attempt to fix ReadTheDocs build issue

# Conflicts:
#	compiler/res/version.txt

.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
@@ -25,7 +25,10 @@ __pycache__/
 parser.out
 parsetab.py
 .pytest_cache/
-compiler/src/prog8_kotlin.jar
-compiler/src/compiled_java
 .attach_pid*
+compiler/lib/
 
+.gradle
+/prog8compiler.jar
+sd*.img
+*.d64

.idea/.gitignore

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

.idea/codeInsightSettings.xml

@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="JavaProjectCodeInsightSettings">
+    <excluded-names>
+      <name>kotlin.Result</name>
+    </excluded-names>
+  </component>
+</project>

.idea/compiler.xml

@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="CompilerConfiguration">
+    <option name="BUILD_PROCESS_HEAP_SIZE" value="1200" />
+    <bytecodeTargetLevel target="11" />
+  </component>
+</project>

.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/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_antlr4.xml

@@ -0,0 +1,19 @@
+<component name="libraryTable">
+  <library name="antlr.antlr4" type="repository">
+    <properties maven-id="org.antlr:antlr4:4.9.2">
+      <exclude>
+        <dependency maven-id="com.ibm.icu:icu4j" />
+      </exclude>
+    </properties>
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4/4.9.2/antlr4-4.9.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4-runtime/4.9.2/antlr4-runtime-4.9.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr-runtime/3.5.2/antlr-runtime-3.5.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/ST4/4.3/ST4-4.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/abego/treelayout/org.abego.treelayout.core/1.0.3/org.abego.treelayout.core-1.0.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/glassfish/javax.json/1.0.4/javax.json-1.0.4.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/github_hypfvieh_dbus_java.xml

@@ -0,0 +1,25 @@
+<component name="libraryTable">
+  <library name="github.hypfvieh.dbus.java" type="repository">
+    <properties maven-id="com.github.hypfvieh:dbus-java:3.3.1" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/hypfvieh/dbus-java/3.3.1/dbus-java-3.3.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-unixsocket/0.38.6/jnr-unixsocket-0.38.6.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-ffi/2.2.2/jnr-ffi-2.2.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jffi/1.3.1/jffi-1.3.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jffi/1.3.1/jffi-1.3.1-native.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/ow2/asm/asm/9.1/asm-9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/ow2/asm/asm-commons/9.1/asm-commons-9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/ow2/asm/asm-analysis/9.1/asm-analysis-9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/ow2/asm/asm-tree/9.1/asm-tree-9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/ow2/asm/asm-util/9.1/asm-util-9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-a64asm/1.0.0/jnr-a64asm-1.0.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-x86asm/1.0.2/jnr-x86asm-1.0.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-constants/0.10.1/jnr-constants-0.10.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-enxio/0.32.4/jnr-enxio-0.32.4.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/jnr/jnr-posix/3.1.5/jnr-posix-3.1.5.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-api/1.7.30/slf4j-api-1.7.30.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/glassfish_javax_json.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="glassfish.javax.json" type="repository">
+    <properties include-transitive-deps="false" maven-id="org.glassfish:javax.json:1.1.4" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/glassfish/javax.json/1.1.4/javax.json-1.1.4.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/hamcrest.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="hamcrest" type="repository">
+    <properties maven-id="org.hamcrest:hamcrest:2.2" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/hamcrest/hamcrest/2.2/hamcrest-2.2.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/jetbrains_kotlinx_cli_jvm.xml

@@ -0,0 +1,10 @@
+<component name="libraryTable">
+  <library name="jetbrains.kotlinx.cli.jvm" type="repository">
+    <properties include-transitive-deps="false" maven-id="org.jetbrains.kotlinx:kotlinx-cli-jvm:0.3.3" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-cli-jvm/0.3.3/kotlinx-cli-jvm-0.3.3.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/junit_jupiter.xml

@@ -0,0 +1,17 @@
+<component name="libraryTable">
+  <library name="junit.jupiter" type="repository">
+    <properties maven-id="org.junit.jupiter:junit-jupiter:5.7.2" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/jupiter/junit-jupiter/5.7.2/junit-jupiter-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/jupiter/junit-jupiter-api/5.7.2/junit-jupiter-api-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/apiguardian/apiguardian-api/1.1.0/apiguardian-api-1.1.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-commons/1.7.2/junit-platform-commons-1.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/jupiter/junit-jupiter-params/5.7.2/junit-jupiter-params-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/jupiter/junit-jupiter-engine/5.7.2/junit-jupiter-engine-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-engine/1.7.2/junit-platform-engine-1.7.2.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/michael_bull_kotlin_result_jvm.xml

@@ -0,0 +1,15 @@
+<component name="libraryTable">
+  <library name="michael.bull.kotlin.result.jvm" type="repository">
+    <properties maven-id="com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.12" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/com/michael-bull/kotlin-result/kotlin-result-jvm/1.1.12/kotlin-result-jvm-1.1.12.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.5.10/kotlin-stdlib-jdk8-1.5.10.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.5.10/kotlin-stdlib-1.5.10.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.5.10/kotlin-stdlib-jdk7-1.5.10.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.5.10/kotlin-stdlib-common-1.5.10.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/slf4j_simple.xml

@@ -0,0 +1,11 @@
+<component name="libraryTable">
+  <library name="slf4j.simple" type="repository">
+    <properties maven-id="org.slf4j:slf4j-simple:1.7.30" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-simple/1.7.30/slf4j-simple-1.7.30.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-api/1.7.30/slf4j-api-1.7.30.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/takes.xml

@@ -0,0 +1,13 @@
+<component name="libraryTable">
+  <library name="takes" type="repository">
+    <properties maven-id="org.takes:takes:1.19" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/takes/takes/1.19/takes-1.19.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/cactoos/cactoos/0.42/cactoos-0.42.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/apache/commons/commons-text/1.4/commons-text-1.4.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/apache/commons/commons-lang3/3.7/commons-lang3-3.7.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </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/Prog8ANTLR.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" project-jdk-name="Kotlin SDK" project-jdk-type="KotlinSDK">
+    <output url="file://$PROJECT_DIR$/out" />
+  </component>
+</project>

.idea/modules.xml

@@ -0,0 +1,17 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="ProjectModuleManager">
+    <modules>
+      <module fileurl="file://$PROJECT_DIR$/codeGeneration/codeGeneration.iml" filepath="$PROJECT_DIR$/codeGeneration/codeGeneration.iml" />
+      <module fileurl="file://$PROJECT_DIR$/codeOptimizers/codeOptimizers.iml" filepath="$PROJECT_DIR$/codeOptimizers/codeOptimizers.iml" />
+      <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://$PROJECT_DIR$/compilerInterfaces/compilerInterfaces.iml" filepath="$PROJECT_DIR$/compilerInterfaces/compilerInterfaces.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>

.readthedocs.yaml

@@ -0,0 +1,29 @@
+# .readthedocs.yaml
+# Read the Docs configuration file
+# See https://docs.readthedocs.io/en/stable/config-file/v2.html for details
+
+# Required
+version: 2
+
+# Set the version of Python and other tools you might need
+build:
+  os: ubuntu-20.04
+  tools:
+    python: "3.9"
+    # You can also specify other tool versions:
+    # nodejs: "16"
+    # rust: "1.55"
+    # golang: "1.17"
+
+# Build documentation in the docs/ directory with Sphinx
+sphinx:
+  configuration: docs/source/conf.py
+
+# If using Sphinx, optionally build your docs in additional formats such as PDF
+formats:
+  - pdf
+
+# Optionally declare the Python requirements required to build your docs
+python:
+  install:
+    - requirements: docs/requirements.txt

CompilerDevelopment.md

@@ -0,0 +1,32 @@
+#### Just a few remarks upfront:
+* There is the (gradle/IDEA) module `parser`: that's the parser generated by ANTLR4, in Java. The only file to be edited here is the grammar, `prog8.g4`.
+* Then we have the module `compilerAst` - in Kotlin - which uses `parser` and adds AST nodes. Here we put our additions to the generated thing, *including any tests of the parsing stage*.
+  - the name is a bit misleading, as this module isn't (or, resp. shouldn't be; see below) about *compiling*, only the parsing stage
+  - also, the tree that comes out isn't much of an *abstraction*, but rather still more or less a parse tree (this might very well change).
+  - **However, let's not *yet* rename the module.** We'll find a good name during refactoring.
+
+#### Problems with `compilerAst`:
+* `ModuleImporter.kt`, doing (Prog8-) module resolution. That's not the parser's job.
+* `ParsingFailedError` (in `ModuleParsing.kt`): this exception (it is actually *not* a `java.lang.Error`...) is thrown in a number of places, where other exceptions would make more sense. For example: not finding a file should just yield a `NoSuchFileException`, not this one. The other problem with it is that it does not provide any additional information about the source of parsing error, in particular a `Position`.
+* During parsing, character literals are turned into UBYTEs (since there is no basic type e.g. CHAR). That's bad because it depends on a specific character encoding (`IStringEncoding` in `compilerAst/src/prog8/ast/AstToplevel.kt`) of/for some target platform. Note that *strings* are indeed encoded later, in the `compiler` module.
+* The same argument applies to `IMemSizer`, and - not entirely sure about that - `IBuiltinFunctions`.
+
+#### Steps to take, in conceptual (!) order:
+1. introduce an abstraction `SourceCode` that encapsulates the origin and actual loading of Prog8 source code
+   - from the local file system (use case: user programs)
+   - from resources (prog8lib)
+   - from plain strings (for testing)
+2. add subclass `ParseError : ParsingFailedError` which adds information about the *source of parsing error* (`SourceCode` and `Position`). We cannot just replace `ParsingFailedError`  right away because it is so widely used (even in the `compiler` module). Therefore we'll just subclass for the time being, add more and more tests requiring the new one to be thrown (or, resp., NOT to be thrown), and gradually transition.
+3. introduce a minimal interface to the outside, input: `SourceCode`, output: a tree with a `Module` node as the root
+   - this will be the Kotlin singleton `Prog8Parser` with the main method `parseModule`
+   - plus, optionally, method's for registering/unregistering a listener with the parser
+   - the *only* exception ever thrown / reported to listeners (TBD) will be `ParseError`
+   - anything related to the lexer, error strategies, character/token streams is hidden from the outside
+   - to make a clear distinction between the *generated* parser (and lexer) vs. `Prog8Parser`, and to discourage directly using the generated stuff, we'll rename the existing `prog8Parser`/`prog8Lexer` to `Prog8ANTLRParser` and `Prog8ANTLRLexer` and move them to package `prog8.parser.generated`
+4. introduce AST node `CharLiteral` and keep them until after identifier resolution and type checking; insert there an AST transformation step that turns them in UBYTE constants (literals)
+5. remove uses of `IStringEncoding` from module `compilerAst` - none should be necessary anymore
+6. move `IStringEncoding` to module `compiler`
+7. same with `ModuleImporter`, then rewrite that (addressing #46)
+8. refactor AST nodes and grammar: less generated parse tree nodes (`XyzContext`), less intermediary stuff (private classes in `Antrl2Kotlin.kt`), more compact code. Also: nicer names such as simply `StringLiteral` instead of `StringLiteralValue`
+9. re-think `IStringEncoding` to address #38
+

LICENSE

@@ -1,3 +1,10 @@
+
+This sofware license is for Prog8 the compiler + associated libraries.
+The software generated by running the compiler is excluded from this.
+
+
+
+
                     GNU GENERAL PUBLIC LICENSE
                        Version 3, 29 June 2007
 

README.md

@@ -1,101 +1,152 @@
-Prog8 - Structured Programming Language for 8-bit 6502/6510 microprocessors
-===========================================================================
+[![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/
+
+Software license
+----------------
+GNU GPL 3.0, see file LICENSE
+
+- prog8 (the compiler + libraries) is licensed under GNU GPL 3.0
+- *exception:* the resulting files created by running the compiler are free to use in whatever way desired.
+
+
+What does Prog8 provide?
+------------------------
+
+- reduction of source code length over raw assembly
+- fast execution speed due to compilation to native assembly code. It's possible to write certain raster interrupt 'demoscene' effects purely in Prog8.
 - 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)
+- floating point math is supported if the target system provides floating point library routines (C64 and Cx16 both do)
+- strings can contain escaped characters but also many symbols directly if they have a petscii equivalent, such as "♠♥♣♦π▚●○╳". Characters like ^, _, \, {, } and | are also accepted and converted to the closest petscii equivalents.
+- automatic static 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
+- programs can be run multiple times without reloading because of automatic variable (re)initializations.
+- conditional branches
+- 'when' statement to provide a concise jump table alternative to if/elseif chains
+- 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
+- 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:
+
+![wizzine screen](docs/source/_static/wizzine.png)
+
+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.gradle

@@ -0,0 +1,10 @@
+plugins {
+    id "org.jetbrains.kotlin.jvm" version "$kotlinVersion" apply false
+}
+
+allprojects {
+    repositories {
+        mavenLocal()
+        mavenCentral()
+    }
+}

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 -package prog8.parser ./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!"

codeGeneration/build.gradle

@@ -0,0 +1,56 @@
+
+plugins {
+    id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm"
+}
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+dependencies {
+    implementation project(':compilerInterfaces')
+    implementation project(':compilerAst')
+    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
+    // implementation "org.jetbrains.kotlin:kotlin-reflect"
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.12"
+
+    testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
+    testImplementation 'org.junit.jupiter:junit-jupiter-api:5.7.2'
+    testImplementation 'org.hamcrest:hamcrest:2.2'
+    testRuntimeOnly 'org.junit.jupiter:junit-jupiter-engine:5.7.2'
+
+}
+
+sourceSets {
+    main {
+        java {
+            srcDirs = ["${project.projectDir}/src"]
+        }
+        resources {
+            srcDirs = ["${project.projectDir}/res"]
+        }
+    }
+    test {
+        java {
+            srcDirs = ["${project.projectDir}/test"]
+        }
+    }
+}
+
+
+test {
+    // Enable JUnit 5 (Gradle 4.6+).
+    useJUnitPlatform()
+
+    // Always run tests, even when nothing changed.
+    dependsOn 'cleanTest'
+
+    // Show test results.
+    testLogging {
+        events "skipped", "failed"
+    }
+}

codeGeneration/codeGeneration.iml

@@ -0,0 +1,19 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="JAVA_MODULE" version="4">
+  <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://$MODULE_DIR$/test" isTestSource="true" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
+    </content>
+    <orderEntry type="jdk" jdkName="openjdk-11" jdkType="JavaSDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+    <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
+    <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
+    <orderEntry type="module" module-name="compilerInterfaces" />
+    <orderEntry type="module" module-name="compilerAst" />
+    <orderEntry type="library" scope="TEST" name="hamcrest" level="project" />
+    <orderEntry type="library" name="junit.jupiter" level="project" />
+  </component>
+</module>

codeGeneration/src/prog8/compiler/target/AssemblyError.kt

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

codeGeneration/src/prog8/compiler/target/C64Target.kt

@@ -0,0 +1,35 @@
+package prog8.compiler.target
+
+import com.github.michaelbull.result.fold
+import prog8.ast.base.ByteDatatypes
+import prog8.ast.base.DataType
+import prog8.ast.base.PassByReferenceDatatypes
+import prog8.ast.base.WordDatatypes
+import prog8.compiler.target.c64.C64MachineDefinition
+import prog8.compiler.target.cbm.Petscii
+import prog8.compilerinterface.ICompilationTarget
+
+
+object C64Target: ICompilationTarget {
+    override val name = "c64"
+    override val machine = C64MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean): List<Short> {
+        val coded = if (altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+        return coded.fold(
+            failure = { throw it },
+            success = { it }
+        )
+    }
+    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
+        }
+    }
+}

codeGeneration/src/prog8/compiler/target/Cx16Target.kt

@@ -0,0 +1,36 @@
+package prog8.compiler.target
+
+import com.github.michaelbull.result.fold
+import prog8.ast.base.ByteDatatypes
+import prog8.ast.base.DataType
+import prog8.ast.base.PassByReferenceDatatypes
+import prog8.ast.base.WordDatatypes
+import prog8.compiler.target.cbm.Petscii
+import prog8.compiler.target.cx16.CX16MachineDefinition
+import prog8.compilerinterface.ICompilationTarget
+
+
+object Cx16Target: ICompilationTarget {
+    override val name = "cx16"
+    override val machine = CX16MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean): List<Short> {
+        val coded= if (altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+        return coded.fold(
+            failure = { throw it },
+            success = { it }
+        )
+    }
+    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
+        }
+    }
+}
+

