-        if(this == zero) return 0.0
-        val exp = b0 - 128
-        val sign = (b1.toInt() and 0x80) > 0
-        val number = 0x80000000L.or(b1.toLong() shl 24).or(b2.toLong() shl 16).or(b3.toLong() shl 8).or(b4.toLong())
-        val result = number.toDouble() * (2.0).pow(exp) / 0x100000000
-        return if(sign) -result else result
-    }
-}
-
-object Charset {
-    private val normalImg = ImageIO.read(javaClass.getResource("/charset/c64/charset-normal.png"))
-    private val shiftedImg = ImageIO.read(javaClass.getResource("/charset/c64/charset-shifted.png"))
-
-    private fun scanChars(img: BufferedImage): Array<BufferedImage> {
-
-        val transparent = BufferedImage(img.width, img.height, BufferedImage.TYPE_INT_ARGB)
-        transparent.createGraphics().drawImage(img, 0, 0, null)
-
-        val black = Color(0,0,0).rgb
-        val nopixel = Color(0,0,0,0).rgb
-        for(y in 0 until transparent.height) {
-            for(x in 0 until transparent.width) {
-                val col = transparent.getRGB(x, y)
-                if(col==black)
-                    transparent.setRGB(x, y, nopixel)
-            }
-        }
-
-        val numColumns = transparent.width / 8
-        val charImages = (0..255).map {
-            val charX = it % numColumns
-            val charY = it/ numColumns
-            transparent.getSubimage(charX*8, charY*8, 8, 8)
-        }
-        return charImages.toTypedArray()
-    }
-
-    val normalChars = scanChars(normalImg)
-    val shiftedChars = scanChars(shiftedImg)
-}

compiler/src/prog8/compiler/target/cpu6502/codegen/AsmOptimizer.kt

@@ -0,0 +1,242 @@
+package prog8.compiler.target.cpu6502.codegen
+
+
+// note: see https://wiki.nesdev.com/w/index.php/6502_assembly_optimisations
+
+
+fun optimizeAssembly(lines: MutableList<String>): Int {
+
+    var numberOfOptimizations = 0
+
+    var linesByFour = getLinesBy(lines, 4)
+
+    var mods = optimizeUselessStackByteWrites(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods = optimizeIncDec(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods = optimizeCmpSequence(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods = optimizeStoreLoadSame(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    mods= optimizeJsrRts(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
+    var linesByFourteen = getLinesBy(lines, 14)
+    mods = optimizeSameAssignments(linesByFourteen)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFourteen = getLinesBy(lines, 14)
+        numberOfOptimizations++
+    }
+
+    // TODO more assembly optimizations
+
+    return numberOfOptimizations
+}
+
+private class Modification(val lineIndex: Int, val remove: Boolean, val replacement: String?)
+
+private fun apply(modifications: List<Modification>, lines: MutableList<String>) {
+    for (modification in modifications.sortedBy { it.lineIndex }.reversed()) {
+        if(modification.remove)
+            lines.removeAt(modification.lineIndex)
+        else
+            lines[modification.lineIndex] = modification.replacement!!
+    }
+}
+
+private fun getLinesBy(lines: MutableList<String>, windowSize: Int) =
+// all lines (that aren't empty or comments) in sliding windows of certain size
+        lines.withIndex().filter { it.value.isNotBlank() && !it.value.trimStart().startsWith(';') }.windowed(windowSize, partialWindows = false)
+
+private fun optimizeCmpSequence(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // the when statement (on bytes) generates a sequence of:
+    //	 lda  $ce01,x
+    //	 cmp  #$20
+    //	 beq  check_prog8_s72choice_32
+    //	 lda  $ce01,x
+    //	 cmp  #$21
+    //	 beq  check_prog8_s73choice_33
+    // the repeated lda can be removed
+    val mods = mutableListOf<Modification>()
+    for(lines in linesByFour) {
+        if(lines[0].value.trim()=="lda  P8ESTACK_LO+1,x" &&
+                lines[1].value.trim().startsWith("cmp ") &&
+                lines[2].value.trim().startsWith("beq ") &&
+                lines[3].value.trim()=="lda  P8ESTACK_LO+1,x") {
+            mods.add(Modification(lines[3].index, true, null)) // remove the second lda
+        }
+    }
+    return mods
+}
+
+private fun optimizeUselessStackByteWrites(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // sta on stack, dex, inx, lda from stack -> eliminate this useless stack byte write
+    // this is a lot harder for word values because the instruction sequence varies.
+    val mods = mutableListOf<Modification>()
+    for(lines in linesByFour) {
+        if(lines[0].value.trim()=="sta  P8ESTACK_LO,x" &&
+                lines[1].value.trim()=="dex" &&
+                lines[2].value.trim()=="inx" &&
+                lines[3].value.trim()=="lda  P8ESTACK_LO,x") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, true, null))
+            mods.add(Modification(lines[3].index, true, null))
+        }
+    }
+    return mods
+}
+
+private fun optimizeSameAssignments(linesByFourteen: List<List<IndexedValue<String>>>): List<Modification> {
+
+    // optimize sequential assignments of the isSameAs value to various targets (bytes, words, floats)
+    // the float one is the one that requires 2*7=14 lines of code to check...
+    // @todo a better place to do this is in the Compiler instead and transform the Ast, or the AsmGen, and never even create the inefficient asm in the first place...
+
+    val mods = mutableListOf<Modification>()
+    for (pair in linesByFourteen) {
+        val first = pair[0].value.trimStart()
+        val second = pair[1].value.trimStart()
+        val third = pair[2].value.trimStart()
+        val fourth = pair[3].value.trimStart()
+        val fifth = pair[4].value.trimStart()
+        val sixth = pair[5].value.trimStart()
+        val seventh = pair[6].value.trimStart()
+        val eighth = pair[7].value.trimStart()
+
+        if(first.startsWith("lda") && second.startsWith("ldy") && third.startsWith("sta") && fourth.startsWith("sty") &&
+                fifth.startsWith("lda") && sixth.startsWith("ldy") && seventh.startsWith("sta") && eighth.startsWith("sty")) {
+            val firstvalue = first.substring(4)
+            val secondvalue = second.substring(4)
+            val thirdvalue = fifth.substring(4)
+            val fourthvalue = sixth.substring(4)
+            if(firstvalue==thirdvalue && secondvalue==fourthvalue) {
+                // lda/ldy   sta/sty   twice the isSameAs word -->  remove second lda/ldy pair (fifth and sixth lines)
+                mods.add(Modification(pair[4].index, true, null))
+                mods.add(Modification(pair[5].index, true, null))
+            }
+        }
+
+        if(first.startsWith("lda") && second.startsWith("sta") && third.startsWith("lda") && fourth.startsWith("sta")) {
+            val firstvalue = first.substring(4)
+            val secondvalue = third.substring(4)
+            if(firstvalue==secondvalue) {
+                // lda value / sta ? / lda isSameAs-value / sta ?  -> remove second lda (third line)
+                mods.add(Modification(pair[2].index, true, null))
+            }
+        }
+
+        if(first.startsWith("lda") && second.startsWith("ldy") && third.startsWith("sta") && fourth.startsWith("sty") &&
+                fifth.startsWith("lda") && sixth.startsWith("ldy") &&
+                (seventh.startsWith("jsr  floats.copy_float") || seventh.startsWith("jsr  cx16flt.copy_float"))) {
+
+            val nineth = pair[8].value.trimStart()
+            val tenth = pair[9].value.trimStart()
+            val eleventh = pair[10].value.trimStart()
+            val twelveth = pair[11].value.trimStart()
+            val thirteenth = pair[12].value.trimStart()
+            val fourteenth = pair[13].value.trimStart()
+
+            if(eighth.startsWith("lda") && nineth.startsWith("ldy") && tenth.startsWith("sta") && eleventh.startsWith("sty") &&
+                    twelveth.startsWith("lda") && thirteenth.startsWith("ldy") &&
+                    (fourteenth.startsWith("jsr  floats.copy_float") || fourteenth.startsWith("jsr  cx16flt.copy_float"))) {
+
+                if(first.substring(4) == eighth.substring(4) && second.substring(4)==nineth.substring(4)) {
+                    // identical float init
+                    mods.add(Modification(pair[7].index, true, null))
+                    mods.add(Modification(pair[8].index, true, null))
+                    mods.add(Modification(pair[9].index, true, null))
+                    mods.add(Modification(pair[10].index, true, null))
+                }
+            }
+        }
+    }
+    return mods
+}
+
+private fun optimizeStoreLoadSame(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can OFTEN be eliminated
+    // TODO this is not true if X is not a regular RAM memory address (but instead mapped I/O or ROM)
+    val mods = mutableListOf<Modification>()
+    for (pair in linesByFour) {
+        val first = pair[0].value.trimStart()
+        val second = pair[1].value.trimStart()
+
+        if ((first.startsWith("sta ") && second.startsWith("lda ")) ||
+                (first.startsWith("stx ") && second.startsWith("ldx ")) ||
+                (first.startsWith("sty ") && second.startsWith("ldy ")) ||
+                (first.startsWith("lda ") && second.startsWith("lda ")) ||
+                (first.startsWith("ldy ") && second.startsWith("ldy ")) ||
+                (first.startsWith("ldx ") && second.startsWith("ldx ")) ||
+                (first.startsWith("sta ") && second.startsWith("lda ")) ||
+                (first.startsWith("sty ") && second.startsWith("ldy ")) ||
+                (first.startsWith("stx ") && second.startsWith("ldx "))
+        ) {
+            val third = pair[2].value.trimStart()
+            if(!third.startsWith("b")) {
+                // no branch instruction follows, we can potentiall remove the load instruction
+                val firstLoc = first.substring(4).trimStart()
+                val secondLoc = second.substring(4).trimStart()
+                if (firstLoc == secondLoc) {
+                    mods.add(Modification(pair[1].index, true, null))
+                }
+            }
+        }
+    }
+    return mods
+}
+
+private fun optimizeIncDec(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // sometimes, iny+dey / inx+dex / dey+iny / dex+inx sequences are generated, these can be eliminated.