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

@@ -0,0 +1,206 @@
+package prog8.compiler.target.c64
+
+import prog8.compiler.target.cbm.viceMonListPostfix
+import prog8.compilerinterface.*
+import java.io.IOException
+import java.nio.file.Path
+import kotlin.math.absoluteValue
+import kotlin.math.pow
+
+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(selectedEmulator: Int, programNameWithPath: Path) {
+        if(selectedEmulator!=1) {
+            System.err.println("The c64 target only supports the main emulator (Vice).")
+            return
+        }
+
+        for(emulator in listOf("x64sc", "x64")) {
+            println("\nStarting C-64 emulator $emulator...")
+            val cmdline = listOf(emulator, "-silent", "-moncommands", "${programNameWithPath}.$viceMonListPostfix",
+                    "-autostartprgmode", "1", "-autostart-warp", "-autostart", "${programNameWithPath}.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")
+
+
+    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 arrayOf(ZeropageType.FLOATSAFE, ZeropageType.BASICSAFE, ZeropageType.DONTUSE ))
+                throw InternalCompilerException("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(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 zeropage 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()
+                }
+            }
+
+            removeReservedFromFreePool()
+        }
+    }
+
+    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 InternalCompilerException("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 InternalCompilerException("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"
+        }
+    }
+}

codeGeneration/src/prog8/compiler/target/cbm/AssemblyProgram.kt

@@ -0,0 +1,103 @@
+package prog8.compiler.target.cbm
+
+import com.github.michaelbull.result.Ok
+import com.github.michaelbull.result.Result
+import com.github.michaelbull.result.mapError
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.IAssemblyProgram
+import prog8.compilerinterface.OutputType
+import prog8.compilerinterface.generatedLabelPrefix
+import prog8.parser.SourceCode
+import java.io.File
+import java.nio.file.Path
+import kotlin.io.path.Path
+import kotlin.io.path.isRegularFile
+
+
+internal const val viceMonListPostfix = "vice-mon-list"
+
+class AssemblyProgram(
+        override val valid: Boolean,
+        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.$viceMonListPostfix")
+
+    override fun assemble(quiet: Boolean, options: CompilationOptions): Int {
+        // 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")
+
+        if(quiet)
+            command.add("--quiet")
+
+        val outFile = when (options.output) {
+            OutputType.PRG -> {
+                command.add("--cbm-prg")
+                println("\nCreating prg for target $compTarget.")
+                prgFile
+            }
+            OutputType.RAW -> {
+                command.add("--nostart")
+                println("\nCreating raw binary for target $compTarget.")
+                binFile
+            }
+        }
+        command.addAll(listOf("--output", outFile.toString(), assemblyFile.toString()))
+
+        val proc = ProcessBuilder(command).inheritIO().start()
+        val result = proc.waitFor()
+        if (result == 0) {
+            removeGeneratedLabelsFromMonlist()
+            generateBreakpointList()
+        }
+        return result
+    }
+
+    private fun removeGeneratedLabelsFromMonlist() {
+        val pattern = Regex("""al (\w+) \S+${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 viceMonListFile.toFile().readLines()) {
+            val match = pattern.matchEntire(line)
+            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, "; $num breakpoints have been defined")
+        breakpoints.add(2, "del")
+        viceMonListFile.toFile().appendText(breakpoints.joinToString("\n") + "\n")
+    }
+}
+
+
+internal fun loadAsmIncludeFile(filename: String, source: SourceCode): Result<String, NoSuchFileException> {
+    return if (filename.startsWith(SourceCode.libraryFilePrefix)) {
+        return com.github.michaelbull.result.runCatching {
+            SourceCode.Resource("/prog8lib/${filename.substring(SourceCode.libraryFilePrefix.length)}").readText()
+        }.mapError { NoSuchFileException(File(filename)) }
+    } else {
+        val sib = Path(source.origin).resolveSibling(filename)
+        if (sib.isRegularFile())
+            Ok(SourceCode.File(sib).readText())
+        else
+            Ok(SourceCode.File(Path(filename)).readText())
+    }
+}

codeGeneration/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> {
+    // 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) but how does this code know?
+    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
+}

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

@@ -0,0 +1,593 @@
+package prog8.compiler.target.cpu6502.codegen
+
+import prog8.ast.Program
+import prog8.ast.base.ArrayToElementTypes
+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.target.AssemblyError
+import prog8.compilerinterface.toConstantIntegerRange
+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.getOr(DataType.UNDEFINED), stmt.iterable as RangeExpr)
+                } else {
+                    translateForOverConstRange(stmt, iterableDt.getOr(DataType.UNDEFINED), range)
+                }
+            }
+            is IdentifierReference -> {
+                translateForOverIterableVar(stmt, iterableDt.getOr(DataType.UNDEFINED), stmt.iterable as IdentifierReference)
+            }
+            else -> throw AssemblyError("can't iterate over ${stmt.iterable.javaClass} - should have been replaced by a variable")
+        }
+    }
+
+    private fun translateForOverNonconstRange(stmt: ForLoop, iterableDt: DataType, range: RangeExpr) {
+        val loopLabel = asmgen.makeLabel("for_loop")
+        val endLabel = asmgen.makeLabel("for_end")
+        val modifiedLabel = asmgen.makeLabel("for_modified")
+        val modifiedLabel2 = asmgen.makeLabel("for_modifiedb")
+        asmgen.loopEndLabels.push(endLabel)
+        val stepsize=range.step.constValue(program)!!.number.toInt()
+
+        if(stepsize < -1) {
+            val limit = range.to.constValue(program)?.number?.toDouble()
+            if(limit==0.0)
+                throw AssemblyError("for unsigned loop variable it's not possible to count down with step != -1 from a non-const value to exactly zero due to value wrapping")
+        }
+
+        when(iterableDt) {
+            DataType.ARRAY_B, DataType.ARRAY_UB -> {
+                if (stepsize==1 || stepsize==-1) {
+
+                    // bytes array, step 1 or -1
+
+                    val incdec = if(stepsize==1) "inc" else "dec"
+                    // loop over byte range via loopvar
+                    val varname = asmgen.asmVariableName(stmt.loopVar)
+                    asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayToElementTypes.getValue(iterableDt), null)
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.body)
+                    asmgen.out("""
+                        lda  $varname
+$modifiedLabel          cmp  #0         ; modified 
+                        beq  $endLabel
+                        $incdec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
+
+                } else {
+
+                    // bytes, step >= 2 or <= -2
+
+                    // loop over byte range via loopvar
+                    val varname = asmgen.asmVariableName(stmt.loopVar)
+                    asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayToElementTypes.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  $endLabel""")
+                }
+                if(length>=16 && asmgen.zeropage.hasByteAvailable()) {
+                    // allocate index var on ZP
+                    val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
+                    asmgen.out("""$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  $endLabel""")
+                }
+                if(length>=16 && asmgen.zeropage.hasByteAvailable()) {
+                    // allocate index var on ZP
+                    val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
+                    asmgen.out("""$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).getOr(DataType.UNDEFINED), stmt.definingSubroutine)
+}

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

@@ -0,0 +1,372 @@
+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.InlineAssembly
+import prog8.ast.statements.RegisterOrStatusflag
+import prog8.ast.statements.Subroutine
+import prog8.ast.statements.SubroutineParameter
+import prog8.compiler.target.AssemblyError
+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
+import prog8.compilerinterface.CpuType
+
+
+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
+            if(regSaveOnStack)
+                asmgen.saveRegisterStack(CpuRegister.X, sub.shouldKeepA().saveOnEntry)
+            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
+            if(regSaveOnStack)
+                asmgen.restoreRegisterStack(CpuRegister.X, sub.shouldKeepA().saveOnReturn)
+            else
+                asmgen.restoreRegisterLocal(CpuRegister.X)
+        }
+    }
+
+    internal fun translateFunctionCall(stmt: IFunctionCall) {
+        // Output only the code to set up 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 {
+                require(sub.isAsmSubroutine)
+                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) {
+            asmgen.out("  jsr  $subName")
+        } else {
+            // inline the subroutine.
+            // we do this by copying the subroutine's statements at the call site.
+            // NOTE: *if* there is a return statement, it will be the only one, and the very last statement of the subroutine
+            // (this condition has been enforced by an ast check earlier)
+
+            // note: for now, this is only reliably supported for asmsubs.
+            if(!sub.isAsmSubroutine)
+                throw AssemblyError("can only reliably inline asmsub routines at this time")
+
+            asmgen.out("  \t; inlined routine follows: ${sub.name}")
+            val assembly = sub.statements.single() as InlineAssembly
+            asmgen.translate(assembly)
+            asmgen.out("  \t; inlined routine end: ${sub.name}")
+        }
+
+        // remember: dealing with the X register and/or dealing with return values is the responsibility of the caller
+    }
+
+    private fun registerArgsViaStackEvaluation(stmt: IFunctionCall, sub: Subroutine) {
+        // this is called when one or more of the arguments are 'complex' and
+        // cannot be assigned to a register easily or risk clobbering other registers.
+        // TODO find another way to prepare the arguments, without using the eval stack
+
+        if(sub.parameters.isEmpty())
+            return
+
+
+        // 1. load all arguments reversed onto the stack: first arg goes last (is on top).
+
+        for (arg in stmt.args.reversed())
+            asmgen.translateExpression(arg)
+
+        // TODO here's an alternative to the above, but for now generates bigger code due to intermediate register steps:
+//        for (arg in stmt.args.reversed()) {
+//            // note this stuff below is needed to (eventually) avoid calling asmgen.translateExpression()
+//            // TODO also This STILL requires the translateNormalAssignment() to be fixed to avoid stack eval for expressions...
+//            val dt = arg.inferType(program).getOr(DataType.UNDEFINED)
+//            asmgen.assignExpressionTo(arg, AsmAssignTarget(TargetStorageKind.STACK, program, asmgen, dt, sub))
+//        }
+
+        var argForCarry: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForXregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForAregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+
+        asmgen.out("  inx")     // align estack pointer
+
+        for(argi in stmt.args.zip(sub.asmParameterRegisters).withIndex()) {
+            val plusIdxStr = if(argi.index==0) "" else "+${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 arrayOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) -> {
+                    require(argForXregister==null)
+                    argForXregister = argi
+                }
+                argi.value.second.registerOrPair in arrayOf(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.${argi.value.second.registerOrPair.toString().lowercase()}
+                            """)
+                            if (asmgen.isTargetCpu(CpuType.CPU65c02))
+                                asmgen.out(
+                                    "  stz  cx16.${
+                                        argi.value.second.registerOrPair.toString().lowercase()
+                                    }+1")
+                            else
+                                asmgen.out(
+                                    "  lda  #0 |  sta  cx16.${
+                                        argi.value.second.registerOrPair.toString().lowercase()
+                                    }+1")
+                        }
+                        in WordDatatypes, in IterableDatatypes ->
+                            asmgen.out(
+                                """
+                                lda  P8ESTACK_LO$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().lowercase()}
+                                lda  P8ESTACK_HI$plusIdxStr,x
+                                sta  cx16.${
+                                    argi.value.second.registerOrPair.toString().lowercase()
+                                }+1
+                            """)
+                        else -> throw AssemblyError("weird dt")
+                    }
+                }
+                else -> throw AssemblyError("weird argument")
+            }
+        }
+
+        if(argForCarry!=null) {
+            val plusIdxStr = if(argForCarry.index==0) "" else "+${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.getOr(DataType.UNDEFINED)
+        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.getOr(DataType.UNDEFINED)
+        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
+                asmgen.assignExpressionToVariable(value, "P8ZP_SCRATCH_W1", DataType.UBYTE, sub)
+                asmgen.signExtendVariableLsb("P8ZP_SCRATCH_W1", valueDt)
+                asmgen.assignVariableToRegister("P8ZP_SCRATCH_W1", 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, false, 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
+    }
+}

codeGeneration/src/prog8/compiler/target/cpu6502/codegen/PostIncrDecrAsmGen.kt

@@ -0,0 +1,128 @@
+package prog8.compiler.target.cpu6502.codegen
+
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.PostIncrDecr
+import prog8.ast.toHex
+import prog8.compiler.target.AssemblyError
+
+
+internal class PostIncrDecrAsmGen(private val program: Program, private val asmgen: AsmGen) {
+    internal fun translate(stmt: PostIncrDecr) {
+        val incr = stmt.operator=="++"
+        val targetIdent = stmt.target.identifier
+        val targetMemory = stmt.target.memoryAddress
+        val targetArrayIdx = stmt.target.arrayindexed
+        val scope = stmt.definingSubroutine
+        when {
+            targetIdent!=null -> {
+                val what = asmgen.asmVariableName(targetIdent)
+                when (stmt.target.inferType(program).getOr(DataType.UNDEFINED)) {
+                    in ByteDatatypes -> asmgen.out(if (incr) "  inc  $what" else "  dec  $what")
+                    in WordDatatypes -> {
+                        if(incr)
+                            asmgen.out(" inc  $what |  bne  + |  inc  $what+1 |+")
+                        else
+                            asmgen.out("""
+        lda  $what
+        bne  +
+        dec  $what+1
++       dec  $what 
+""")
+                    }
+                    DataType.FLOAT -> {
+                        asmgen.out("  lda  #<$what |  ldy  #>$what")
+                        asmgen.out(if(incr) "  jsr  floats.inc_var_f" else "  jsr  floats.dec_var_f")
+                    }
+                    else -> throw AssemblyError("need numeric type")
+                }
+            }
+            targetMemory!=null -> {
+                when (val addressExpr = targetMemory.addressExpression) {
+                    is NumericLiteralValue -> {
+                        val what = addressExpr.number.toHex()
+                        asmgen.out(if(incr) "  inc  $what" else "  dec  $what")
+                    }
+                    is IdentifierReference -> {
+                        val what = asmgen.asmVariableName(addressExpr)
+                        asmgen.out("  lda  $what |  sta  (+) +1 |  lda  $what+1 |  sta  (+) +2")
+                        if(incr)
+                            asmgen.out("+\tinc  ${'$'}ffff\t; modified")
+                        else
+                            asmgen.out("+\tdec  ${'$'}ffff\t; modified")
+                    }
+                    else -> {
+                        asmgen.assignExpressionToRegister(addressExpr, RegisterOrPair.AY)
+                        asmgen.out("  sta  (+) + 1 |  sty  (+) + 2")
+                        if(incr)
+                            asmgen.out("+\tinc  ${'$'}ffff\t; modified")
+                        else
+                            asmgen.out("+\tdec  ${'$'}ffff\t; modified")
+                    }
+                }
+            }
+            targetArrayIdx!=null -> {
+                val asmArrayvarname = asmgen.asmVariableName(targetArrayIdx.arrayvar)
+                val elementDt = targetArrayIdx.inferType(program).getOr(DataType.UNDEFINED)
+                val constIndex = targetArrayIdx.indexer.constIndex()
+                if(constIndex!=null) {
+                    val indexValue = constIndex * program.memsizer.memorySize(elementDt)
+                    when(elementDt) {
+                        in ByteDatatypes -> asmgen.out(if (incr) "  inc  $asmArrayvarname+$indexValue" else "  dec  $asmArrayvarname+$indexValue")
+                        in WordDatatypes -> {
+                            if(incr)
+                                asmgen.out(" inc  $asmArrayvarname+$indexValue |  bne  + |  inc  $asmArrayvarname+$indexValue+1 |+")
+                            else
+                                asmgen.out("""
+    lda  $asmArrayvarname+$indexValue
+    bne  +
+    dec  $asmArrayvarname+$indexValue+1
++       dec  $asmArrayvarname+$indexValue 
+""")
+                        }
+                        DataType.FLOAT -> {
+                            asmgen.out("  lda  #<$asmArrayvarname+$indexValue |  ldy  #>$asmArrayvarname+$indexValue")
+                            asmgen.out(if(incr) "  jsr  floats.inc_var_f" else "  jsr  floats.dec_var_f")
+                        }
+                        else -> throw AssemblyError("need numeric type")
+                    }
+                }
+                else
+                {
+                    asmgen.loadScaledArrayIndexIntoRegister(targetArrayIdx, elementDt, CpuRegister.A)
+                    asmgen.saveRegisterLocal(CpuRegister.X, scope!!)
+                    asmgen.out("  tax")
+                    when(elementDt) {
+                        in ByteDatatypes -> {
+                            asmgen.out(if(incr) "  inc  $asmArrayvarname,x" else "  dec  $asmArrayvarname,x")
+                        }
+                        in WordDatatypes -> {
+                            if(incr)
+                                asmgen.out(" inc  $asmArrayvarname,x |  bne  + |  inc  $asmArrayvarname+1,x |+")
+                            else
+                                asmgen.out("""
+    lda  $asmArrayvarname,x
+    bne  +
+    dec  $asmArrayvarname+1,x
++       dec  $asmArrayvarname,x 
+""")
+                        }
+                        DataType.FLOAT -> {
+                            asmgen.out("""
+                    ldy  #>$asmArrayvarname
+                    clc
+                    adc  #<$asmArrayvarname
+                    bcc  +
+                    iny
++                       jsr  floats.inc_var_f""")
+                        }
+                        else -> throw AssemblyError("weird array elt dt")
+                    }
+                    asmgen.restoreRegisterLocal(CpuRegister.X)
+                }
+            }
+        }
+    }
+}

codeGeneration/src/prog8/compiler/target/cpu6502/codegen/assignment/AsmAssignment.kt

@@ -0,0 +1,215 @@
+package prog8.compiler.target.cpu6502.codegen.assignment
+
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.compilerinterface.IMemSizer
+import prog8.compiler.target.AssemblyError
+import prog8.compiler.target.cpu6502.codegen.AsmGen
+
+
+internal enum class TargetStorageKind {
+    VARIABLE,
+    ARRAY,
+    MEMORY,
+    REGISTER,
+    STACK
+}
+
+internal enum class SourceStorageKind {
+    LITERALNUMBER,
+    VARIABLE,
+    ARRAY,
+    MEMORY,
+    REGISTER,
+    STACK,              // value is already present on stack
+    EXPRESSION,         // expression in ast-form, still to be evaluated
+}
+
+internal class AsmAssignTarget(val kind: TargetStorageKind,
+                               private val program: Program,
+                               private val asmgen: AsmGen,
+                               val datatype: DataType,
+                               val scope: Subroutine?,
+                               private val variableAsmName: String? = null,
+                               val array: ArrayIndexedExpression? = null,
+                               val memory: DirectMemoryWrite? = null,
+                               val register: RegisterOrPair? = null,
+                               val origAstTarget: AssignTarget? = null
+                               )
+{
+    val constMemoryAddress by lazy { memory?.addressExpression?.constValue(program)?.number?.toInt() ?: 0}
+    val constArrayIndexValue by lazy { array?.indexer?.constIndex() }
+    val asmVarname: String by lazy {
+        if (array == null)
+            variableAsmName!!
+        else
+            asmgen.asmVariableName(array.arrayvar)
+    }
+
+    lateinit var origAssign: AsmAssignment
+
+    init {
+        if(register!=null && datatype !in NumericDatatypes)
+            throw AssemblyError("register must be integer or float type")
+    }
+
+    companion object {
+        fun fromAstAssignment(assign: Assignment, program: Program, asmgen: AsmGen): AsmAssignTarget = with(assign.target) {
+            val idt = inferType(program)
+            if(!idt.isKnown)
+                throw AssemblyError("unknown dt")
+            val dt = idt.getOr(DataType.UNDEFINED)
+            when {
+                identifier != null -> AsmAssignTarget(TargetStorageKind.VARIABLE, program, asmgen, dt, assign.definingSubroutine, variableAsmName = asmgen.asmVariableName(identifier!!), origAstTarget =  this)
+                arrayindexed != null -> AsmAssignTarget(TargetStorageKind.ARRAY, program, asmgen, dt, assign.definingSubroutine, array = arrayindexed, origAstTarget =  this)
+                memoryAddress != null -> AsmAssignTarget(TargetStorageKind.MEMORY, program, asmgen, dt, assign.definingSubroutine, memory =  memoryAddress, origAstTarget =  this)
+                else -> throw AssemblyError("weird target")
+            }
+        }
+
+        fun fromRegisters(registers: RegisterOrPair, signed: Boolean, scope: Subroutine?, program: Program, asmgen: AsmGen): AsmAssignTarget =
+                when(registers) {
+                    RegisterOrPair.A,
+                    RegisterOrPair.X,
+                    RegisterOrPair.Y -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, if(signed) DataType.BYTE else DataType.UBYTE, scope, register = registers)
+                    RegisterOrPair.AX,
+                    RegisterOrPair.AY,
+                    RegisterOrPair.XY -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, if(signed) DataType.WORD else DataType.UWORD, scope, register = registers)
+                    RegisterOrPair.FAC1,
+                    RegisterOrPair.FAC2 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, DataType.FLOAT, scope, register = registers)
+                    RegisterOrPair.R0,
+                    RegisterOrPair.R1,
+                    RegisterOrPair.R2,
+                    RegisterOrPair.R3,
+                    RegisterOrPair.R4,
+                    RegisterOrPair.R5,
+                    RegisterOrPair.R6,
+                    RegisterOrPair.R7,
+                    RegisterOrPair.R8,
+                    RegisterOrPair.R9,
+                    RegisterOrPair.R10,
+                    RegisterOrPair.R11,
+                    RegisterOrPair.R12,
+                    RegisterOrPair.R13,
+                    RegisterOrPair.R14,
+                    RegisterOrPair.R15 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, if(signed) DataType.WORD else DataType.UWORD, scope, register = registers)
+                }
+    }
+}
+
+internal class AsmAssignSource(val kind: SourceStorageKind,
+                               private val program: Program,
+                               private val asmgen: AsmGen,
+                               val datatype: DataType,
+                               private val variableAsmName: String? = null,
+                               val array: ArrayIndexedExpression? = null,
+                               val memory: DirectMemoryRead? = null,
+                               val register: RegisterOrPair? = null,
+                               val number: NumericLiteralValue? = null,
+                               val expression: Expression? = null
+)
+{
+    val constMemoryAddress by lazy { memory?.addressExpression?.constValue(program)?.number?.toInt() ?: 0}
+    val constArrayIndexValue by lazy { array?.indexer?.constIndex() }
+
+    val asmVarname: String
+        get() = if(array==null)
+            variableAsmName!!
+        else
+            asmgen.asmVariableName(array.arrayvar)
+
+    companion object {
+        fun fromAstSource(indexer: ArrayIndex, program: Program, asmgen: AsmGen): AsmAssignSource = fromAstSource(indexer.indexExpr, program, asmgen)
+
+        fun fromAstSource(value: Expression, program: Program, asmgen: AsmGen): AsmAssignSource {
+            val cv = value.constValue(program)
+            if(cv!=null)
+                return AsmAssignSource(SourceStorageKind.LITERALNUMBER, program, asmgen, cv.type, number = cv)
+
+            return when(value) {
+                is NumericLiteralValue -> AsmAssignSource(SourceStorageKind.LITERALNUMBER, program, asmgen, value.type, number = cv)
+                is StringLiteralValue -> throw AssemblyError("string literal value should not occur anymore for asm generation")
+                is ArrayLiteralValue -> throw AssemblyError("array literal value should not occur anymore for asm generation")
+                is IdentifierReference -> {
+                    val dt = value.inferType(program).getOr(DataType.UNDEFINED)
+                    val varName=asmgen.asmVariableName(value)
+                    // special case: "cx16.r[0-15]" are 16-bits virtual registers of the commander X16 system
+                    if(dt == DataType.UWORD && varName.lowercase().startsWith("cx16.r")) {
+                        val regStr = varName.lowercase().substring(5)
+                        val reg = RegisterOrPair.valueOf(regStr.uppercase())
+                        AsmAssignSource(SourceStorageKind.REGISTER, program, asmgen, dt, register = reg)
+                    } else {
+                        AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, dt, variableAsmName = varName)
+                    }
+                }
+                is DirectMemoryRead -> {
+                    AsmAssignSource(SourceStorageKind.MEMORY, program, asmgen, DataType.UBYTE, memory = value)
+                }
+                is ArrayIndexedExpression -> {
+                    val dt = value.inferType(program).getOr(DataType.UNDEFINED)
+                    AsmAssignSource(SourceStorageKind.ARRAY, program, asmgen, dt, array = value)
+                }
+                is FunctionCall -> {
+                    when (val sub = value.target.targetStatement(program)) {
+                        is Subroutine -> {
+                            val returnType = sub.returntypes.zip(sub.asmReturnvaluesRegisters).firstOrNull { rr -> rr.second.registerOrPair != null || rr.second.statusflag!=null }?.first
+                                    ?: throw AssemblyError("can't translate zero return values in assignment")
+
+                            AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, returnType, expression = value)
+                        }
+                        is BuiltinFunctionStatementPlaceholder -> {
+                            val returnType = value.inferType(program)
+                            if(!returnType.isKnown)
+                                throw AssemblyError("unknown dt")
+                            AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, returnType.getOr(DataType.UNDEFINED), expression = value)
+                        }
+                        else -> {
+                            throw AssemblyError("weird call")
+                        }
+                    }
+                }
+                else -> {
+                    val dt = value.inferType(program)
+                    if(!dt.isKnown)
+                        throw AssemblyError("unknown dt")
+                    AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, dt.getOr(DataType.UNDEFINED), expression = value)
+                }
+            }
+        }
+    }
+
+    fun adjustSignedUnsigned(target: AsmAssignTarget): AsmAssignSource {
+        // allow some signed/unsigned relaxations
+
+        fun withAdjustedDt(newType: DataType) =
+                AsmAssignSource(kind, program, asmgen, newType, variableAsmName, array, memory, register, number, expression)
+
+        if(target.datatype!=datatype) {
+            if(target.datatype in ByteDatatypes && datatype in ByteDatatypes) {
+                return withAdjustedDt(target.datatype)
+            } else if(target.datatype in WordDatatypes && datatype in WordDatatypes) {
+                return withAdjustedDt(target.datatype)
+            }
+        }
+        return this
+    }
+
+}
+
+
+internal class AsmAssignment(val source: AsmAssignSource,
+                             val target: AsmAssignTarget,
+                             val isAugmentable: Boolean,
+                             memsizer: IMemSizer,
+                             val position: Position) {
+
+    init {
+        if(target.register !in arrayOf(RegisterOrPair.XY, RegisterOrPair.AX, RegisterOrPair.AY))
+            require(source.datatype != DataType.UNDEFINED) { "must not be placeholder/undefined datatype" }
+            require(memsizer.memorySize(source.datatype) <= memsizer.memorySize(target.datatype)) {
+                "source storage size must be less or equal to target datatype storage size"
+            }
+    }
+}

codeGeneration/src/prog8/compiler/target/cx16/CX16MachineDefinition.kt

@@ -0,0 +1,124 @@
+package prog8.compiler.target.cx16
+
+import prog8.compiler.target.c64.C64MachineDefinition
+import prog8.compiler.target.cbm.viceMonListPostfix
+import prog8.compilerinterface.*
+import java.io.IOException
+import java.nio.file.Path
+
+
+object CX16MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU65c02
+
+    // 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 = 0x8000
+
+    // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
+    override val ESTACK_LO = 0x0400        //  $0400-$04ff inclusive
+    override val ESTACK_HI = 0x0500        //  $0500-$05ff inclusive
+
+    override lateinit var zeropage: Zeropage
+
+    override fun getFloat(num: Number) = C64MachineDefinition.Mflpt5.fromNumber(num)
+    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
+    }
+
+    override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
+        val emulatorName: String
+        val extraArgs: List<String>
+
+        when(selectedEmulator) {
+            1 -> {
+                emulatorName = "x16emu"
+                extraArgs = emptyList()
+            }
+            2 -> {
+                emulatorName = "box16"
+                extraArgs = listOf("-sym", "${programNameWithPath}.$viceMonListPostfix")
+            }
+            else -> {
+                System.err.println("Cx16 target only supports x16emu and box16 emulators.")
+                return
+            }
+        }
+
+        for(emulator in listOf(emulatorName)) {
+            println("\nStarting Commander X16 emulator $emulator...")
+            val cmdline = listOf(emulator, "-scale", "2", "-run", "-prg", "${programNameWithPath}.prg") + extraArgs
+            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 < 0x9f00 || address in 0xa000..0xbfff
+
+    override fun initializeZeropage(compilerOptions: CompilationOptions) {
+        zeropage = CX16Zeropage(compilerOptions)
+    }
+
+    // 6502 opcodes (including aliases and illegal opcodes), these cannot be used as variable or label names
+    override val opcodeNames = setOf("adc", "and", "asl", "bcc", "bcs",
+            "beq", "bge", "bit", "blt", "bmi", "bne", "bpl", "brk", "bvc", "bvs", "clc",
+            "cld", "cli", "clv", "cmp", "cpx", "cpy", "dec", "dex", "dey",
+            "eor", "gcc", "gcs", "geq", "gge", "glt", "gmi", "gne", "gpl", "gvc", "gvs",
+            "inc", "inx", "iny", "jmp", "jsr",
+            "lda", "ldx", "ldy", "lsr", "nop", "ora", "pha", "php",
+            "pla", "plp", "rol", "ror", "rti", "rts", "sbc",
+            "sec", "sed", "sei",
+            "sta", "stx", "sty", "tax", "tay", "tsx", "txa", "txs", "tya",
+            "bra", "phx", "phy", "plx", "ply", "stz", "trb", "tsb", "bbr", "bbs",
+            "rmb", "smb", "stp", "wai")
+
+
+    class CX16Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+        override val SCRATCH_B1 = 0x7a      // temp storage for a single byte
+        override val SCRATCH_REG = 0x7b     // temp storage for a register, must be B1+1
+        override val SCRATCH_W1 = 0x7c      // temp storage 1 for a word  $7c+$7d
+        override val SCRATCH_W2 = 0x7e      // temp storage 2 for a word  $7e+$7f
+
+
+        init {
+            if (options.floats && options.zeropage !in arrayOf(ZeropageType.BASICSAFE, ZeropageType.DONTUSE ))
+                throw InternalCompilerException("when floats are enabled, zero page type should be 'basicsafe' or 'dontuse'")
+
+            // the addresses 0x02 to 0x21 (inclusive) are taken for sixteen virtual 16-bit api registers.
+
+            when (options.zeropage) {
+                ZeropageType.FULL -> {
+                    free.addAll(0x22..0xff)
+                }
+                ZeropageType.KERNALSAFE -> {
+                    free.addAll(0x22..0x7f)
+                    free.addAll(0xa9..0xff)
+                }
+                ZeropageType.BASICSAFE -> {
+                    free.addAll(0x22..0x7f)
+                }
+                ZeropageType.DONTUSE -> {
+                    free.clear() // don't use zeropage at all
+                }
+                else -> throw InternalCompilerException("for this machine target, zero page type 'floatsafe' is not available. ${options.zeropage}")
+            }
+
+            removeReservedFromFreePool()
+        }
+    }
+}

codeGeneration/test/AsmGenTests.kt

@@ -0,0 +1,151 @@
+package prog8tests.asmgen
+
+import org.hamcrest.MatcherAssert.assertThat
+import org.hamcrest.Matchers.equalTo
+import org.junit.jupiter.api.Test
+import org.junit.jupiter.api.TestInstance
+import prog8.ast.Module
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.base.Position
+import prog8.ast.base.RegisterOrPair
+import prog8.ast.base.VarDeclType
+import prog8.ast.expressions.AddressOf
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.*
+import prog8.compiler.target.C64Target
+import prog8.compiler.target.c64.C64MachineDefinition
+import prog8.compiler.target.cpu6502.codegen.AsmGen
+import prog8.compilerinterface.*
+import prog8.parser.SourceCode
+import prog8tests.asmgen.helpers.DummyFunctions
+import prog8tests.asmgen.helpers.DummyMemsizer
+import prog8tests.asmgen.helpers.DummyStringEncoder
+import prog8tests.asmgen.helpers.ErrorReporterForTests
+import java.nio.file.Path
+
+
+@TestInstance(TestInstance.Lifecycle.PER_CLASS)
+class TestAsmGen6502 {
+
+    private fun createTestProgram(): Program {
+        /*
+main  {
+
+label_outside:
+    uword var_outside
+
+    sub start () {
+        uword localvar = 1234
+        uword tgt
+
+locallabel:
+        tgt = localvar
+        tgt = &locallabel
+        tgt = &var_outside
+        tgt = &label_outside
+        tgt = &main.start.localvar
+        tgt = &main.start.locallabel
+        tgt = &main.var_outside
+        tgt = &main.label_outside
+    }
+}
+
+         */
+        val varInSub = VarDecl(VarDeclType.VAR, DataType.UWORD, ZeropageWish.DONTCARE, null, "localvar", NumericLiteralValue.optimalInteger(1234, Position.DUMMY), false, false, false, Position.DUMMY)
+        val var2InSub = VarDecl(VarDeclType.VAR, DataType.UWORD, ZeropageWish.DONTCARE, null, "tgt", null, false, false, false, Position.DUMMY)
+        val labelInSub = Label("locallabel", Position.DUMMY)
+
+        val tgt = AssignTarget(IdentifierReference(listOf("tgt"), Position.DUMMY), null, null, Position.DUMMY)
+        val assign1 = Assignment(tgt, IdentifierReference(listOf("localvar"), Position.DUMMY), Position.DUMMY)
+        val assign2 = Assignment(tgt, AddressOf(IdentifierReference(listOf("locallabel"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign3 = Assignment(tgt, AddressOf(IdentifierReference(listOf("var_outside"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign4 = Assignment(tgt, AddressOf(IdentifierReference(listOf("label_outside"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign5 = Assignment(tgt, AddressOf(IdentifierReference(listOf("main","start","localvar"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign6 = Assignment(tgt, AddressOf(IdentifierReference(listOf("main","start","locallabel"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign7 = Assignment(tgt, AddressOf(IdentifierReference(listOf("main","var_outside"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+        val assign8 = Assignment(tgt, AddressOf(IdentifierReference(listOf("main","label_outside"), Position.DUMMY), Position.DUMMY), Position.DUMMY)
+
+        val statements = mutableListOf(varInSub, var2InSub, labelInSub, assign1, assign2, assign3, assign4, assign5, assign6, assign7, assign8)
+        val subroutine = Subroutine("start", mutableListOf(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, statements, Position.DUMMY)
+        val labelInBlock = Label("label_outside", Position.DUMMY)
+        val varInBlock = VarDecl(VarDeclType.VAR, DataType.UWORD, ZeropageWish.DONTCARE, null, "var_outside", null, false, false, false, Position.DUMMY)
+        val block = Block("main", null, mutableListOf(labelInBlock, varInBlock, subroutine), false, Position.DUMMY)
+
+        val module = Module(mutableListOf(block), Position.DUMMY, SourceCode.Generated("test"))
+        val program = Program("test", DummyFunctions, DummyMemsizer, DummyStringEncoder)
+            .addModule(module)
+        module.linkIntoProgram(program)
+        return program
+    }
+
+    private fun createTestAsmGen(program: Program): AsmGen {
+        val errors = ErrorReporterForTests()
+        val options = CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, true, C64Target)
+        val zp = C64MachineDefinition.C64Zeropage(options)
+        val asmgen = AsmGen(program, errors, zp, options, C64Target, Path.of(""))
+        return asmgen
+    }
+
+    @Test
+    fun testSymbolNameFromStrings() {
+        val program = createTestProgram()
+        val asmgen = createTestAsmGen(program)
+
+        assertThat(asmgen.asmSymbolName("name"), equalTo("name"))
+        assertThat(asmgen.asmSymbolName("<name>"), equalTo("prog8_name"))
+        assertThat(asmgen.asmSymbolName(RegisterOrPair.R15), equalTo("cx16.r15"))
+        assertThat(asmgen.asmSymbolName(listOf("a", "b", "name")), equalTo("a.b.name"))
+        assertThat(asmgen.asmVariableName("name"), equalTo("name"))
+        assertThat(asmgen.asmVariableName("<name>"), equalTo("prog8_name"))
+        assertThat(asmgen.asmVariableName(listOf("a", "b", "name")), equalTo("a.b.name"))
+    }
+
+    @Test
+    fun testSymbolNameFromVarIdentifier() {
+        val program = createTestProgram()
+        val asmgen = createTestAsmGen(program)
+        val sub = program.entrypoint
+
+        // local variable
+        val localvarIdent = sub.statements.filterIsInstance<Assignment>().first { it.value is IdentifierReference }.value as IdentifierReference
+        assertThat(asmgen.asmSymbolName(localvarIdent), equalTo("localvar"))
+        assertThat(asmgen.asmVariableName(localvarIdent), equalTo("localvar"))
+        val localvarIdentScoped = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("main", "start", "localvar") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(localvarIdentScoped), equalTo("main.start.localvar"))
+        assertThat(asmgen.asmVariableName(localvarIdentScoped), equalTo("main.start.localvar"))
+
+        // variable from outer scope (note that for Variables, no scoping prefix symbols are required,
+        //   because they're not outputted as locally scoped symbols for the assembler
+        val scopedVarIdent = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("var_outside") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(scopedVarIdent), equalTo("main.var_outside"))
+        assertThat(asmgen.asmVariableName(scopedVarIdent), equalTo("var_outside"))
+        val scopedVarIdentScoped = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("main", "var_outside") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(scopedVarIdentScoped), equalTo("main.var_outside"))
+        assertThat(asmgen.asmVariableName(scopedVarIdentScoped), equalTo("main.var_outside"))
+    }
+
+    @Test
+    fun testSymbolNameFromLabelIdentifier() {
+        val program = createTestProgram()
+        val asmgen = createTestAsmGen(program)
+        val sub = program.entrypoint
+
+        // local label
+        val localLabelIdent = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("locallabel") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(localLabelIdent), equalTo("_locallabel"))
+        assertThat("as a variable it uses different naming rules (no underscore prefix)", asmgen.asmVariableName(localLabelIdent), equalTo("locallabel"))
+        val localLabelIdentScoped = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("main","start","locallabel") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(localLabelIdentScoped), equalTo("main.start._locallabel"))
+        assertThat("as a variable it uses different naming rules (no underscore prefix)", asmgen.asmVariableName(localLabelIdentScoped), equalTo("main.start.locallabel"))
+
+        // label from outer scope needs sope prefixes because it is outputted as a locally scoped symbol for the assembler
+        val scopedLabelIdent = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("label_outside") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(scopedLabelIdent), equalTo("main._label_outside"))
+        assertThat("as a variable it uses different naming rules (no underscore prefix)", asmgen.asmVariableName(scopedLabelIdent), equalTo("label_outside"))
+        val scopedLabelIdentScoped = (sub.statements.filterIsInstance<Assignment>().first { (it.value as? AddressOf)?.identifier?.nameInSource==listOf("main","label_outside") }.value as AddressOf).identifier
+        assertThat(asmgen.asmSymbolName(scopedLabelIdentScoped), equalTo("main._label_outside"))
+        assertThat("as a variable it uses different naming rules (no underscore prefix)", asmgen.asmVariableName(scopedLabelIdentScoped), equalTo("main.label_outside"))
+    }
+}