+    val mods = mutableListOf<Modification>()
+    for (pair in linesByFour) {
+        val first = pair[0].value
+        val second = pair[1].value
+        if ((" iny" in first || "\tiny" in first) && (" dey" in second || "\tdey" in second)
+                || (" inx" in first || "\tinx" in first) && (" dex" in second || "\tdex" in second)
+                || (" dey" in first || "\tdey" in first) && (" iny" in second || "\tiny" in second)
+                || (" dex" in first || "\tdex" in first) && (" inx" in second || "\tinx" in second)) {
+            mods.add(Modification(pair[0].index, true, null))
+            mods.add(Modification(pair[1].index, true, null))
+        }
+    }
+    return mods
+}
+
+private fun optimizeJsrRts(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // jsr Sub + rts -> jmp Sub
+    val mods = mutableListOf<Modification>()
+    for (pair in linesByFour) {
+        val first = pair[0].value
+        val second = pair[1].value
+        if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {
+            mods += Modification(pair[0].index, false, pair[0].value.replace("jsr", "jmp"))
+            mods += Modification(pair[1].index, true, null)
+        }
+    }
+    return mods
+}

compiler/src/prog8/compiler/target/cpu6502/codegen/ForLoopsAsmGen.kt

@@ -0,0 +1,585 @@
+package prog8.compiler.target.cpu6502.codegen
+
+import prog8.ast.Program
+import prog8.ast.base.ArrayElementTypes
+import prog8.ast.base.DataType
+import prog8.ast.base.RegisterOrPair
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.RangeExpr
+import prog8.ast.statements.ForLoop
+import prog8.ast.toHex
+import prog8.compiler.AssemblyError
+import kotlin.math.absoluteValue
+
+internal class ForLoopsAsmGen(private val program: Program, private val asmgen: AsmGen) {
+
+    internal fun translate(stmt: ForLoop) {
+        val iterableDt = stmt.iterable.inferType(program)
+        if(!iterableDt.isKnown)
+            throw AssemblyError("unknown dt")
+        when(stmt.iterable) {
+            is RangeExpr -> {
+                val range = (stmt.iterable as RangeExpr).toConstantIntegerRange()
+                if(range==null) {
+                    translateForOverNonconstRange(stmt, iterableDt.typeOrElse(DataType.STRUCT), stmt.iterable as RangeExpr)
+                } else {
+                    translateForOverConstRange(stmt, iterableDt.typeOrElse(DataType.STRUCT), range)
+                }
+            }
+            is IdentifierReference -> {
+                translateForOverIterableVar(stmt, iterableDt.typeOrElse(DataType.STRUCT), stmt.iterable as IdentifierReference)
+            }
+            else -> throw AssemblyError("can't iterate over ${stmt.iterable.javaClass} - should have been replaced by a variable")
+        }
+    }
+
+    private fun translateForOverNonconstRange(stmt: ForLoop, iterableDt: DataType, range: RangeExpr) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        val modifiedLabel = asmgen.makeLabel("for_modified")
+        val modifiedLabel2 = asmgen.makeLabel("for_modifiedb")
+        asmgen.loopEndLabels.push(endLabel)
+        val stepsize=range.step.constValue(program)!!.number.toInt()
+        when(iterableDt) {
+            DataType.ARRAY_B, DataType.ARRAY_UB -> {
+                if (stepsize==1 || stepsize==-1) {
+
+                    // bytes array, step 1 or -1
+
+                    val incdec = if(stepsize==1) "inc" else "dec"
+                    // loop over byte range via loopvar
+                    val varname = asmgen.asmVariableName(stmt.loopVar)
+                    asmgen.assignExpressionToVariable(range.from, varname, ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.body)
+                    asmgen.out("""
+                        lda  $varname
+$modifiedLabel          cmp  #0         ; modified 
+                        beq  $endLabel
+                        $incdec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
+
+                } else {
+
+                    // bytes, step >= 2 or <= -2
+
+                    // loop over byte range via loopvar
+                    val varname = asmgen.asmVariableName(stmt.loopVar)
+                    asmgen.assignExpressionToVariable(range.from, varname, ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.body)
+                    if(stepsize>0) {
+                        asmgen.out("""
+                            lda  $varname
+                            clc
+                            adc  #$stepsize
+                            sta  $varname
+$modifiedLabel              cmp  #0    ; modified
+                            bmi  $loopLabel
+                            beq  $loopLabel""")
+                    } else {
+                        asmgen.out("""
+                            lda  $varname
+                            sec
+                            sbc  #${stepsize.absoluteValue}
+                            sta  $varname
+$modifiedLabel              cmp  #0     ; modified
+                            bpl  $loopLabel""")
+                    }
+                    asmgen.out(endLabel)
+                }
+            }
+            DataType.ARRAY_W, DataType.ARRAY_UW -> {
+                when {
+
+                    // words, step 1 or -1
+
+                    stepsize == 1 || stepsize == -1 -> {
+                        val varname = asmgen.asmVariableName(stmt.loopVar)
+                        assignLoopvar(stmt, range)
+                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                        asmgen.out("""
+                            sty  $modifiedLabel+1
+                            sta  $modifiedLabel2+1
+$loopLabel""")
+                        asmgen.translate(stmt.body)
+                        asmgen.out("""
+                lda  $varname+1
+$modifiedLabel  cmp  #0    ; modified 
+                bne  +
+                lda  $varname
+$modifiedLabel2 cmp  #0    ; modified 
+                beq  $endLabel""")
+                        if(stepsize==1) {
+                            asmgen.out("""
++               inc  $varname
+                bne  $loopLabel
+                inc  $varname+1""")
+                            asmgen.jmp(loopLabel)
+                        } else {
+                            asmgen.out("""
++               lda  $varname
+                bne  +
+                dec  $varname+1
++               dec  $varname""")
+                            asmgen.jmp(loopLabel)
+                        }
+                        asmgen.out(endLabel)
+                    }
+                    stepsize > 0 -> {
+
+                        // (u)words, step >= 2
+                        val varname = asmgen.asmVariableName(stmt.loopVar)
+                        assignLoopvar(stmt, range)
+                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                        asmgen.out("""
+                            sty  $modifiedLabel+1
+                            sta  $modifiedLabel2+1
+$loopLabel""")
+                        asmgen.translate(stmt.body)
+
+                        if (iterableDt == DataType.ARRAY_UW) {
+                            asmgen.out("""
+                lda  $varname
+                clc
+                adc  #<$stepsize
+                sta  $varname
+                lda  $varname+1
+                adc  #>$stepsize
+                sta  $varname+1
+$modifiedLabel  cmp  #0     ; modified
+                bcc  $loopLabel
+                bne  $endLabel