codeGeneration/test/helpers/Dummies.kt

@@ -0,0 +1,37 @@
+package prog8tests.asmgen.helpers
+
+import prog8.ast.IBuiltinFunctions
+import prog8.ast.base.Position
+import prog8.ast.expressions.Expression
+import prog8.ast.expressions.InferredTypes
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.compilerinterface.IMemSizer
+import prog8.ast.base.DataType
+import prog8.compilerinterface.IStringEncoding
+
+
+internal val DummyFunctions = object : IBuiltinFunctions {
+    override val names: Set<String> = emptySet()
+    override val purefunctionNames: Set<String> = emptySet()
+    override fun constValue(
+        name: String,
+        args: List<Expression>,
+        position: Position,
+    ): NumericLiteralValue? = null
+
+    override fun returnType(name: String, args: MutableList<Expression>) = InferredTypes.InferredType.unknown()
+}
+
+internal val DummyMemsizer = object : IMemSizer {
+    override fun memorySize(dt: DataType): Int = 0
+}
+
+internal val DummyStringEncoder = object : IStringEncoding {
+    override fun encodeString(str: String, altEncoding: Boolean): List<Short> {
+        return emptyList()
+    }
+
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean): String {
+        return ""
+    }
+}

codeGeneration/test/helpers/ErrorReporterForTests.kt

@@ -0,0 +1,30 @@
+package prog8tests.asmgen.helpers
+
+import prog8.ast.base.Position
+import prog8.compilerinterface.IErrorReporter
+
+internal class ErrorReporterForTests(private val throwExceptionAtReportIfErrors: Boolean=true): IErrorReporter {
+
+
+    val errors = mutableListOf<String>()
+    val warnings = mutableListOf<String>()
+
+    override fun err(msg: String, position: Position) {
+        errors.add("${position.toClickableStr()} $msg")
+    }
+
+    override fun warn(msg: String, position: Position) {
+        warnings.add("${position.toClickableStr()} $msg")
+    }
+
+    override fun noErrors(): Boolean  = errors.isEmpty()
+
+    override fun report() {
+        warnings.forEach { println("UNITTEST COMPILATION REPORT: WARNING: $it") }
+        errors.forEach { println("UNITTEST COMPILATION REPORT: ERROR: $it") }
+        if(throwExceptionAtReportIfErrors)
+            finalizeNumErrors(errors.size, warnings.size)
+        errors.clear()
+        warnings.clear()
+    }
+}

codeOptimizers/build.gradle

@@ -0,0 +1,54 @@
+
+plugins {
+    id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm"
+}
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+dependencies {
+    implementation project(':compilerInterfaces')
+    implementation project(':compilerAst')
+    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
+    // implementation "org.jetbrains.kotlin:kotlin-reflect"
+
+    testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
+    testImplementation 'org.junit.jupiter:junit-jupiter-api:5.7.2'
+    testImplementation 'org.hamcrest:hamcrest:2.2'
+    testRuntimeOnly 'org.junit.jupiter:junit-jupiter-engine:5.7.2'
+
+}
+
+sourceSets {
+    main {
+        java {
+            srcDirs = ["${project.projectDir}/src"]
+        }
+        resources {
+            srcDirs = ["${project.projectDir}/res"]
+        }
+    }
+    test {
+        java {
+            srcDirs = ["${project.projectDir}/test"]
+        }
+    }
+}
+
+test {
+    // Enable JUnit 5 (Gradle 4.6+).
+    useJUnitPlatform()
+
+    // Always run tests, even when nothing changed.
+    dependsOn 'cleanTest'
+
+    // Show test results.
+    testLogging {
+        events "skipped", "failed"
+    }
+}

codeOptimizers/codeOptimizers.iml

@@ -0,0 +1,18 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="JAVA_MODULE" version="4">
+  <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://$MODULE_DIR$/test" isTestSource="true" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
+    </content>
+    <orderEntry type="jdk" jdkName="openjdk-11" jdkType="JavaSDK" />
+    <orderEntry type="sourceFolder" forTests="false" />
+    <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
+    <orderEntry type="module" module-name="compilerInterfaces" />
+    <orderEntry type="module" module-name="compilerAst" />
+    <orderEntry type="library" scope="TEST" name="hamcrest" level="project" />
+    <orderEntry type="library" name="junit.jupiter" level="project" />
+  </component>
+</module>

codeOptimizers/src/prog8/optimizer/BinExprSplitter.kt

@@ -0,0 +1,94 @@
+package prog8.optimizer
+
+import prog8.ast.IStatementContainer
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.augmentAssignmentOperators
+import prog8.ast.statements.AssignTarget
+import prog8.ast.statements.Assignment
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.isInRegularRAMof
+
+
+class BinExprSplitter(private val program: Program, private val options: CompilationOptions, private val compTarget: ICompilationTarget) : AstWalker() {
+
+//    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+// TODO somehow if we do this, the resulting code for some programs (cube3d.p8) gets hundreds of bytes larger...: [ IS THIS STILL TRUE AFTER ALL CHANGES? ]
+//        if(decl.type==VarDeclType.VAR ) {
+//            val binExpr = decl.value as? BinaryExpression
+//            if (binExpr != null && binExpr.operator in augmentAssignmentOperators) {
+//                // split into a vardecl with just the left expression, and an aug. assignment with the right expression.
+//                val augExpr = BinaryExpression(IdentifierReference(listOf(decl.name), decl.position), binExpr.operator, binExpr.right, binExpr.position)
+//                val target = AssignTarget(IdentifierReference(listOf(decl.name), decl.position), null, null, decl.position)
+//                val assign = Assignment(target, augExpr, binExpr.position)
+//                println("SPLIT VARDECL $decl")
+//                return listOf(
+//                        IAstModification.SetExpression({ decl.value = it }, binExpr.left, decl),
+//                        IAstModification.InsertAfter(decl, assign, parent)
+//                )
+//            }
+//        }
+//        return noModifications
+//    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+
+        val binExpr = assignment.value as? BinaryExpression
+        if (binExpr != null) {
+
+            if(binExpr.inferType(program).istype(DataType.FLOAT) && !options.optimizeFloatExpressions)
+                return noModifications
+
+
+/*
+
+Reduce the complexity of a (binary) expression that has to be evaluated on the eval stack,
+by attempting to splitting it up into individual simple steps.
+We only consider a binary expression *one* level deep (so the operands must not be a combined expression)
+
+
+X =      BinExpr                                    X   =   LeftExpr
+        <operator>                                     followed by
+          /   \             IF 'X' not used         X   =   BinExpr
+         /     \             IN expression ==>             <operator>
+        /       \                                           /   \
+    LeftExpr.  RightExpr.                                  /     \
+                                                          X     RightExpr.
+
+
+ */
+            if(binExpr.operator in augmentAssignmentOperators && isSimpleTarget(assignment.target)) {
+                if(assignment.target isSameAs binExpr.left || assignment.target isSameAs binExpr.right)
+                    return noModifications
+
+                if(binExpr.right.isSimple && !assignment.isAugmentable) {
+                    val firstAssign = Assignment(assignment.target.copy(), binExpr.left, binExpr.left.position)
+                    val targetExpr = assignment.target.toExpression()
+                    val augExpr = BinaryExpression(targetExpr, binExpr.operator, binExpr.right, binExpr.right.position)
+                    return listOf(
+                        IAstModification.ReplaceNode(binExpr, augExpr, assignment),
+                        IAstModification.InsertBefore(assignment, firstAssign, assignment.parent as IStatementContainer)
+                    )
+                }
+            }
+
+            // TODO further unraveling of binary expression trees into flat statements.
+            // however this should probably be done in a more generic way to also service
+            // the expressiontrees that are not used in an assignment statement...
+        }
+
+        return noModifications
+    }
+
+    private fun isSimpleTarget(target: AssignTarget) =
+            if (target.identifier!=null || target.memoryAddress!=null)
+                target.isInRegularRAMof(compTarget.machine)
+            else
+                false
+
+}

codeOptimizers/src/prog8/optimizer/ConstExprEvaluator.kt

@@ -0,0 +1,278 @@
+package prog8.optimizer
+
+import prog8.ast.base.*
+import prog8.ast.expressions.Expression
+import prog8.ast.expressions.NumericLiteralValue
+import kotlin.math.pow
+
+
+class ConstExprEvaluator {
+
+    fun evaluate(left: NumericLiteralValue, operator: String, right: NumericLiteralValue): Expression {
+        try {
+            return when(operator) {
+                "+" -> plus(left, right)
+                "-" -> minus(left, right)
+                "*" -> multiply(left, right)
+                "/" -> divide(left, right)
+                "%" -> remainder(left, right)
+                "**" -> power(left, right)
+                "&" -> bitwiseand(left, right)
+                "|" -> bitwiseor(left, right)
+                "^" -> bitwisexor(left, right)
+                "and" -> logicaland(left, right)
+                "or" -> logicalor(left, right)
+                "xor" -> logicalxor(left, right)
+                "<" -> NumericLiteralValue.fromBoolean(left < right, left.position)
+                ">" -> NumericLiteralValue.fromBoolean(left > right, left.position)
+                "<=" -> NumericLiteralValue.fromBoolean(left <= right, left.position)
+                ">=" -> NumericLiteralValue.fromBoolean(left >= right, left.position)
+                "==" -> NumericLiteralValue.fromBoolean(left == right, left.position)
+                "!=" -> NumericLiteralValue.fromBoolean(left != right, left.position)
+                "<<" -> shiftedleft(left, right)
+                ">>" -> shiftedright(left, right)
+                else -> throw FatalAstException("const evaluation for invalid operator $operator")
+            }
+        } catch (ax: FatalAstException) {
+            throw ExpressionError(ax.message, left.position)
+        }
+    }
+
+    private fun shiftedright(left: NumericLiteralValue, amount: NumericLiteralValue): Expression {
+        if(left.type !in IntegerDatatypes || amount.type !in IntegerDatatypes)
+            throw ExpressionError("cannot compute $left >> $amount", left.position)
+        val result =
+                if(left.type== DataType.UBYTE || left.type== DataType.UWORD)
+                    left.number.toInt().ushr(amount.number.toInt())
+                else
+                    left.number.toInt().shr(amount.number.toInt())
+        return NumericLiteralValue(left.type, result, left.position)
+    }
+
+    private fun shiftedleft(left: NumericLiteralValue, amount: NumericLiteralValue): Expression {
+        if(left.type !in IntegerDatatypes || amount.type !in IntegerDatatypes)
+            throw ExpressionError("cannot compute $left << $amount", left.position)
+        val result = left.number.toInt().shl(amount.number.toInt())
+        return NumericLiteralValue(left.type, result, left.position)
+    }
+
+    private fun logicalxor(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot compute $left locical-bitxor $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean((left.number.toInt() != 0) xor (right.number.toInt() != 0), left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean((left.number.toInt() != 0) xor (right.number.toDouble() != 0.0), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean((left.number.toDouble() != 0.0) xor (right.number.toInt() != 0), left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean((left.number.toDouble() != 0.0) xor (right.number.toDouble() != 0.0), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun logicalor(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot compute $left locical-or $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 || right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 || right.number.toDouble() != 0.0, left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number.toDouble() != 0.0 || right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toDouble() != 0.0 || right.number.toDouble() != 0.0, left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun logicaland(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot compute $left locical-and $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 && right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 && right.number.toDouble() != 0.0, left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number.toDouble() != 0.0 && right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toDouble() != 0.0 && right.number.toDouble() != 0.0, left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun bitwisexor(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        if(left.type== DataType.UBYTE) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() xor (right.number.toInt() and 255)).toShort(), left.position)
+            }
+        } else if(left.type== DataType.UWORD) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UWORD, left.number.toInt() xor right.number.toInt(), left.position)
+            }
+        }
+        throw ExpressionError("cannot calculate $left ^ $right", left.position)
+    }
+
+    private fun bitwiseor(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        if(left.type== DataType.UBYTE) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() or (right.number.toInt() and 255)).toShort(), left.position)
+            }
+        } else if(left.type== DataType.UWORD) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UWORD, left.number.toInt() or right.number.toInt(), left.position)
+            }
+        }
+        throw ExpressionError("cannot calculate $left | $right", left.position)
+    }
+
+    private fun bitwiseand(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        if(left.type== DataType.UBYTE) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() and (right.number.toInt() and 255)).toShort(), left.position)
+            }
+        } else if(left.type== DataType.UWORD) {
+            if(right.type in IntegerDatatypes) {
+                return NumericLiteralValue(DataType.UWORD, left.number.toInt() and right.number.toInt(), left.position)
+            }
+        }
+        throw ExpressionError("cannot calculate $left & $right", left.position)
+    }
+
+    private fun power(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot calculate $left ** $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.optimalNumeric(left.number.toInt().toDouble().pow(right.number.toInt()), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt().toDouble().pow(right.number.toDouble()), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble().pow(right.number.toInt()), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble().pow(right.number.toDouble()), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun plus(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot add $left and $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.optimalInteger(left.number.toInt() + right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() + right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() + right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() + right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun minus(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot subtract $left and $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.optimalInteger(left.number.toInt() - right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() - right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() - right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() - right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun multiply(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot multiply ${left.type} and ${right.type}"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue.optimalInteger(left.number.toInt() * right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() * right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() * right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toDouble() * right.number.toDouble(), left.position)
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun divideByZeroError(pos: Position): Unit =
+            throw ExpressionError("division by zero", pos)
+
+    private fun divide(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot divide $left by $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> {
+                    if(right.number.toInt()==0) divideByZeroError(right.position)
+                    val result: Int = left.number.toInt() / right.number.toInt()
+                    NumericLiteralValue.optimalInteger(result, left.position)
+                }
+                DataType.FLOAT -> {
+                    if(right.number.toDouble()==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toInt() / right.number.toDouble(), left.position)
+                }
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> {
+                    if(right.number.toInt()==0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toDouble() / right.number.toInt(), left.position)
+                }
+                DataType.FLOAT -> {
+                    if(right.number.toDouble()==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toDouble() / right.number.toDouble(), left.position)
+                }
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+
+    private fun remainder(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
+        val error = "cannot compute remainder of $left by $right"
+        return when (left.type) {
+            in IntegerDatatypes -> when (right.type) {
+                in IntegerDatatypes -> {
+                    if(right.number.toInt()==0) divideByZeroError(right.position)
+                    NumericLiteralValue.optimalNumeric(left.number.toInt().toDouble() % right.number.toInt().toDouble(), left.position)
+                }
+                DataType.FLOAT -> {
+                    if(right.number.toDouble()==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toInt() % right.number.toDouble(), left.position)
+                }
+                else -> throw ExpressionError(error, left.position)
+            }
+            DataType.FLOAT -> when (right.type) {
+                in IntegerDatatypes -> {
+                    if(right.number.toInt()==0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toDouble() % right.number.toInt(), left.position)
+                }
+                DataType.FLOAT -> {
+                    if(right.number.toDouble()==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toDouble() % right.number.toDouble(), left.position)
+                }
+                else -> throw ExpressionError(error, left.position)
+            }
+            else -> throw ExpressionError(error, left.position)
+        }
+    }
+}

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -0,0 +1,502 @@
+package prog8.optimizer
+
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.Assignment
+import prog8.ast.statements.ForLoop
+import prog8.ast.statements.VarDecl
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import kotlin.math.pow
+
+
+class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
+
+    override fun before(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
+        // @( &thing )  -->  thing
+        val addrOf = memread.addressExpression as? AddressOf
+        return if(addrOf!=null)
+            listOf(IAstModification.ReplaceNode(memread, addrOf.identifier, parent))
+        else
+            noModifications
+    }
+
+    override fun after(expr: PrefixExpression, parent: Node): Iterable<IAstModification> {
+        // Try to turn a unary prefix expression into a single constant value.
+        // Compile-time constant sub expressions will be evaluated on the spot.
+        // For instance, the expression for "- 4.5" will be optimized into the float literal -4.5
+        val subexpr = expr.expression
+        if (subexpr is NumericLiteralValue) {
+            // accept prefixed literal values (such as -3, not true)
+            return when (expr.operator) {
+                "+" -> listOf(IAstModification.ReplaceNode(expr, subexpr, parent))
+                "-" -> when (subexpr.type) {
+                    in IntegerDatatypes -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue.optimalInteger(-subexpr.number.toInt(), subexpr.position),
+                                parent))
+                    }
+                    DataType.FLOAT -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue(DataType.FLOAT, -subexpr.number.toDouble(), subexpr.position),
+                                parent))
+                    }
+                    else -> throw ExpressionError("can only take negative of int or float", subexpr.position)
+                }
+                "~" -> when (subexpr.type) {
+                    DataType.BYTE -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue(DataType.BYTE, subexpr.number.toInt().inv(), subexpr.position),
+                                parent))
+                    }
+                    DataType.UBYTE -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue(DataType.UBYTE, subexpr.number.toInt().inv() and 255, subexpr.position),
+                                parent))
+                    }
+                    DataType.WORD -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue(DataType.WORD, subexpr.number.toInt().inv(), subexpr.position),
+                                parent))
+                    }
+                    DataType.UWORD -> {
+                        listOf(IAstModification.ReplaceNode(expr,
+                                NumericLiteralValue(DataType.UWORD, subexpr.number.toInt().inv() and 65535, subexpr.position),
+                                parent))
+                    }
+                    else -> throw ExpressionError("can only take bitwise inversion of int", subexpr.position)
+                }
+                "not" -> {
+                    listOf(IAstModification.ReplaceNode(expr,
+                            NumericLiteralValue.fromBoolean(subexpr.number.toDouble() == 0.0, subexpr.position),
+                            parent))
+                }
+                else -> throw ExpressionError(expr.operator, subexpr.position)
+            }
+        }
+        return noModifications
+    }
+
+    /**
+     * Try to constfold a binary expression.
+     * Compile-time constant sub expressions will be evaluated on the spot.
+     * For instance, "9 * (4 + 2)" will be optimized into the integer literal 54.
+     *
+     * More complex stuff: reordering to group constants:
+     * If one of our operands is a Constant,
+     *   and the other operand is a Binary expression,
+     *   and one of ITS operands is a Constant,
+     *   and ITS other operand is NOT a Constant,
+     *   ...it may be possible to rewrite the expression to group the two Constants together,
+     *      to allow them to be const-folded away.
+     *
+     *  examples include:
+     *        (X / c1) * c2  ->  X / (c2/c1)
+     *        (X + c1) - c2  ->  X + (c1-c2)
+     */
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        val leftconst = expr.left.constValue(program)
+        val rightconst = expr.right.constValue(program)
+        val modifications = mutableListOf<IAstModification>()
+
+        if(expr.operator == "**" && leftconst!=null) {
+            // optimize various simple cases of ** :
+            //  optimize away 1 ** x into just 1 and 0 ** x into just 0
+            //  optimize 2 ** x into (1<<x)  if both operands are integer.
+            val leftDt = leftconst.inferType(program).getOr(DataType.UNDEFINED)
+            when (leftconst.number.toDouble()) {
+                0.0 -> {
+                    val value = NumericLiteralValue(leftDt, 0, expr.position)
+                    modifications += IAstModification.ReplaceNode(expr, value, parent)
+                }
+                1.0 -> {
+                    val value = NumericLiteralValue(leftDt, 1, expr.position)
+                    modifications += IAstModification.ReplaceNode(expr, value, parent)
+                }
+                2.0 -> {
+                    if(rightconst!=null) {
+                        val value = NumericLiteralValue(leftDt, 2.0.pow(rightconst.number.toDouble()), expr.position)
+                        modifications += IAstModification.ReplaceNode(expr, value, parent)
+                    } else {
+                        val rightDt = expr.right.inferType(program).getOr(DataType.UNDEFINED)
+                        if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                            val targetDt =
+                                when (parent) {
+                                    is Assignment -> parent.target.inferType(program).getOr(DataType.UNDEFINED)
+                                    is VarDecl -> parent.datatype
+                                    else -> leftDt
+                                }
+                            val one = NumericLiteralValue(targetDt, 1, expr.position)
+                            val shift = BinaryExpression(one, "<<", expr.right, expr.position)
+                            modifications += IAstModification.ReplaceNode(expr, shift, parent)
+                        }
+                    }
+                }
+            }
+        }
+
+        if(expr.inferType(program) istype DataType.FLOAT) {
+            val subExpr: BinaryExpression? = when {
+                leftconst != null -> expr.right as? BinaryExpression
+                rightconst != null -> expr.left as? BinaryExpression
+                else -> null
+            }
+            if (subExpr != null) {
+                val subleftconst = subExpr.left.constValue(program)
+                val subrightconst = subExpr.right.constValue(program)
+                if ((subleftconst != null && subrightconst == null) || (subleftconst == null && subrightconst != null)) {
+                    // try reordering.
+                    val change = groupTwoFloatConstsTogether(
+                        expr, subExpr,
+                        leftconst != null, rightconst != null,
+                        subleftconst != null, subrightconst != null
+                    )
+                    if (change != null)
+                        modifications += change
+                }
+            }
+        }
+
+        // const fold when both operands are a const
+        if(leftconst != null && rightconst != null) {
+            val evaluator = ConstExprEvaluator()
+            val result = evaluator.evaluate(leftconst, expr.operator, rightconst)
+            modifications += IAstModification.ReplaceNode(expr, result, parent)
+        }
+
+        return modifications
+    }
+
+    override fun after(array: ArrayLiteralValue, parent: Node): Iterable<IAstModification> {
+        // because constant folding can result in arrays that are now suddenly capable
+        // of telling the type of all their elements (for instance, when they contained -2 which
+        // was a prefix expression earlier), we recalculate the array's datatype.
+        if(array.type.isKnown)
+            return noModifications
+
+        // if the array literalvalue is inside an array vardecl, take the type from that
+        // otherwise infer it from the elements of the array
+        val vardeclType = (array.parent as? VarDecl)?.datatype
+        if(vardeclType!=null) {
+            val newArray = array.cast(vardeclType)
+            if (newArray != null && newArray != array)
+                return listOf(IAstModification.ReplaceNode(array, newArray, parent))
+        } else {
+            val arrayDt = array.guessDatatype(program)
+            if (arrayDt.isKnown) {
+                val newArray = array.cast(arrayDt.getOr(DataType.UNDEFINED))
+                if (newArray != null && newArray != array)
+                    return listOf(IAstModification.ReplaceNode(array, newArray, parent))
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        // the args of a fuction are constfolded via recursion already.
+        val constvalue = functionCall.constValue(program)
+        return if(constvalue!=null)
+            listOf(IAstModification.ReplaceNode(functionCall, constvalue, parent))
+        else
+            noModifications
+    }
+
+    override fun after(forLoop: ForLoop, parent: Node): Iterable<IAstModification> {
+        fun adjustRangeDt(rangeFrom: NumericLiteralValue, targetDt: DataType, rangeTo: NumericLiteralValue, stepLiteral: NumericLiteralValue?, range: RangeExpr): RangeExpr? {
+            val fromCast = rangeFrom.cast(targetDt)
+            val toCast = rangeTo.cast(targetDt)
+            if(!fromCast.isValid || !toCast.isValid)
+                return null
+
+            val newStep =
+                if(stepLiteral!=null) {
+                    val stepCast = stepLiteral.cast(targetDt)
+                    if(stepCast.isValid)
+                        stepCast.valueOrZero()
+                    else
+                        range.step
+                } else {
+                    range.step
+                }
+
+            return RangeExpr(fromCast.valueOrZero(), toCast.valueOrZero(), newStep, range.position)
+        }
+
+        // adjust the datatype of a range expression in for loops to the loop variable.
+        val iterableRange = forLoop.iterable as? RangeExpr ?: return noModifications
+        val rangeFrom = iterableRange.from as? NumericLiteralValue
+        val rangeTo = iterableRange.to as? NumericLiteralValue
+        if(rangeFrom==null || rangeTo==null) return noModifications
+
+        val loopvar = forLoop.loopVar.targetVarDecl(program) ?: throw UndefinedSymbolError(forLoop.loopVar)
+
+        val stepLiteral = iterableRange.step as? NumericLiteralValue
+        when(loopvar.datatype) {
+            DataType.UBYTE -> {
+                if(rangeFrom.type!= DataType.UBYTE) {
+                    // attempt to translate the iterable into ubyte values
+                    val newIter = adjustRangeDt(rangeFrom, loopvar.datatype, rangeTo, stepLiteral, iterableRange)
+                    if(newIter!=null)
+                        return listOf(IAstModification.ReplaceNode(forLoop.iterable, newIter, forLoop))
+                }
+            }
+            DataType.BYTE -> {
+                if(rangeFrom.type!= DataType.BYTE) {
+                    // attempt to translate the iterable into byte values
+                    val newIter = adjustRangeDt(rangeFrom, loopvar.datatype, rangeTo, stepLiteral, iterableRange)
+                    if(newIter!=null)
+                        return listOf(IAstModification.ReplaceNode(forLoop.iterable, newIter, forLoop))
+                }
+            }
+            DataType.UWORD -> {
+                if(rangeFrom.type!= DataType.UWORD) {
+                    // attempt to translate the iterable into uword values
+                    val newIter = adjustRangeDt(rangeFrom, loopvar.datatype, rangeTo, stepLiteral, iterableRange)
+                    if(newIter!=null)
+                        return listOf(IAstModification.ReplaceNode(forLoop.iterable, newIter, forLoop))
+                }
+            }
+            DataType.WORD -> {
+                if(rangeFrom.type!= DataType.WORD) {
+                    // attempt to translate the iterable into word values
+                    val newIter = adjustRangeDt(rangeFrom, loopvar.datatype, rangeTo, stepLiteral, iterableRange)
+                    if(newIter!=null)
+                        return listOf(IAstModification.ReplaceNode(forLoop.iterable, newIter, forLoop))
+                }
+            }
+            else -> throw FatalAstException("invalid loopvar datatype $loopvar")
+        }
+
+        return noModifications
+    }
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        val numval = decl.value as? NumericLiteralValue
+        if(decl.type== VarDeclType.CONST && numval!=null) {
+            val valueDt = numval.inferType(program)
+            if(valueDt isnot decl.datatype) {
+                val cast = numval.cast(decl.datatype)
+                if(cast.isValid)
+                    return listOf(IAstModification.ReplaceNode(numval, cast.valueOrZero(), decl))
+            }
+        }
+        return noModifications
+    }
+
+    private class ShuffleOperands(val expr: BinaryExpression,
+                                  val exprOperator: String?,
+                                  val subExpr: BinaryExpression,
+                                  val newExprLeft: Expression?,
+                                  val newExprRight: Expression?,
+                                  val newSubexprLeft: Expression?,
+                                  val newSubexprRight: Expression?
+                                  ): IAstModification {
+        override fun perform() {
+            if(exprOperator!=null) expr.operator = exprOperator
+            if(newExprLeft!=null) expr.left = newExprLeft
+            if(newExprRight!=null) expr.right = newExprRight
+            if(newSubexprLeft!=null) subExpr.left = newSubexprLeft
+            if(newSubexprRight!=null) subExpr.right = newSubexprRight
+        }
+    }
+
+    private fun groupTwoFloatConstsTogether(expr: BinaryExpression,
+       subExpr: BinaryExpression,
+       leftIsConst: Boolean,
+       rightIsConst: Boolean,
+       subleftIsConst: Boolean,
+       subrightIsConst: Boolean): IAstModification?
+    {
+        // NOTE: THIS IS ONLY VALID ON FLOATING POINT CONSTANTS
+
+        // todo: this implements only a small set of possible reorderings at this time
+        if(expr.operator==subExpr.operator) {
+            // both operators are the same.
+
+            // If associative,  we can simply shuffle the const operands around to optimize.
+            if(expr.operator in associativeOperators) {
+                return if(leftIsConst) {
+                    if(subleftIsConst)
+                        ShuffleOperands(expr, null, subExpr, subExpr.right, null, null, expr.left)
+                    else
+                        ShuffleOperands(expr, null, subExpr, subExpr.left, null, expr.left, null)
+                } else {
+                    if(subleftIsConst)
+                        ShuffleOperands(expr, null, subExpr, null, subExpr.right, null, expr.right)
+                    else
+                        ShuffleOperands(expr, null, subExpr, null, subExpr.left, expr.right, null)
+                }
+            }
+
+            // If - or /,  we simetimes must reorder more, and flip operators (- -> +, / -> *)
+            if(expr.operator=="-" || expr.operator=="/") {
+                if(leftIsConst) {
+                    return if (subleftIsConst) {
+                        ShuffleOperands(expr, if (expr.operator == "-") "+" else "*", subExpr, subExpr.right, null, expr.left, subExpr.left)
+                    } else {
+                        IAstModification.ReplaceNode(expr,
+                                BinaryExpression(
+                                        BinaryExpression(expr.left, if (expr.operator == "-") "+" else "*", subExpr.right, subExpr.position),
+                                        expr.operator, subExpr.left, expr.position),
+                                expr.parent)
+                    }
+                } else {
+                    return if(subleftIsConst) {
+                        return ShuffleOperands(expr, null, subExpr, null, subExpr.right, null, expr.right)
+                    } else {
+                        IAstModification.ReplaceNode(expr,
+                                BinaryExpression(
+                                        subExpr.left, expr.operator,
+                                        BinaryExpression(expr.right, if (expr.operator == "-") "+" else "*", subExpr.right, subExpr.position),
+                                        expr.position),
+                                expr.parent)
+                    }
+                }
+            }
+            return null
+
+        }
+        else
+        {
+
+            if(expr.operator=="/" && subExpr.operator=="*") {
+                if(leftIsConst) {
+                    val change = if(subleftIsConst) {
+                        // C1/(C2*V) -> (C1/C2)/V
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "/", subExpr.left, subExpr.position),
+                                "/",
+                                subExpr.right, expr.position)
+                    } else {
+                        // C1/(V*C2) -> (C1/C2)/V
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "/", subExpr.right, subExpr.position),
+                                "/",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                } else {
+                    val change = if(subleftIsConst) {
+                        // (C1*V)/C2 -> (C1/C2)*V
+                        BinaryExpression(
+                                BinaryExpression(subExpr.left, "/", expr.right, subExpr.position),
+                                "*",
+                                subExpr.right, expr.position)
+                    } else {
+                        // (V*C1)/C2 -> (C1/C2)*V
+                        BinaryExpression(
+                                BinaryExpression(subExpr.right, "/", expr.right, subExpr.position),
+                                "*",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                }
+            }
+            else if(expr.operator=="*" && subExpr.operator=="/") {
+                if(leftIsConst) {
+                    val change = if(subleftIsConst) {
+                        // C1*(C2/V) -> (C1*C2)/V
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "*", subExpr.left, subExpr.position),
+                                "/",
+                                subExpr.right, expr.position)
+                    } else {
+                        // C1*(V/C2) -> (C1/C2)*V
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "/", subExpr.right, subExpr.position),
+                                "*",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                } else {
+                    val change = if(subleftIsConst) {
+                        // (C1/V)*C2 -> (C1*C2)/V
+                        BinaryExpression(
+                                BinaryExpression(subExpr.left, "*", expr.right, subExpr.position),
+                                "/",
+                                subExpr.right, expr.position)
+                    } else {
+                        // (V/C1)*C2 -> (C1/C2)*V
+                        BinaryExpression(
+                                BinaryExpression(expr.right, "/", subExpr.right, subExpr.position),
+                                "*",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                }
+            }
+            else if(expr.operator=="+" && subExpr.operator=="-") {
+                if(leftIsConst){
+                    val change = if(subleftIsConst){
+                        // c1+(c2-v)  ->  (c1+c2)-v
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "+", subExpr.left, subExpr.position),
+                                "-",
+                                subExpr.right, expr.position)
+                    } else {
+                        // c1+(v-c2)  ->  v+(c1-c2)
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "-", subExpr.right, subExpr.position),
+                                "+",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                } else {
+                    val change = if(subleftIsConst) {
+                        // (c1-v)+c2  ->  (c1+c2)-v
+                        BinaryExpression(
+                                BinaryExpression(subExpr.left, "+", expr.right, subExpr.position),
+                                "-",
+                                subExpr.right, expr.position)
+                    } else {
+                        // (v-c1)+c2  ->  v+(c2-c1)
+                        BinaryExpression(
+                                BinaryExpression(expr.right, "-", subExpr.right, subExpr.position),
+                                "+",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                }
+            }
+            else if(expr.operator=="-" && subExpr.operator=="+") {
+                if(leftIsConst) {
+                    val change = if(subleftIsConst) {
+                        // c1-(c2+v)  ->  (c1-c2)-v
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "-", subExpr.left, subExpr.position),
+                                "-",
+                                subExpr.right, expr.position)
+                    } else {
+                        // c1-(v+c2)  ->  (c1-c2)-v
+                        BinaryExpression(
+                                BinaryExpression(expr.left, "-", subExpr.right, subExpr.position),
+                                "-",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                } else {
+                    val change = if(subleftIsConst) {
+                        // (c1+v)-c2  ->  v+(c1-c2)
+                        BinaryExpression(
+                                BinaryExpression(subExpr.left, "-", expr.right, subExpr.position),
+                                "+",
+                                subExpr.right, expr.position)
+                    } else {
+                        // (v+c1)-c2  ->  v+(c1-c2)
+                        BinaryExpression(
+                                BinaryExpression(subExpr.right, "-", expr.right, subExpr.position),
+                                "+",
+                                subExpr.left, expr.position)
+                    }
+                    return IAstModification.ReplaceNode(expr, change, expr.parent)
+                }
+            }
+
+            return null
+        }
+    }
+
+}