+$modifiedLabel2 lda  #0     ; modified
+                cmp  $varname
+                bcc  $endLabel
+                bcs  $loopLabel
+$endLabel""")
+                        } else {
+                            asmgen.out("""
+                lda  $varname
+                clc
+                adc  #<$stepsize
+                sta  $varname
+                lda  $varname+1
+                adc  #>$stepsize
+                sta  $varname+1
+$modifiedLabel2 lda  #0   ; modified
+                cmp  $varname
+$modifiedLabel  lda  #0   ; modified
+                sbc  $varname+1
+                bvc  +
+                eor  #$80
++               bpl  $loopLabel                
+$endLabel""")
+                        }
+                    }
+                    else -> {
+
+                        // (u)words, step <= -2
+                        val varname = asmgen.asmVariableName(stmt.loopVar)
+                        assignLoopvar(stmt, range)
+                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                        asmgen.out("""
+                            sty  $modifiedLabel+1
+                            sta  $modifiedLabel2+1
+$loopLabel""")
+                        asmgen.translate(stmt.body)
+
+                        if(iterableDt==DataType.ARRAY_UW) {
+                            asmgen.out("""
+                lda  $varname
+                sec
+                sbc  #<${stepsize.absoluteValue}
+                sta  $varname
+                lda  $varname+1
+                sbc  #>${stepsize.absoluteValue}
+                sta  $varname+1
+$modifiedLabel  cmp  #0    ; modified                
+                bcc  $endLabel
+                bne  $loopLabel
+                lda  $varname
+$modifiedLabel2 cmp  #0    ; modified 
+                bcs  $loopLabel
+$endLabel""")
+                        } else {
+                            asmgen.out("""
+                lda  $varname
+                sec
+                sbc  #<${stepsize.absoluteValue}
+                sta  $varname
+                pha
+                lda  $varname+1
+                sbc  #>${stepsize.absoluteValue}
+                sta  $varname+1
+                pla
+$modifiedLabel2 cmp  #0    ; modified 
+                lda  $varname+1
+$modifiedLabel  sbc  #0    ; modified
+                bvc  +
+                eor  #$80
++               bpl  $loopLabel                
+$endLabel""")
+                        }
+                    }
+                }
+            }
+            else -> throw AssemblyError("range expression can only be byte or word")
+        }
+
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForOverIterableVar(stmt: ForLoop, iterableDt: DataType, ident: IdentifierReference) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        val iterableName = asmgen.asmVariableName(ident)
+        val decl = ident.targetVarDecl(program)!!
+        when(iterableDt) {
+            DataType.STR -> {
+                asmgen.out("""
+                    lda  #<$iterableName
+                    ldy  #>$iterableName
+                    sta  $loopLabel+1
+                    sty  $loopLabel+2
+$loopLabel          lda  ${65535.toHex()}       ; modified
+                    beq  $endLabel
+                    sta  ${asmgen.asmVariableName(stmt.loopVar)}""")
+                asmgen.translate(stmt.body)
+                asmgen.out("""
+                    inc  $loopLabel+1
+                    bne  $loopLabel
+                    inc  $loopLabel+2
+                    bne  $loopLabel
+$endLabel""")
+            }
+            DataType.ARRAY_UB, DataType.ARRAY_B -> {
+                val length = decl.arraysize!!.constIndex()!!
+                val indexVar = asmgen.makeLabel("for_index")
+                asmgen.out("""
+                    ldy  #0
+$loopLabel          sty  $indexVar
+                    lda  $iterableName,y
+                    sta  ${asmgen.asmVariableName(stmt.loopVar)}""")
+                asmgen.translate(stmt.body)
+                if(length<=255) {
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        cpy  #$length
+                        beq  $endLabel
+                        bne  $loopLabel""")
+                } else {
+                    // length is 256
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        bne  $loopLabel
+                        beq  $endLabel""")
+                }
+                if(length>=16 && asmgen.zeropage.available() > 0) {
+                    // allocate index var on ZP
+                    val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
+                    asmgen.out("""$indexVar = $zpAddr  ; auto zp UBYTE""")
+                } else {
+                    asmgen.out("""
+$indexVar           .byte  0""")
+                }
+                asmgen.out(endLabel)
+            }
+            DataType.ARRAY_W, DataType.ARRAY_UW -> {
+                val length = decl.arraysize!!.constIndex()!! * 2
+                val indexVar = asmgen.makeLabel("for_index")
+                val loopvarName = asmgen.asmVariableName(stmt.loopVar)
+                asmgen.out("""
+                    ldy  #0
+$loopLabel          sty  $indexVar
+                    lda  $iterableName,y
+                    sta  $loopvarName
+                    lda  $iterableName+1,y
+                    sta  $loopvarName+1""")
+                asmgen.translate(stmt.body)
+                if(length<=127) {
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        iny
+                        cpy  #$length
+                        beq  $endLabel
+                        bne  $loopLabel""")
+                } else {
+                    // length is 128 words, 256 bytes
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        iny
+                        bne  $loopLabel
+                        beq  $endLabel""")
+                }
+                if(length>=16 && asmgen.zeropage.available() > 0) {
+                    // allocate index var on ZP
+                    val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
+                    asmgen.out("""$indexVar = $zpAddr  ; auto zp UBYTE""")
+                } else {
+                    asmgen.out("""
+$indexVar           .byte  0""")
+                }
+                asmgen.out(endLabel)
+            }
+            DataType.ARRAY_F -> {
+                throw AssemblyError("for loop with floating point variables is not supported")
+            }
+            else -> throw AssemblyError("can't iterate over $iterableDt")
+        }
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForOverConstRange(stmt: ForLoop, iterableDt: DataType, range: IntProgression) {
+        if (range.isEmpty() || range.step==0)
+            throw AssemblyError("empty range or step 0")
+        if(iterableDt==DataType.ARRAY_B || iterableDt==DataType.ARRAY_UB) {
+            if(range.step==1 && range.last>range.first) return translateForSimpleByteRangeAsc(stmt, range)
+            if(range.step==-1 && range.last<range.first) return translateForSimpleByteRangeDesc(stmt, range)
+        }
+        else if(iterableDt==DataType.ARRAY_W || iterableDt==DataType.ARRAY_UW) {
+            if(range.step==1 && range.last>range.first) return translateForSimpleWordRangeAsc(stmt, range)
+            if(range.step==-1 && range.last<range.first) return translateForSimpleWordRangeDesc(stmt, range)
+        }
+
+        // not one of the easy cases, generate more complex code...