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -0,0 +1,232 @@
+package prog8.optimizer
+
+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.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
+import prog8.compilerinterface.size
+import prog8.compilerinterface.toConstantIntegerRange
+
+// Fix up the literal value's type to match that of the vardecl
+//   (also check range literal operands types before they get expanded into arrays for instance)
+class VarConstantValueTypeAdjuster(private val program: Program, private val errors: IErrorReporter) : AstWalker() {
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+
+        if(decl.parent is AnonymousScope)
+            throw FatalAstException("vardecl may no longer occur in anonymousscope")
+
+        try {
+            val declConstValue = decl.value?.constValue(program)
+            if(declConstValue!=null && (decl.type==VarDeclType.VAR || decl.type==VarDeclType.CONST)
+                && declConstValue.inferType(program) isnot decl.datatype) {
+                // cast the numeric literal to the appropriate datatype of the variable
+                val cast = declConstValue.cast(decl.datatype)
+                if(cast.isValid)
+                    return listOf(IAstModification.ReplaceNode(decl.value!!, cast.valueOrZero(), decl))
+            }
+        } catch (x: UndefinedSymbolError) {
+            errors.err(x.message, x.position)
+        }
+
+        return noModifications
+    }
+
+    override fun after(range: RangeExpr, parent: Node): Iterable<IAstModification> {
+        val from = range.from.constValue(program)?.number?.toDouble()
+        val to = range.to.constValue(program)?.number?.toDouble()
+        val step = range.step.constValue(program)?.number?.toDouble()
+
+        if(from==null) {
+            if(!range.from.inferType(program).isInteger)
+                errors.err("range expression from value must be integer", range.from.position)
+        } else if(from-from.toInt()>0) {
+            errors.err("range expression from value must be integer", range.from.position)
+        }
+
+        if(to==null) {
+            if(!range.to.inferType(program).isInteger)
+                errors.err("range expression to value must be integer", range.to.position)
+        } else if(to-to.toInt()>0) {
+            errors.err("range expression to value must be integer", range.to.position)
+        }
+
+        if(step==null) {
+            if(!range.step.inferType(program).isInteger)
+                errors.err("range expression step value must be integer", range.step.position)
+        } else if(step-step.toInt()>0) {
+            errors.err("range expression step value must be integer", range.step.position)
+        }
+
+        return noModifications
+    }
+}
+
+
+// Replace all constant identifiers with their actual value,
+// and the array var initializer values and sizes.
+// This is needed because further constant optimizations depend on those.
+internal class ConstantIdentifierReplacer(private val program: Program, private val errors: IErrorReporter, private val compTarget: ICompilationTarget) : AstWalker() {
+
+    override fun after(identifier: IdentifierReference, parent: Node): Iterable<IAstModification> {
+        // replace identifiers that refer to const value, with the value itself
+        // if it's a simple type and if it's not a left hand side variable
+        if(identifier.parent is AssignTarget)
+            return noModifications
+        var forloop = identifier.parent as? ForLoop
+        if(forloop==null)
+            forloop = identifier.parent.parent as? ForLoop
+        if(forloop!=null && identifier===forloop.loopVar)
+            return noModifications
+
+        try {
+            val cval = identifier.constValue(program) ?: return noModifications
+            return when (cval.type) {
+                in NumericDatatypes -> listOf(
+                    IAstModification.ReplaceNode(
+                        identifier,
+                        NumericLiteralValue(cval.type, cval.number, identifier.position),
+                        identifier.parent
+                    )
+                )
+                in PassByReferenceDatatypes -> throw FatalAstException("pass-by-reference type should not be considered a constant")
+                else -> noModifications
+            }
+        } catch (x: UndefinedSymbolError) {
+            errors.err(x.message, x.position)
+            return noModifications
+        }
+    }
+
+    override fun before(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        // the initializer value can't refer to the variable itself (recursive definition)
+        // TODO: use call graph for this?
+        if(decl.value?.referencesIdentifier(decl.name) == true || decl.arraysize?.indexExpr?.referencesIdentifier(decl.name) == true) {
+            errors.err("recursive var declaration", decl.position)
+            return noModifications
+        }
+
+        if(decl.type== VarDeclType.CONST || decl.type== VarDeclType.VAR) {
+            if(decl.isArray){
+                val arraysize = decl.arraysize
+                if(arraysize==null) {
+                    // for arrays that have no size specifier attempt to deduce the size
+                    val arrayval = decl.value as? ArrayLiteralValue
+                    if(arrayval!=null) {
+                        return listOf(IAstModification.SetExpression(
+                                { decl.arraysize = ArrayIndex(it, decl.position) },
+                                NumericLiteralValue.optimalInteger(arrayval.value.size, decl.position),
+                                decl
+                        ))
+                    }
+                }
+            }
+
+            when(decl.datatype) {
+                DataType.FLOAT -> {
+                    // vardecl: for scalar float vars, promote constant integer initialization values to floats
+                    val litval = decl.value as? NumericLiteralValue
+                    if (litval!=null && litval.type in IntegerDatatypes) {
+                        val newValue = NumericLiteralValue(DataType.FLOAT, litval.number.toDouble(), litval.position)
+                        return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
+                    }
+                }
+                DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W -> {
+                    val rangeExpr = decl.value as? RangeExpr
+                    if(rangeExpr!=null) {
+                        // convert the initializer range expression to an actual array
+                        val declArraySize = decl.arraysize?.constIndex()
+                        if(declArraySize!=null && declArraySize!=rangeExpr.size())
+                            errors.err("range expression size doesn't match declared array size", decl.value?.position!!)
+                        val constRange = rangeExpr.toConstantIntegerRange()
+                        if(constRange!=null) {
+                            val eltType = rangeExpr.inferType(program).getOr(DataType.UBYTE)
+                            val newValue = if(eltType in ByteDatatypes) {
+                                ArrayLiteralValue(InferredTypes.InferredType.known(decl.datatype),
+                                        constRange.map { NumericLiteralValue(eltType, it.toShort(), decl.value!!.position) }.toTypedArray(),
+                                        position = decl.value!!.position)
+                            } else {
+                                ArrayLiteralValue(InferredTypes.InferredType.known(decl.datatype),
+                                        constRange.map { NumericLiteralValue(eltType, it, decl.value!!.position) }.toTypedArray(),
+                                        position = decl.value!!.position)
+                            }
+                            return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
+                        }
+                    }
+                    val numericLv = decl.value as? NumericLiteralValue
+                    if(numericLv!=null && numericLv.type== DataType.FLOAT)
+                        errors.err("arraysize requires only integers here", numericLv.position)
+                    val size = decl.arraysize?.constIndex() ?: return noModifications
+                    if (rangeExpr==null && numericLv!=null) {
+                        // arraysize initializer is empty or a single int, and we know the size; create the arraysize.
+                        val fillvalue = numericLv.number.toInt()
+                        when(decl.datatype){
+                            DataType.ARRAY_UB -> {
+                                if(fillvalue !in 0..255)
+                                    errors.err("ubyte value overflow", numericLv.position)
+                            }
+                            DataType.ARRAY_B -> {
+                                if(fillvalue !in -128..127)
+                                    errors.err("byte value overflow", numericLv.position)
+                            }
+                            DataType.ARRAY_UW -> {
+                                if(fillvalue !in 0..65535)
+                                    errors.err("uword value overflow", numericLv.position)
+                            }
+                            DataType.ARRAY_W -> {
+                                if(fillvalue !in -32768..32767)
+                                    errors.err("word value overflow", numericLv.position)
+                            }
+                            else -> {}
+                        }
+                        // create the array itself, filled with the fillvalue.
+                        val array = Array(size) {fillvalue}.map { NumericLiteralValue(ArrayToElementTypes.getValue(decl.datatype), it, numericLv.position) }.toTypedArray<Expression>()
+                        val refValue = ArrayLiteralValue(InferredTypes.InferredType.known(decl.datatype), array, position = numericLv.position)
+                        return listOf(IAstModification.ReplaceNode(decl.value!!, refValue, decl))
+                    }
+                }
+                DataType.ARRAY_F -> {
+                    val rangeExpr = decl.value as? RangeExpr
+                    if(rangeExpr!=null) {
+                        // convert the initializer range expression to an actual array of floats
+                        val declArraySize = decl.arraysize?.constIndex()
+                        if(declArraySize!=null && declArraySize!=rangeExpr.size())
+                            errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
+                        val constRange = rangeExpr.toConstantIntegerRange()
+                        if(constRange!=null) {
+                            val newValue = ArrayLiteralValue(InferredTypes.InferredType.known(DataType.ARRAY_F),
+                                    constRange.map { NumericLiteralValue(DataType.FLOAT, it.toDouble(), decl.value!!.position) }.toTypedArray(),
+                                    position = decl.value!!.position)
+                            return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
+                        }
+                    }
+
+                    val numericLv = decl.value as? NumericLiteralValue
+                    val size = decl.arraysize?.constIndex() ?: return noModifications
+                    if(rangeExpr==null && numericLv!=null) {
+                        // arraysize initializer is a single int, and we know the size.
+                        val fillvalue = numericLv.number.toDouble()
+                        if (fillvalue < compTarget.machine.FLOAT_MAX_NEGATIVE || fillvalue > compTarget.machine.FLOAT_MAX_POSITIVE)
+                            errors.err("float value overflow", numericLv.position)
+                        else {
+                            // create the array itself, filled with the fillvalue.
+                            val array = Array(size) {fillvalue}.map { NumericLiteralValue(DataType.FLOAT, it, numericLv.position) }.toTypedArray<Expression>()
+                            val refValue = ArrayLiteralValue(InferredTypes.InferredType.known(DataType.ARRAY_F), array, position = numericLv.position)
+                            return listOf(IAstModification.ReplaceNode(decl.value!!, refValue, decl))
+                        }
+                    }
+                }
+                else -> {
+                    // nothing to do for this type
+                }
+            }
+        }
+
+        return noModifications
+    }
+}

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -0,0 +1,721 @@
+package prog8.optimizer
+
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.base.FatalAstException
+import prog8.ast.base.IntegerDatatypes
+import prog8.ast.base.NumericDatatypes
+import prog8.ast.expressions.*
+import prog8.ast.statements.Assignment
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import kotlin.math.abs
+import kotlin.math.log2
+import kotlin.math.pow
+
+/*
+    todo add more peephole expression optimizations
+
+    Investigate what optimizations binaryen has, also see  https://egorbo.com/peephole-optimizations.html
+
+*(&X)  =>  X
+X % 1  => 0
+X / 1  =>  X
+X ^ -1  =>  ~x
+X >= 1  =>  X > 0
+X <  1  =>  X <= 0
+X + С1 == C2  =>  X == C2 - C1
+((X + C1) + C2)  =>  (X + (C1 + C2))
+((X + C1) + (Y + C2))  =>  ((X + Y) + (C1 + C2))
+
+
+ */
+
+
+class ExpressionSimplifier(private val program: Program) : AstWalker() {
+    private val powersOfTwo = (1..16).map { (2.0).pow(it) }.toSet()
+    private val negativePowersOfTwo = powersOfTwo.map { -it }.toSet()
+
+    override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
+        val mods = mutableListOf<IAstModification>()
+
+        // try to statically convert a literal value into one of the desired type
+        val literal = typecast.expression as? NumericLiteralValue
+        if (literal != null) {
+            val newLiteral = literal.cast(typecast.type)
+            if (newLiteral.isValid && newLiteral.valueOrZero() !== literal)
+                mods += IAstModification.ReplaceNode(typecast.expression, newLiteral.valueOrZero(), typecast)
+        }
+
+        // remove redundant nested typecasts
+        val subTypecast = typecast.expression as? TypecastExpression
+        if (subTypecast != null) {
+            // remove the sub-typecast if its datatype is larger than the outer typecast
+            if(subTypecast.type largerThan typecast.type) {
+                mods += IAstModification.ReplaceNode(typecast.expression, subTypecast.expression, typecast)
+            }
+        } else {
+            if (typecast.expression.inferType(program) istype typecast.type) {
+                // remove duplicate cast
+                mods += IAstModification.ReplaceNode(typecast, typecast.expression, parent)
+            }
+        }
+
+        return mods
+    }
+
+    override fun before(expr: PrefixExpression, parent: Node): Iterable<IAstModification> {
+        if (expr.operator == "+") {
+            // +X --> X
+            return listOf(IAstModification.ReplaceNode(expr, expr.expression, parent))
+        } else if (expr.operator == "not") {
+            when(expr.expression) {
+                is PrefixExpression -> {
+                    // NOT(NOT(...)) -> ...
+                    val pe = expr.expression as PrefixExpression
+                    if(pe.operator == "not")
+                        return listOf(IAstModification.ReplaceNode(expr, pe.expression, parent))
+                }
+                is BinaryExpression -> {
+                    // NOT (xxxx)  ->   invert the xxxx
+                    val be = expr.expression as BinaryExpression
+                    val newExpr = when (be.operator) {
+                        "<" -> BinaryExpression(be.left, ">=", be.right, be.position)
+                        ">" -> BinaryExpression(be.left, "<=", be.right, be.position)
+                        "<=" -> BinaryExpression(be.left, ">", be.right, be.position)
+                        ">=" -> BinaryExpression(be.left, "<", be.right, be.position)
+                        "==" -> BinaryExpression(be.left, "!=", be.right, be.position)
+                        "!=" -> BinaryExpression(be.left, "==", be.right, be.position)
+                        else -> null
+                    }
+
+                    if (newExpr != null)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                }
+                else -> return noModifications
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        val leftVal = expr.left.constValue(program)
+        val rightVal = expr.right.constValue(program)
+
+        val leftIDt = expr.left.inferType(program)
+        val rightIDt = expr.right.inferType(program)
+        if (!leftIDt.isKnown || !rightIDt.isKnown)
+            throw FatalAstException("can't determine datatype of both expression operands $expr")
+
+        // ConstValue <associativeoperator> X -->  X <associativeoperator> ConstValue
+        if (leftVal != null && expr.operator in associativeOperators && rightVal == null)
+            return listOf(IAstModification.SwapOperands(expr))
+
+        // NonBinaryExpression  <associativeoperator>  BinaryExpression  -->  BinaryExpression  <associativeoperator>  NonBinaryExpression
+        if (expr.operator in associativeOperators && expr.left !is BinaryExpression && expr.right is BinaryExpression) {
+            if(parent !is Assignment || !(expr.left isSameAs parent.target))
+                return listOf(IAstModification.SwapOperands(expr))
+        }
+
+        // X + (-A)  -->  X - A
+        if (expr.operator == "+" && (expr.right as? PrefixExpression)?.operator == "-") {
+            return listOf(IAstModification.ReplaceNode(
+                    expr,
+                    BinaryExpression(expr.left, "-", (expr.right as PrefixExpression).expression, expr.position),
+                    parent
+            ))
+        }
+
+        // (-A) + X  -->  X - A
+        if (expr.operator == "+" && (expr.left as? PrefixExpression)?.operator == "-") {
+            return listOf(IAstModification.ReplaceNode(
+                    expr,
+                    BinaryExpression(expr.right, "-", (expr.left as PrefixExpression).expression, expr.position),
+                    parent
+            ))
+        }
+
+        // X - (-A)  -->  X + A
+        if (expr.operator == "-" && (expr.right as? PrefixExpression)?.operator == "-") {
+            return listOf(IAstModification.ReplaceNode(
+                    expr,
+                    BinaryExpression(expr.left, "+", (expr.right as PrefixExpression).expression, expr.position),
+                    parent
+            ))
+        }
+
+        val leftDt = leftIDt.getOr(DataType.UNDEFINED)
+        val rightDt = rightIDt.getOr(DataType.UNDEFINED)
+
+        if (expr.operator == "+" || expr.operator == "-"
+                && leftVal == null && rightVal == null
+                && leftDt in NumericDatatypes && rightDt in NumericDatatypes) {
+            val leftBinExpr = expr.left as? BinaryExpression
+            val rightBinExpr = expr.right as? BinaryExpression
+            if (leftBinExpr?.operator == "*") {
+                if (expr.operator == "+") {
+                    // Y*X + X  ->  X*(Y + 1)
+                    // X*Y + X  ->  X*(Y + 1)
+                    val x = expr.right
+                    val y = determineY(x, leftBinExpr)
+                    if (y != null) {
+                        val yPlus1 = BinaryExpression(y, "+", NumericLiteralValue(leftDt, 1, y.position), y.position)
+                        val newExpr = BinaryExpression(x, "*", yPlus1, x.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    }
+                } else {
+                    // Y*X - X  ->  X*(Y - 1)
+                    // X*Y - X  ->  X*(Y - 1)
+                    val x = expr.right
+                    val y = determineY(x, leftBinExpr)
+                    if (y != null) {
+                        val yMinus1 = BinaryExpression(y, "-", NumericLiteralValue(leftDt, 1, y.position), y.position)
+                        val newExpr = BinaryExpression(x, "*", yMinus1, x.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    }
+                }
+            } else if (rightBinExpr?.operator == "*") {
+                if (expr.operator == "+") {
+                    // X + Y*X  ->  X*(Y + 1)
+                    // X + X*Y  ->  X*(Y + 1)
+                    val x = expr.left
+                    val y = determineY(x, rightBinExpr)
+                    if (y != null) {
+                        val yPlus1 = BinaryExpression(y, "+", NumericLiteralValue.optimalInteger(1, y.position), y.position)
+                        val newExpr = BinaryExpression(x, "*", yPlus1, x.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    }
+                }
+            }
+        }
+
+        if(expr.operator == ">=" && rightVal?.number == 0) {
+            if (leftDt == DataType.UBYTE || leftDt == DataType.UWORD) {
+                // unsigned >= 0 --> true
+                return listOf(IAstModification.ReplaceNode(expr, NumericLiteralValue.fromBoolean(true, expr.position), parent))
+            }
+        }
+
+        if(expr.operator == "<" && rightVal?.number == 0) {
+            if (leftDt == DataType.UBYTE || leftDt == DataType.UWORD) {
+                // unsigned < 0 --> false
+                return listOf(IAstModification.ReplaceNode(expr, NumericLiteralValue.fromBoolean(false, expr.position), parent))
+            }
+            when(leftDt) {
+                DataType.BYTE -> {
+                    // signed < 0  -->  signed & $80
+                    return listOf(IAstModification.ReplaceNode(
+                            expr,
+                            BinaryExpression(expr.left, "&", NumericLiteralValue.optimalInteger(0x80, expr.position), expr.position),
+                            parent
+                    ))
+                }
+                DataType.WORD -> {
+                    // signedw < 0 -->  msb(signedw) & $80
+                    return listOf(IAstModification.ReplaceNode(
+                            expr,
+                            BinaryExpression(FunctionCall(IdentifierReference(listOf("msb"), expr.position),
+                                    mutableListOf(expr.left),
+                                    expr.position
+                            ), "&", NumericLiteralValue.optimalInteger(0x80, expr.position), expr.position),
+                            parent
+                    ))
+                }
+                else -> {}
+            }
+        }
+
+        // simplify when a term is constant and directly determines the outcome
+        val constTrue = NumericLiteralValue.fromBoolean(true, expr.position)
+        val constFalse = NumericLiteralValue.fromBoolean(false, expr.position)
+        val newExpr: Expression? = when (expr.operator) {
+            "or" -> {
+                if ((leftVal != null && leftVal.asBooleanValue) || (rightVal != null && rightVal.asBooleanValue))
+                    constTrue
+                else if (leftVal != null && !leftVal.asBooleanValue)
+                    expr.right
+                else if (rightVal != null && !rightVal.asBooleanValue)
+                    expr.left
+                else
+                    null
+            }
+            "and" -> {
+                if ((leftVal != null && !leftVal.asBooleanValue) || (rightVal != null && !rightVal.asBooleanValue))
+                    constFalse
+                else if (leftVal != null && leftVal.asBooleanValue)
+                    expr.right
+                else if (rightVal != null && rightVal.asBooleanValue)
+                    expr.left
+                else
+                    null
+            }
+            "xor" -> {
+                if (leftVal != null && !leftVal.asBooleanValue)
+                    expr.right
+                else if (rightVal != null && !rightVal.asBooleanValue)
+                    expr.left
+                else if (leftVal != null && leftVal.asBooleanValue)
+                    PrefixExpression("not", expr.right, expr.right.position)
+                else if (rightVal != null && rightVal.asBooleanValue)
+                    PrefixExpression("not", expr.left, expr.left.position)
+                else
+                    null
+            }
+            "|", "^" -> {
+                if (leftVal != null && !leftVal.asBooleanValue)
+                    expr.right
+                else if (rightVal != null && !rightVal.asBooleanValue)
+                    expr.left
+                else
+                    null
+            }
+            "&" -> {
+                if (leftVal != null && !leftVal.asBooleanValue)
+                    constFalse
+                else if (rightVal != null && !rightVal.asBooleanValue)
+                    constFalse
+                else
+                    null
+            }
+            "*" -> optimizeMultiplication(expr, leftVal, rightVal)
+            "/" -> optimizeDivision(expr, leftVal, rightVal)
+            "+" -> optimizeAdd(expr, leftVal, rightVal)
+            "-" -> optimizeSub(expr, leftVal, rightVal)
+            "**" -> optimizePower(expr, leftVal, rightVal)
+            "%" -> optimizeRemainder(expr, leftVal, rightVal)
+            ">>" -> optimizeShiftRight(expr, rightVal)
+            "<<" -> optimizeShiftLeft(expr, rightVal)
+            else -> null
+        }
+
+        if(newExpr != null)
+            return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+
+        return noModifications
+    }
+
+    override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        if(functionCall.target.nameInSource == listOf("lsb")) {
+            val arg = functionCall.args[0]
+            if(arg is TypecastExpression) {
+                val valueDt = arg.expression.inferType(program)
+                if (valueDt istype DataType.BYTE || valueDt istype DataType.UBYTE) {
+                    // useless lsb() of byte value that was typecasted to word
+                    return listOf(IAstModification.ReplaceNode(functionCall, arg.expression, parent))
+                }
+            } else {
+                val argDt = arg.inferType(program)
+                if (argDt istype DataType.BYTE || argDt istype DataType.UBYTE) {
+                    // useless lsb() of byte value
+                    return listOf(IAstModification.ReplaceNode(functionCall, arg, parent))
+                }
+            }
+        }
+        else if(functionCall.target.nameInSource == listOf("msb")) {
+            val arg = functionCall.args[0]
+            if(arg is TypecastExpression) {
+                val valueDt = arg.expression.inferType(program)
+                if (valueDt istype DataType.BYTE || valueDt istype DataType.UBYTE) {
+                    // useless msb() of byte value that was typecasted to word, replace with 0
+                    return listOf(IAstModification.ReplaceNode(
+                            functionCall,
+                            NumericLiteralValue(valueDt.getOr(DataType.UBYTE), 0, arg.expression.position),
+                            parent))
+                }
+            } else {
+                val argDt = arg.inferType(program)
+                if (argDt istype DataType.BYTE || argDt istype DataType.UBYTE) {
+                    // useless msb() of byte value, replace with 0
+                    return listOf(IAstModification.ReplaceNode(
+                            functionCall,
+                            NumericLiteralValue(argDt.getOr(DataType.UBYTE), 0, arg.position),
+                            parent))
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun determineY(x: Expression, subBinExpr: BinaryExpression): Expression? {
+        return when {
+            subBinExpr.left isSameAs x -> subBinExpr.right
+            subBinExpr.right isSameAs x -> subBinExpr.left
+            else -> null
+        }
+    }
+
+    private fun optimizeAdd(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if(expr.left.isSameAs(expr.right)) {
+            // optimize X+X into X *2
+            expr.operator = "*"
+            expr.right = NumericLiteralValue.optimalInteger(2, expr.right.position)
+            expr.right.linkParents(expr)
+            return expr
+        }
+
+        if (leftVal == null && rightVal == null)
+            return null
+
+        val (expr2, _, rightVal2) = reorderAssociativeWithConstant(expr, leftVal)
+        if (rightVal2 != null) {
+            // right value is a constant, see if we can optimize
+            val rightConst: NumericLiteralValue = rightVal2
+            when (rightConst.number.toDouble()) {
+                0.0 -> {
+                    // left
+                    return expr2.left
+                }
+            }
+        }
+        // no need to check for left val constant (because of associativity)
+
+        val rnum = rightVal?.number?.toDouble()
+        if(rnum!=null && rnum<0.0) {
+            expr.operator = "-"
+            expr.right = NumericLiteralValue(rightVal.type, -rnum, rightVal.position)
+            return expr
+        }
+
+        return null
+    }
+
+    private fun optimizeSub(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if(expr.left.isSameAs(expr.right)) {
+            // optimize X-X into 0
+            return NumericLiteralValue.optimalInteger(0, expr.position)
+        }
+
+        if (leftVal == null && rightVal == null)
+            return null
+
+        if (rightVal != null) {
+            // right value is a constant, see if we can optimize
+            val rnum = rightVal.number.toDouble()
+            if (rnum == 0.0) {
+                // left
+                return expr.left
+            }
+
+            if(rnum<0.0) {
+                expr.operator = "+"
+                expr.right = NumericLiteralValue(rightVal.type, -rnum, rightVal.position)
+                return expr
+            }
+        }
+        if (leftVal != null) {
+            // left value is a constant, see if we can optimize
+            when (leftVal.number.toDouble()) {
+                0.0 -> {
+                    // -right
+                    return PrefixExpression("-", expr.right, expr.position)
+                }
+            }
+        }
+
+
+        return null
+    }
+
+    private fun optimizePower(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if (leftVal == null && rightVal == null)
+            return null
+
+        if (rightVal != null) {
+            // right value is a constant, see if we can optimize
+            val rightConst: NumericLiteralValue = rightVal
+            when (rightConst.number.toDouble()) {
+                -3.0 -> {
+                    // -1/(left*left*left)
+                    return BinaryExpression(NumericLiteralValue(DataType.FLOAT, -1.0, expr.position), "/",
+                            BinaryExpression(expr.left, "*", BinaryExpression(expr.left, "*", expr.left, expr.position), expr.position),
+                            expr.position)
+                }
+                -2.0 -> {
+                    // -1/(left*left)
+                    return BinaryExpression(NumericLiteralValue(DataType.FLOAT, -1.0, expr.position), "/",
+                            BinaryExpression(expr.left, "*", expr.left, expr.position),
+                            expr.position)
+                }
+                -1.0 -> {
+                    // -1/left
+                    return BinaryExpression(NumericLiteralValue(DataType.FLOAT, -1.0, expr.position), "/",
+                            expr.left, expr.position)
+                }
+                0.0 -> {
+                    // 1
+                    return NumericLiteralValue(rightConst.type, 1, expr.position)
+                }
+                0.5 -> {
+                    // sqrt(left)
+                    return FunctionCall(IdentifierReference(listOf("sqrt"), expr.position), mutableListOf(expr.left), expr.position)
+                }
+                1.0 -> {
+                    // left
+                    return expr.left
+                }
+                2.0 -> {
+                    // left*left
+                    return BinaryExpression(expr.left, "*", expr.left, expr.position)
+                }
+                3.0 -> {
+                    // left*left*left
+                    return BinaryExpression(expr.left, "*", BinaryExpression(expr.left, "*", expr.left, expr.position), expr.position)
+                }
+            }
+        }
+        if (leftVal != null) {
+            // left value is a constant, see if we can optimize
+            when (leftVal.number.toDouble()) {
+                -1.0 -> {
+                    // -1
+                    return NumericLiteralValue(DataType.FLOAT, -1.0, expr.position)
+                }
+                0.0 -> {
+                    // 0
+                    return NumericLiteralValue(leftVal.type, 0, expr.position)
+                }
+                1.0 -> {
+                    //1
+                    return NumericLiteralValue(leftVal.type, 1, expr.position)
+                }
+
+            }
+        }
+
+        return null
+    }
+
+    private fun optimizeRemainder(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if (leftVal == null && rightVal == null)
+            return null
+
+        // simplify assignments  A = B <operator> C
+
+        val cv = rightVal?.number?.toInt()?.toDouble()
+        when (expr.operator) {
+            "%" -> {
+                if (cv == 1.0) {
+                    val idt = expr.inferType(program)
+                    if(!idt.isKnown)
+                        throw FatalAstException("unknown dt")
+                    return NumericLiteralValue(idt.getOr(DataType.UNDEFINED), 0, expr.position)
+                } else if (cv in powersOfTwo) {
+                    expr.operator = "&"
+                    expr.right = NumericLiteralValue.optimalInteger(cv!!.toInt()-1, expr.position)
+                    return null
+                }
+            }
+        }
+        return null
+
+    }
+
+    private fun optimizeDivision(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if (leftVal == null && rightVal == null)
+            return null
+
+        // cannot shuffle assiciativity with division!
+        if (rightVal != null) {
+            // right value is a constant, see if we can optimize
+            val rightConst: NumericLiteralValue = rightVal
+            val cv = rightConst.number.toDouble()
+            val leftIDt = expr.left.inferType(program)
+            if (!leftIDt.isKnown)
+                return null
+            val leftDt = leftIDt.getOr(DataType.UNDEFINED)
+            when (cv) {
+                -1.0 -> {
+                    //  '/' -> -left
+                    if (expr.operator == "/") {
+                        return PrefixExpression("-", expr.left, expr.position)
+                    }
+                }
+                1.0 -> {
+                    //  '/' -> left
+                    if (expr.operator == "/") {
+                        return expr.left
+                    }
+                }
+                in powersOfTwo -> {
+                    if (leftDt in IntegerDatatypes) {
+                        // divided by a power of two => shift right
+                        val numshifts = log2(cv).toInt()
+                        return BinaryExpression(expr.left, ">>", NumericLiteralValue.optimalInteger(numshifts, expr.position), expr.position)
+                    }
+                }
+                in negativePowersOfTwo -> {
+                    if (leftDt in IntegerDatatypes) {
+                        // divided by a negative power of two => negate, then shift right
+                        val numshifts = log2(-cv).toInt()
+                        return BinaryExpression(PrefixExpression("-", expr.left, expr.position), ">>", NumericLiteralValue.optimalInteger(numshifts, expr.position), expr.position)
+                    }
+                }
+            }
+
+            if (leftDt == DataType.UBYTE) {
+                if (abs(rightConst.number.toDouble()) >= 256.0) {
+                    return NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                }
+            } else if (leftDt == DataType.UWORD) {
+                if (abs(rightConst.number.toDouble()) >= 65536.0) {
+                    return NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                }
+            }
+        }
+
+        if (leftVal != null) {
+            // left value is a constant, see if we can optimize
+            when (leftVal.number.toDouble()) {
+                0.0 -> {
+                    // 0
+                    return NumericLiteralValue(leftVal.type, 0, expr.position)
+                }
+            }
+        }
+
+        return null
+    }
+
+    private fun optimizeMultiplication(expr: BinaryExpression, leftVal: NumericLiteralValue?, rightVal: NumericLiteralValue?): Expression? {
+        if (leftVal == null && rightVal == null)
+            return null
+
+        val (expr2, _, rightVal2) = reorderAssociativeWithConstant(expr, leftVal)
+        if (rightVal2 != null) {
+            // right value is a constant, see if we can optimize
+            val leftValue: Expression = expr2.left
+            val rightConst: NumericLiteralValue = rightVal2
+            when (val cv = rightConst.number.toDouble()) {
+                -1.0 -> {
+                    // -left
+                    return PrefixExpression("-", leftValue, expr.position)
+                }
+                0.0 -> {
+                    // 0
+                    return NumericLiteralValue(rightConst.type, 0, expr.position)
+                }
+                1.0 -> {
+                    // left
+                    return expr2.left
+                }
+                in powersOfTwo -> {
+                    if (leftValue.inferType(program).isInteger) {
+                        // times a power of two => shift left
+                        val numshifts = log2(cv).toInt()
+                        return BinaryExpression(expr2.left, "<<", NumericLiteralValue.optimalInteger(numshifts, expr.position), expr.position)
+                    }
+                }
+                in negativePowersOfTwo -> {
+                    if (leftValue.inferType(program).isInteger) {
+                        // times a negative power of two => negate, then shift left
+                        val numshifts = log2(-cv).toInt()
+                        return BinaryExpression(PrefixExpression("-", expr2.left, expr.position), "<<", NumericLiteralValue.optimalInteger(numshifts, expr.position), expr.position)
+                    }
+                }
+            }
+        }
+        // no need to check for left val constant (because of associativity)
+
+        return null
+    }
+
+    private fun optimizeShiftLeft(expr: BinaryExpression, amountLv: NumericLiteralValue?): Expression? {
+        if (amountLv == null)
+            return null
+
+        val amount = amountLv.number.toInt()
+        if (amount == 0) {
+            return expr.left
+        }
+        val targetIDt = expr.left.inferType(program)
+        if(!targetIDt.isKnown)
+            throw FatalAstException("unknown dt")
+        when (val targetDt = targetIDt.getOr(DataType.UNDEFINED)) {
+            DataType.UBYTE, DataType.BYTE -> {
+                if (amount >= 8) {
+                    return NumericLiteralValue(targetDt, 0, expr.position)
+                }
+            }
+            DataType.UWORD, DataType.WORD -> {
+                if (amount >= 16) {
+                    return NumericLiteralValue(targetDt, 0, expr.position)
+                } else if (amount >= 8) {
+                    val lsb = FunctionCall(IdentifierReference(listOf("lsb"), expr.position), mutableListOf(expr.left), expr.position)
+                    if (amount == 8) {
+                        return FunctionCall(IdentifierReference(listOf("mkword"), expr.position), mutableListOf(lsb, NumericLiteralValue.optimalInteger(0, expr.position)), expr.position)
+                    }
+                    val shifted = BinaryExpression(lsb, "<<", NumericLiteralValue.optimalInteger(amount - 8, expr.position), expr.position)
+                    return FunctionCall(IdentifierReference(listOf("mkword"), expr.position), mutableListOf(shifted, NumericLiteralValue.optimalInteger(0, expr.position)), expr.position)
+                }
+            }
+            else -> {
+            }
+        }
+        return null
+    }
+
+    private fun optimizeShiftRight(expr: BinaryExpression, amountLv: NumericLiteralValue?): Expression? {
+        if (amountLv == null)
+            return null
+
+        val amount = amountLv.number.toInt()
+        if (amount == 0) {
+            return expr.left
+        }
+        val idt = expr.left.inferType(program)
+        if(!idt.isKnown)
+            throw FatalAstException("unknown dt")
+        when (idt.getOr(DataType.UNDEFINED)) {
+            DataType.UBYTE -> {
+                if (amount >= 8) {
+                    return NumericLiteralValue.optimalInteger(0, expr.position)
+                }
+            }
+            DataType.BYTE -> {
+                if (amount > 8) {
+                    expr.right = NumericLiteralValue.optimalInteger(8, expr.right.position)
+                    return null
+                }
+            }
+            DataType.UWORD -> {
+                if (amount >= 16) {
+                    return NumericLiteralValue.optimalInteger(0, expr.position)
+                }
+                else if (amount >= 8) {
+                    val msb = FunctionCall(IdentifierReference(listOf("msb"), expr.position), mutableListOf(expr.left), expr.position)
+                    if (amount == 8) {
+                        // mkword(0, msb(v))
+                        val zero = NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                        return FunctionCall(IdentifierReference(listOf("mkword"), expr.position), mutableListOf(zero, msb), expr.position)
+                    }
+                    return TypecastExpression(BinaryExpression(msb, ">>", NumericLiteralValue.optimalInteger(amount - 8, expr.position), expr.position), DataType.UWORD, true, expr.position)
+                }
+            }
+            DataType.WORD -> {
+                if (amount > 16) {
+                    expr.right = NumericLiteralValue.optimalInteger(16, expr.right.position)
+                    return null
+                }
+            }
+            else -> {
+            }
+        }
+        return null
+    }
+
+    private fun reorderAssociativeWithConstant(expr: BinaryExpression, leftVal: NumericLiteralValue?): BinExprWithConstants {
+        if (expr.operator in associativeOperators && leftVal != null) {
+            // swap left and right so that right is always the constant
+            val tmp = expr.left
+            expr.left = expr.right
+            expr.right = tmp
+            return BinExprWithConstants(expr, expr.right.constValue(program), leftVal)
+        }
+        return BinExprWithConstants(expr, leftVal, expr.right.constValue(program))
+    }
+
+    private data class BinExprWithConstants(val expr: BinaryExpression, val leftVal: NumericLiteralValue?, val rightVal: NumericLiteralValue?)
+
+}