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        when(iterableDt) {
+            DataType.ARRAY_B, DataType.ARRAY_UB -> {
+                // loop over byte range via loopvar, step >= 2 or <= -2
+                val varname = asmgen.asmVariableName(stmt.loopVar)
+                asmgen.out("""
+                            lda  #${range.first}
+                            sta  $varname
+$loopLabel""")
+                asmgen.translate(stmt.body)
+                when (range.step) {
+                    0, 1, -1 -> {
+                        throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $stmt")
+                    }
+                    2 -> {
+                        if(range.last==255 || range.last==254) {
+                            asmgen.out("""
+                                inc  $varname
+                                beq  $endLabel
+                                inc  $varname
+                                bne  $loopLabel""")
+                        } else {
+                            asmgen.out("""
+                                inc  $varname
+                                inc  $varname
+                                lda  $varname
+                                cmp  #${range.last+2}
+                                bne  $loopLabel""")
+                        }
+                    }
+                    -2 -> {
+                        when (range.last) {
+                            0 -> {
+                                asmgen.out("""
+                                    lda  $varname
+                                    beq  $endLabel
+                                    dec  $varname
+                                    dec  $varname""")
+                                asmgen.jmp(loopLabel)
+                            }
+                            1 -> asmgen.out("""
+                                    dec  $varname
+                                    beq  $endLabel
+                                    dec  $varname
+                                    bne  $loopLabel""")
+                            else -> asmgen.out("""
+                                    dec  $varname
+                                    dec  $varname
+                                    lda  $varname
+                                    cmp  #${range.last-2}
+                                    bne  $loopLabel""")
+                        }
+                    }
+                    else -> {
+                        // step <= -3 or >= 3
+                        asmgen.out("""
+                            lda  $varname
+                            cmp  #${range.last}
+                            beq  $endLabel
+                            clc
+                            adc  #${range.step}
+                            sta  $varname""")
+                        asmgen.jmp(loopLabel)
+                    }
+                }
+                asmgen.out(endLabel)
+            }
+            DataType.ARRAY_W, DataType.ARRAY_UW -> {
+                // loop over word range via loopvar, step >= 2 or <= -2
+                val varname = asmgen.asmVariableName(stmt.loopVar)
+                when (range.step) {
+                    0, 1, -1 -> {
+                        throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $stmt")
+                    }
+                    else -> {
+                        // word, step >= 2 or <= -2
+                        // note: range.last has already been adjusted by kotlin itself to actually be the last value of the sequence
+                        asmgen.out("""
+                            lda  #<${range.first}
+                            ldy  #>${range.first}
+                            sta  $varname
+                            sty  $varname+1
+$loopLabel""")
+                        asmgen.translate(stmt.body)
+                        asmgen.out("""
+                            lda  $varname
+                            cmp  #<${range.last}
+                            bne  +
+                            lda  $varname+1
+                            cmp  #>${range.last}
+                            bne  +
+                            beq  $endLabel
++                           lda  $varname
+                            clc
+                            adc  #<${range.step}
+                            sta  $varname
+                            lda  $varname+1
+                            adc  #>${range.step}
+                            sta  $varname+1""")
+                        asmgen.jmp(loopLabel)
+                        asmgen.out(endLabel)
+                    }
+                }
+            }
+            else -> throw AssemblyError("range expression can only be byte or word")
+        }
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForSimpleByteRangeAsc(stmt: ForLoop, range: IntProgression) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        val varname = asmgen.asmVariableName(stmt.loopVar)
+        asmgen.out("""
+                lda  #${range.first}
+                sta  $varname
+$loopLabel""")
+        asmgen.translate(stmt.body)
+        if (range.last == 255) {
+            asmgen.out("""
+                inc  $varname
+                bne  $loopLabel
+$endLabel""")
+        } else {
+            asmgen.out("""
+                inc  $varname
+                lda  $varname
+                cmp  #${range.last+1}
+                bne  $loopLabel
+$endLabel""")
+        }
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForSimpleByteRangeDesc(stmt: ForLoop, range: IntProgression) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        val varname = asmgen.asmVariableName(stmt.loopVar)
+        asmgen.out("""
+                    lda  #${range.first}
+                    sta  $varname
+$loopLabel""")
+        asmgen.translate(stmt.body)
+        when (range.last) {
+            0 -> {
+                asmgen.out("""
+                    lda  $varname
+                    beq  $endLabel
+                    dec  $varname""")
+                asmgen.jmp(loopLabel)
+                asmgen.out(endLabel)
+            }
+            1 -> {
+                asmgen.out("""
+                    dec  $varname
+                    bne  $loopLabel
+$endLabel""")
+            }
+            else -> {
+                asmgen.out("""
+                    dec  $varname
+                    lda  $varname
+                    cmp  #${range.last-1}
+                    bne  $loopLabel
+$endLabel""")
+            }
+        }
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForSimpleWordRangeAsc(stmt: ForLoop, range: IntProgression) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        val varname = asmgen.asmVariableName(stmt.loopVar)
+        asmgen.out("""
+            lda  #<${range.first}
+            ldy  #>${range.first}
+            sta  $varname
+            sty  $varname+1
+$loopLabel""")
+        asmgen.translate(stmt.body)
+        asmgen.out("""
+            lda  $varname
+            cmp  #<${range.last}
+            bne  +
+            lda  $varname+1
+            cmp  #>${range.last}
+            beq  $endLabel
++           inc  $varname
+            bne  $loopLabel
+            inc  $varname+1""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun translateForSimpleWordRangeDesc(stmt: ForLoop, range: IntProgression) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        asmgen.loopEndLabels.push(endLabel)
+        val varname = asmgen.asmVariableName(stmt.loopVar)
+        asmgen.out("""
+            lda  #<${range.first}
+            ldy  #>${range.first}
+            sta  $varname
+            sty  $varname+1
+$loopLabel""")
+        asmgen.translate(stmt.body)
+        asmgen.out("""
+            lda  $varname
+            cmp  #<${range.last}
+            bne  +
+            lda  $varname+1
+            cmp  #>${range.last}
+            beq  $endLabel
++           lda  $varname
+            bne  +
+            dec  $varname+1
++           dec  $varname""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
+        asmgen.loopEndLabels.pop()
+    }
+
+    private fun assignLoopvar(stmt: ForLoop, range: RangeExpr) =
+        asmgen.assignExpressionToVariable(range.from, asmgen.asmVariableName(stmt.loopVar), stmt.loopVarDt(program).typeOrElse(DataType.STRUCT), stmt.definingSubroutine())
+}

compiler/src/prog8/compiler/target/cpu6502/codegen/FunctionCallAsmGen.kt

@@ -0,0 +1,347 @@
+package prog8.compiler.target.cpu6502.codegen
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.compiler.AssemblyError
+import prog8.compiler.target.CpuType
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignSource
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignTarget
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignment
+import prog8.compiler.target.cpu6502.codegen.assignment.TargetStorageKind
+
+
+internal class FunctionCallAsmGen(private val program: Program, private val asmgen: AsmGen) {
+
+    internal fun translateFunctionCallStatement(stmt: IFunctionCall) {
+        saveXbeforeCall(stmt)
+        translateFunctionCall(stmt)
+        restoreXafterCall(stmt)
+        // just ignore any result values from the function call.
+    }
+
+    internal fun saveXbeforeCall(stmt: IFunctionCall) {
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        if(sub.shouldSaveX()) {
+            val regSaveOnStack = sub.asmAddress==null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
+            val (keepAonEntry: Boolean, keepAonReturn: Boolean) = sub.shouldKeepA()
+            if(regSaveOnStack)
+                asmgen.saveRegisterStack(CpuRegister.X, keepAonEntry)
+            else
+                asmgen.saveRegisterLocal(CpuRegister.X, (stmt as Node).definingSubroutine()!!)
+        }
+    }
+
+    internal fun restoreXafterCall(stmt: IFunctionCall) {
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        if(sub.shouldSaveX()) {
+            val regSaveOnStack = sub.asmAddress==null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
+            val (keepAonEntry: Boolean, keepAonReturn: Boolean) = sub.shouldKeepA()
+
+            if(regSaveOnStack)
+                asmgen.restoreRegisterStack(CpuRegister.X, keepAonReturn)
+            else
+                asmgen.restoreRegisterLocal(CpuRegister.X)
+        }
+    }
+
+    internal fun translateFunctionCall(stmt: IFunctionCall) {
+        // Output only the code to setup the parameters and perform the actual call
+        // NOTE: does NOT output the code to deal with the result values!
+        // NOTE: does NOT output code to save/restore the X register for this call! Every caller should deal with this in their own way!!
+        //       (you can use subroutine.shouldSaveX() and saveX()/restoreX() routines as a help for this)
+
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        val subName = asmgen.asmSymbolName(stmt.target)
+        if(stmt.args.isNotEmpty()) {
+
+            if(sub.asmParameterRegisters.isEmpty()) {
+                // via variables
+                for(arg in sub.parameters.withIndex().zip(stmt.args)) {
+                    argumentViaVariable(sub, arg.first, arg.second)
+                }
+            } else {
+                // via registers
+                if(sub.parameters.size==1) {
+                    // just a single parameter, no risk of clobbering registers
+                    argumentViaRegister(sub, IndexedValue(0, sub.parameters.single()), stmt.args[0])
+                } else {
+
+                    fun isNoClobberRisk(expr: Expression): Boolean {
+                        if(expr is AddressOf ||
+                                expr is NumericLiteralValue ||
+                                expr is StringLiteralValue ||
+                                expr is ArrayLiteralValue ||
+                                expr is IdentifierReference)
+                            return true
+
+                        if(expr is FunctionCall) {
+                            if(expr.target.nameInSource==listOf("lsb") || expr.target.nameInSource==listOf("msb"))
+                                return isNoClobberRisk(expr.args[0])
+                            if(expr.target.nameInSource==listOf("mkword"))
+                                return isNoClobberRisk(expr.args[0]) && isNoClobberRisk(expr.args[1])
+                        }
+
+                        return false
+                    }
+
+                    when {
+                        stmt.args.all {isNoClobberRisk(it)} -> {
+                            // There's no risk of clobbering for these simple argument types. Optimize the register loading directly from these values.
+                            // register assignment order: 1) cx16 virtual word registers, 2) actual CPU registers, 3) CPU Carry status flag.
+                            val argsInfo = sub.parameters.withIndex().zip(stmt.args).zip(sub.asmParameterRegisters)
+                            val (cx16virtualRegs, args2) = argsInfo.partition { it.second.registerOrPair in Cx16VirtualRegisters }
+                            val (cpuRegs, statusRegs) = args2.partition { it.second.registerOrPair!=null }
+                            for(arg in cx16virtualRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                            for(arg in cpuRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                            for(arg in statusRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                        }
+                        else -> {
+                            // Risk of clobbering due to complex expression args. Evaluate first, then assign registers.
+                            registerArgsViaStackEvaluation(stmt, sub)
+                        }
+                    }
+                }
+            }
+        }
+
+        if(sub.inline && asmgen.options.optimize) {
+            if(sub.containsDefinedVariables())
+                throw AssemblyError("can't inline sub with vars")
+            if(!sub.isAsmSubroutine && sub.parameters.isNotEmpty())
+                throw AssemblyError("can't inline a non-asm subroutine with parameters")
+            asmgen.out("  \t; inlined routine follows: ${sub.name} from ${sub.position}")
+            val statements = sub.statements.filter { it !is ParameterVarDecl && it !is Directive }
+            statements.forEach { asmgen.translate(it) }
+        }
+        else {
+            asmgen.out("  jsr  $subName")
+        }
+
+        // remember: dealing with the X register and/or dealing with return values is the responsibility of the caller
+    }
+
+    private fun registerArgsViaStackEvaluation(stmt: IFunctionCall, sub: Subroutine) {
+        // this is called when one or more of the arguments are 'complex' and
+        // cannot be assigned to a register easily or risk clobbering other registers.
+
+        if(sub.parameters.isEmpty())
+            return
+
+        // 1. load all arguments reversed onto the stack: first arg goes last (is on top).
+        for (arg in stmt.args.reversed())
+            asmgen.translateExpression(arg)
+
+        var argForCarry: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForXregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForAregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+
+        asmgen.out("  inx")     // align estack pointer
+
+        for(argi in stmt.args.zip(sub.asmParameterRegisters).withIndex()) {
+            val plusIdxStr = if(argi.index==0) "" else "+${argi.index}"
+            when {
+                argi.value.second.statusflag == Statusflag.Pc -> {
+                    require(argForCarry == null)
+                    argForCarry = argi
+                }
+                argi.value.second.statusflag != null -> throw AssemblyError("can only use Carry as status flag parameter")
+                argi.value.second.registerOrPair in setOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) -> {
+                    require(argForXregister==null)
+                    argForXregister = argi
+                }
+                argi.value.second.registerOrPair in setOf(RegisterOrPair.A, RegisterOrPair.AY) -> {
+                    require(argForAregister == null)
+                    argForAregister = argi
+                }
+                argi.value.second.registerOrPair == RegisterOrPair.Y -> {
+                    asmgen.out("  ldy  P8ESTACK_LO$plusIdxStr,x")
+                }
+                argi.value.second.registerOrPair in Cx16VirtualRegisters -> {
+                    // immediately output code to load the virtual register, to avoid clobbering the A register later
+                    when (sub.parameters[argi.index].type) {
+                        in ByteDatatypes -> {
+                            // only load the lsb of the virtual register
+                            asmgen.out("""
+                                lda  P8ESTACK_LO$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}
+                            """)
+                            if (asmgen.isTargetCpu(CpuType.CPU65c02))
+                                asmgen.out("  stz  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1")
+                            else
+                                asmgen.out("  lda  #0 |  sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1")
+                        }
+                        in WordDatatypes, in IterableDatatypes ->
+                            asmgen.out("""
+                                lda  P8ESTACK_LO$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}
+                                lda  P8ESTACK_HI$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1
+                            """)
+                        else -> throw AssemblyError("weird dt")
+                    }
+                }
+                else -> throw AssemblyError("weird argument")
+            }
+        }
+
+        if(argForCarry!=null) {
+            val plusIdxStr = if(argForCarry.index==0) "" else "+${argForCarry.index}"
+            asmgen.out("""
+                lda  P8ESTACK_LO$plusIdxStr,x
+                beq  +
+                sec
+                bcs  ++
++               clc
++               php""")             // push the status flags
+        }
+
+        if(argForAregister!=null) {
+            val plusIdxStr = if(argForAregister.index==0) "" else "+${argForAregister.index}"
+            when(argForAregister.value.second.registerOrPair) {
+                RegisterOrPair.A -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x")
+                RegisterOrPair.AY -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x |  ldy  P8ESTACK_HI$plusIdxStr,x")
+                else -> throw AssemblyError("weird arg")
+            }
+        }
+
+        if(argForXregister!=null) {
+            val plusIdxStr = if(argForXregister.index==0) "" else "+${argForXregister.index}"
+
+            if(argForAregister!=null)
+                asmgen.out("  pha")
+            when(argForXregister.value.second.registerOrPair) {
+                RegisterOrPair.X -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x |  tax")
+                RegisterOrPair.AX -> asmgen.out("  ldy  P8ESTACK_LO$plusIdxStr,x |  lda  P8ESTACK_HI$plusIdxStr,x |  tax |  tya")
+                RegisterOrPair.XY -> asmgen.out("  ldy  P8ESTACK_HI$plusIdxStr,x |  lda  P8ESTACK_LO$plusIdxStr,x |  tax")
+                else -> throw AssemblyError("weird arg")
+            }
+            if(argForAregister!=null)
+                asmgen.out("  pla")
+        } else {
+            repeat(sub.parameters.size - 1) { asmgen.out("  inx") }       // unwind stack
+        }
+
+        if(argForCarry!=null)
+            asmgen.out("  plp")       // set the carry flag back to correct value
+    }
+
+    private fun argumentViaVariable(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
+        // pass parameter via a regular variable (not via registers)
+        val valueIDt = value.inferType(program)
+        if(!valueIDt.isKnown)
+            throw AssemblyError("unknown dt")
+        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
+        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
+            throw AssemblyError("argument type incompatible")
+
+        val varName = asmgen.asmVariableName(sub.scopedname+"."+parameter.value.name)
+        asmgen.assignExpressionToVariable(value, varName, parameter.value.type, sub)
+    }
+
+    private fun argumentViaRegister(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
+        // pass argument via a register parameter
+        val valueIDt = value.inferType(program)
+        if(!valueIDt.isKnown)
+            throw AssemblyError("unknown dt")
+        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
+        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
+            throw AssemblyError("argument type incompatible")
+
+        val paramRegister = sub.asmParameterRegisters[parameter.index]
+        val statusflag = paramRegister.statusflag
+        val register = paramRegister.registerOrPair
+        val requiredDt = parameter.value.type
+        if(requiredDt!=valueDt) {
+            if(valueDt largerThan requiredDt)
+                throw AssemblyError("can only convert byte values to word param types")
+        }
+        if (statusflag!=null) {
+            if(requiredDt!=valueDt)
+                throw AssemblyError("for statusflag, byte value is required")
+            if (statusflag == Statusflag.Pc) {
+                // this param needs to be set last, right before the jsr
+                // for now, this is already enforced on the subroutine definition by the Ast Checker
+                when(value) {
+                    is NumericLiteralValue -> {
+                        val carrySet = value.number.toInt() != 0
+                        asmgen.out(if(carrySet) "  sec" else "  clc")
+                    }
+                    is IdentifierReference -> {
+                        val sourceName = asmgen.asmVariableName(value)
+                        asmgen.out("""
+                pha
+                lda  $sourceName
+                beq  +
+                sec  
+                bcs  ++
+    +           clc
+    +           pla
+    """)
+                    }
+                    else -> {
+                        asmgen.assignExpressionToRegister(value, RegisterOrPair.A)
+                        asmgen.out("""
+                                    beq  +
+                                    sec
+                                    bcs  ++
+    +                               clc
+    +""")
+                    }
+                }
+            } else throw AssemblyError("can only use Carry as status flag parameter")
+        }
+        else {
+            // via register or register pair
+            register!!
+            if(requiredDt largerThan valueDt) {
+                // we need to sign extend the source, do this via temporary word variable
+                val scratchVar = asmgen.asmVariableName("P8ZP_SCRATCH_W1")
+                asmgen.assignExpressionToVariable(value, scratchVar, DataType.UBYTE, sub)
+                asmgen.signExtendVariableLsb(scratchVar, valueDt)
+                asmgen.assignVariableToRegister(scratchVar, register)
+            } else {
+                val target: AsmAssignTarget =
+                    if(parameter.value.type in ByteDatatypes && (register==RegisterOrPair.AX || register == RegisterOrPair.AY || register==RegisterOrPair.XY || register in Cx16VirtualRegisters))
+                        AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, parameter.value.type, sub, register = register)
+                    else
+                        AsmAssignTarget.fromRegisters(register, sub, program, asmgen)
+                val src = if(valueDt in PassByReferenceDatatypes) {
+                    if(value is IdentifierReference) {
+                        val addr = AddressOf(value, Position.DUMMY)
+                        AsmAssignSource.fromAstSource(addr, program, asmgen).adjustSignedUnsigned(target)
+                    } else {
+                        AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
+                    }
+                } else {
+                    AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
+                }
+                asmgen.translateNormalAssignment(AsmAssignment(src, target, false, program.memsizer, Position.DUMMY))
+            }
+        }
+    }
+
+    private fun isArgumentTypeCompatible(argType: DataType, paramType: DataType): Boolean {
+        if(argType isAssignableTo paramType)
+            return true
+        if(argType in ByteDatatypes && paramType in ByteDatatypes)
+            return true
+        if(argType in WordDatatypes && paramType in WordDatatypes)
+            return true
+
+        // we have a special rule for some types.
+        // strings are assignable to UWORD, for example, and vice versa
+        if(argType==DataType.STR && paramType==DataType.UWORD)
+            return true
+        if(argType==DataType.UWORD && paramType == DataType.STR)
+            return true
+
+        return false
+    }
+}