codeOptimizers/src/prog8/optimizer/Extensions.kt

@@ -0,0 +1,67 @@
+package prog8.optimizer
+
+import prog8.ast.IBuiltinFunctions
+import prog8.ast.Program
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
+
+
+fun Program.constantFold(errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val valuetypefixer = VarConstantValueTypeAdjuster(this, errors)
+    valuetypefixer.visit(this)
+    if(errors.noErrors()) {
+        valuetypefixer.applyModifications()
+
+        val replacer = ConstantIdentifierReplacer(this, errors, compTarget)
+        replacer.visit(this)
+        if (errors.noErrors()) {
+            replacer.applyModifications()
+
+            valuetypefixer.visit(this)
+            if(errors.noErrors()) {
+                valuetypefixer.applyModifications()
+
+                val optimizer = ConstantFoldingOptimizer(this)
+                optimizer.visit(this)
+                while (errors.noErrors() && optimizer.applyModifications() > 0) {
+                    optimizer.visit(this)
+                }
+
+                if (errors.noErrors()) {
+                    replacer.visit(this)
+                    replacer.applyModifications()
+                }
+            }
+        }
+    }
+
+    if(errors.noErrors())
+        modules.forEach { it.linkParents(namespace) }   // re-link in final configuration
+}
+
+
+fun Program.optimizeStatements(errors: IErrorReporter,
+                               functions: IBuiltinFunctions,
+                               compTarget: ICompilationTarget
+): Int {
+    val optimizer = StatementOptimizer(this, errors, functions, compTarget)
+    optimizer.visit(this)
+    val optimizationCount = optimizer.applyModifications()
+
+    modules.forEach { it.linkParents(this.namespace) }   // re-link in final configuration
+
+    return optimizationCount
+}
+
+fun Program.simplifyExpressions() : Int {
+    val opti = ExpressionSimplifier(this)
+    opti.visit(this)
+    return opti.applyModifications()
+}
+
+fun Program.splitBinaryExpressions(options: CompilationOptions, compTarget: ICompilationTarget) : Int {
+    val opti = BinExprSplitter(this, options, compTarget)
+    opti.visit(this)
+    return opti.applyModifications()
+}

codeOptimizers/src/prog8/optimizer/StatementOptimizer.kt

@@ -0,0 +1,493 @@
+package prog8.optimizer
+
+import prog8.ast.IBuiltinFunctions
+import prog8.ast.IStatementContainer
+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.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
+import prog8.compilerinterface.size
+import kotlin.math.floor
+
+internal const val retvarName = "prog8_retval"
+
+
+class StatementOptimizer(private val program: Program,
+                                  private val errors: IErrorReporter,
+                                  private val functions: IBuiltinFunctions,
+                                  private val compTarget: ICompilationTarget
+) : AstWalker() {
+
+    private val subsThatNeedReturnVariable = mutableSetOf<Triple<IStatementContainer, DataType, Position>>()
+
+
+    override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        for(returnvar in subsThatNeedReturnVariable) {
+            val decl = VarDecl(VarDeclType.VAR, returnvar.second, ZeropageWish.DONTCARE, null, retvarName, null,
+                isArray = false,
+                autogeneratedDontRemove = true,
+                sharedWithAsm = false,
+                position = returnvar.third
+            )
+            returnvar.first.statements.add(0, decl)
+        }
+        subsThatNeedReturnVariable.clear()
+        return noModifications
+    }
+
+    override fun before(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        // if the first instruction in the called subroutine is a return statement with a simple value,
+        // remove the jump altogeter and inline the returnvalue directly.
+        val subroutine = functionCall.target.targetSubroutine(program)
+        if(subroutine!=null) {
+            val first = subroutine.statements.asSequence().filterNot { it is VarDecl || it is Directive }.firstOrNull()
+            if(first is Return && first.value?.isSimple==true) {
+                val copy = when(val orig = first.value!!) {
+                    is AddressOf -> {
+                        val scoped = scopePrefix(orig.identifier, subroutine)
+                        AddressOf(scoped, orig.position)
+                    }
+                    is DirectMemoryRead -> {
+                        when(val expr = orig.addressExpression) {
+                            is NumericLiteralValue -> DirectMemoryRead(expr.copy(), orig.position)
+                            else -> return noModifications
+                        }
+                    }
+                    is IdentifierReference -> scopePrefix(orig, subroutine)
+                    is NumericLiteralValue -> orig.copy()
+                    is StringLiteralValue -> orig.copy()
+                    else -> return noModifications
+                }
+                return listOf(IAstModification.ReplaceNode(functionCall, copy, parent))
+            }
+        }
+        return noModifications
+    }
+
+    private fun scopePrefix(variable: IdentifierReference, subroutine: Subroutine): IdentifierReference {
+        val scoped = subroutine.makeScopedName(variable.nameInSource.last())
+        return IdentifierReference(scoped.split('.'), variable.position)
+    }
+
+    override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        if(functionCallStatement.target.nameInSource.size==1 && functionCallStatement.target.nameInSource[0] in functions.names) {
+            val functionName = functionCallStatement.target.nameInSource[0]
+            if (functionName in functions.purefunctionNames) {
+                errors.warn("statement has no effect (function return value is discarded)", functionCallStatement.position)
+                return listOf(IAstModification.Remove(functionCallStatement, parent as IStatementContainer))
+            }
+        }
+
+        // printing a literal string of just 2 or 1 characters is replaced by directly outputting those characters
+        // only do this optimization if the arg is a known-constant string literal instead of a user defined variable.
+        if(functionCallStatement.target.nameInSource==listOf("txt", "print")) {
+            val arg = functionCallStatement.args.single()
+            val stringVar: IdentifierReference? = if(arg is AddressOf) {
+                arg.identifier
+            } else {
+                arg as? IdentifierReference
+            }
+            if(stringVar!=null && stringVar.wasStringLiteral(program)) {
+                val string = stringVar.targetVarDecl(program)?.value as? StringLiteralValue
+                if(string!=null) {
+                    val pos = functionCallStatement.position
+                    if (string.value.length == 1) {
+                        val firstCharEncoded = compTarget.encodeString(string.value, string.altEncoding)[0]
+                        val chrout = FunctionCallStatement(
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstCharEncoded.toInt(), pos)),
+                            functionCallStatement.void, pos
+                        )
+                        return listOf(IAstModification.ReplaceNode(functionCallStatement, chrout, parent))
+                    } else if (string.value.length == 2) {
+                        val firstTwoCharsEncoded = compTarget.encodeString(string.value.take(2), string.altEncoding)
+                        val chrout1 = FunctionCallStatement(
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[0].toInt(), pos)),
+                            functionCallStatement.void, pos
+                        )
+                        val chrout2 = FunctionCallStatement(
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[1].toInt(), pos)),
+                            functionCallStatement.void, pos
+                        )
+                        return listOf(
+                            IAstModification.InsertBefore(functionCallStatement, chrout1, parent as IStatementContainer),
+                            IAstModification.ReplaceNode(functionCallStatement, chrout2, parent)
+                        )
+                    }
+                }
+            }
+        }
+
+        // if the first instruction in the called subroutine is a return statement, remove the jump altogeter
+        val subroutine = functionCallStatement.target.targetSubroutine(program)
+        if(subroutine!=null) {
+            val first = subroutine.statements.asSequence().filterNot { it is VarDecl || it is Directive }.firstOrNull()
+            if(first is Return)
+                return listOf(IAstModification.Remove(functionCallStatement, parent as IStatementContainer))
+        }
+
+        return noModifications
+    }
+
+//    override fun before(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+//        // if the first instruction in the called subroutine is a return statement with constant value, replace with the constant value
+//        val subroutine = functionCall.target.targetSubroutine(program)
+//        if(subroutine!=null) {
+//            val first = subroutine.statements.asSequence().filterNot { it is VarDecl || it is Directive }.firstOrNull()
+//            if(first is Return && first.value!=null) {
+//                val constval = first.value?.constValue(program)
+//                if(constval!=null)
+//                    return listOf(IAstModification.ReplaceNode(functionCall, constval, parent))
+//            }
+//        }
+//        return noModifications
+//    }
+
+    override fun after(ifStatement: IfStatement, parent: Node): Iterable<IAstModification> {
+        // remove empty if statements
+        if(ifStatement.truepart.isEmpty() && ifStatement.elsepart.isEmpty())
+            return listOf(IAstModification.Remove(ifStatement, parent as IStatementContainer))
+
+        // empty true part? switch with the else part
+        if(ifStatement.truepart.isEmpty() && ifStatement.elsepart.isNotEmpty()) {
+            val invertedCondition = PrefixExpression("not", ifStatement.condition, ifStatement.condition.position)
+            val emptyscope = AnonymousScope(mutableListOf(), ifStatement.elsepart.position)
+            val truepart = AnonymousScope(ifStatement.elsepart.statements, ifStatement.truepart.position)
+            return listOf(
+                    IAstModification.ReplaceNode(ifStatement.condition, invertedCondition, ifStatement),
+                    IAstModification.ReplaceNode(ifStatement.truepart, truepart, ifStatement),
+                    IAstModification.ReplaceNode(ifStatement.elsepart, emptyscope, ifStatement)
+            )
+        }
+
+        val constvalue = ifStatement.condition.constValue(program)
+        if(constvalue!=null) {
+            return if(constvalue.asBooleanValue){
+                // always true -> keep only if-part
+                errors.warn("condition is always true", ifStatement.position)
+                listOf(IAstModification.ReplaceNode(ifStatement, ifStatement.truepart, parent))
+            } else {
+                // always false -> keep only else-part
+                errors.warn("condition is always false", ifStatement.position)
+                listOf(IAstModification.ReplaceNode(ifStatement, ifStatement.elsepart, parent))
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(forLoop: ForLoop, parent: Node): Iterable<IAstModification> {
+        if(forLoop.body.isEmpty()) {
+            errors.warn("removing empty for loop", forLoop.position)
+            return listOf(IAstModification.Remove(forLoop, parent as IStatementContainer))
+        } else if(forLoop.body.statements.size==1) {
+            val loopvar = forLoop.body.statements[0] as? VarDecl
+            if(loopvar!=null && loopvar.name==forLoop.loopVar.nameInSource.singleOrNull()) {
+                // remove empty for loop (only loopvar decl in it)
+                return listOf(IAstModification.Remove(forLoop, parent as IStatementContainer))
+            }
+        }
+
+        val range = forLoop.iterable as? RangeExpr
+        if(range!=null) {
+            if (range.size() == 1) {
+                // for loop over a (constant) range of just a single value-- optimize the loop away
+                // loopvar/reg = range value , follow by block
+                val scope = AnonymousScope(mutableListOf(), forLoop.position)
+                scope.statements.add(Assignment(AssignTarget(forLoop.loopVar, null, null, forLoop.position), range.from, forLoop.position))
+                scope.statements.addAll(forLoop.body.statements)
+                return listOf(IAstModification.ReplaceNode(forLoop, scope, parent))
+            }
+        }
+        val iterable = (forLoop.iterable as? IdentifierReference)?.targetVarDecl(program)
+        if(iterable!=null) {
+            if(iterable.datatype==DataType.STR) {
+                val sv = iterable.value as StringLiteralValue
+                val size = sv.value.length
+                if(size==1) {
+                    // loop over string of length 1 -> just assign the single character
+                    val character = compTarget.encodeString(sv.value, sv.altEncoding)[0]
+                    val byte = NumericLiteralValue(DataType.UBYTE, character, iterable.position)
+                    val scope = AnonymousScope(mutableListOf(), forLoop.position)
+                    scope.statements.add(Assignment(AssignTarget(forLoop.loopVar, null, null, forLoop.position), byte, forLoop.position))
+                    scope.statements.addAll(forLoop.body.statements)
+                    return listOf(IAstModification.ReplaceNode(forLoop, scope, parent))
+                }
+            }
+            else if(iterable.datatype in ArrayDatatypes) {
+                val size = iterable.arraysize!!.constIndex()
+                if(size==1) {
+                    // loop over array of length 1 -> just assign the single value
+                    val av = (iterable.value as ArrayLiteralValue).value[0].constValue(program)?.number
+                    if(av!=null) {
+                        val scope = AnonymousScope(mutableListOf(), forLoop.position)
+                        scope.statements.add(Assignment(
+                                AssignTarget(forLoop.loopVar, null, null, forLoop.position), NumericLiteralValue.optimalInteger(av.toInt(), iterable.position),
+                                forLoop.position))
+                        scope.statements.addAll(forLoop.body.statements)
+                        return listOf(IAstModification.ReplaceNode(forLoop, scope, parent))
+                    }
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun before(untilLoop: UntilLoop, parent: Node): Iterable<IAstModification> {
+        val constvalue = untilLoop.condition.constValue(program)
+        if(constvalue!=null) {
+            if(constvalue.asBooleanValue) {
+                // always true -> keep only the statement block (if there are no break statements)
+                errors.warn("condition is always true", untilLoop.condition.position)
+                if(!hasBreak(untilLoop.body))
+                    return listOf(IAstModification.ReplaceNode(untilLoop, untilLoop.body, parent))
+            } else {
+                // always false
+                val forever = RepeatLoop(null, untilLoop.body, untilLoop.position)
+                return listOf(IAstModification.ReplaceNode(untilLoop, forever, parent))
+            }
+        }
+        return noModifications
+    }
+
+    override fun before(whileLoop: WhileLoop, parent: Node): Iterable<IAstModification> {
+        val constvalue = whileLoop.condition.constValue(program)
+        if(constvalue!=null) {
+            return if(constvalue.asBooleanValue) {
+                // always true
+                val forever = RepeatLoop(null, whileLoop.body, whileLoop.position)
+                listOf(IAstModification.ReplaceNode(whileLoop, forever, parent))
+            } else {
+                // always false -> remove the while statement altogether
+                errors.warn("condition is always false", whileLoop.condition.position)
+                listOf(IAstModification.Remove(whileLoop, parent as IStatementContainer))
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(repeatLoop: RepeatLoop, parent: Node): Iterable<IAstModification> {
+        val iter = repeatLoop.iterations
+        if(iter!=null) {
+            if(repeatLoop.body.isEmpty()) {
+                errors.warn("empty loop removed", repeatLoop.position)
+                return listOf(IAstModification.Remove(repeatLoop, parent as IStatementContainer))
+            }
+            val iterations = iter.constValue(program)?.number?.toInt()
+            if (iterations == 0) {
+                errors.warn("iterations is always 0, removed loop", iter.position)
+                return listOf(IAstModification.Remove(repeatLoop, parent as IStatementContainer))
+            }
+            if (iterations == 1) {
+                errors.warn("iterations is always 1", iter.position)
+                return listOf(IAstModification.ReplaceNode(repeatLoop, repeatLoop.body, parent))
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(jump: Jump, parent: Node): Iterable<IAstModification> {
+        // if the jump is to the next statement, remove the jump
+        val scope = jump.parent as IStatementContainer
+        val label = jump.identifier?.targetStatement(program)
+        if(label!=null && scope.statements.indexOf(label) == scope.statements.indexOf(jump)+1)
+            return listOf(IAstModification.Remove(jump, scope))
+
+        return noModifications
+    }
+
+    override fun before(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+
+        val binExpr = assignment.value as? BinaryExpression
+        if(binExpr!=null) {
+            if(binExpr.left isSameAs assignment.target) {
+                val rExpr = binExpr.right as? BinaryExpression
+                if(rExpr!=null) {
+                    val op1 = binExpr.operator
+                    val op2 = rExpr.operator
+
+                    if(rExpr.left is NumericLiteralValue && op2 in associativeOperators) {
+                        // associative operator, make sure the constant numeric value is second (right)
+                        return listOf(IAstModification.SwapOperands(rExpr))
+                    }
+
+                    val rNum = (rExpr.right as? NumericLiteralValue)?.number
+                    if(rNum!=null) {
+                        if (op1 == "+" || op1 == "-") {
+                            if (op2 == "+") {
+                                // A = A +/- B + N
+                                val expr2 = BinaryExpression(binExpr.left, binExpr.operator, rExpr.left, binExpr.position)
+                                val addConstant = Assignment(
+                                        assignment.target,
+                                        BinaryExpression(binExpr.left, "+", rExpr.right, rExpr.position),
+                                        assignment.position
+                                )
+                                return listOf(
+                                        IAstModification.ReplaceNode(binExpr, expr2, binExpr.parent),
+                                        IAstModification.InsertAfter(assignment, addConstant, parent as IStatementContainer))
+                            } else if (op2 == "-") {
+                                // A = A +/- B - N
+                                val expr2 = BinaryExpression(binExpr.left, binExpr.operator, rExpr.left, binExpr.position)
+                                val subConstant = Assignment(
+                                        assignment.target,
+                                        BinaryExpression(binExpr.left, "-", rExpr.right, rExpr.position),
+                                        assignment.position
+                                )
+                                return listOf(
+                                        IAstModification.ReplaceNode(binExpr, expr2, binExpr.parent),
+                                        IAstModification.InsertAfter(assignment, subConstant, parent as IStatementContainer))
+                            }
+                        }
+                    }
+                }
+            }
+
+            if(binExpr.operator in associativeOperators && binExpr.right isSameAs assignment.target) {
+                // associative operator, swap the operands so that the assignment target is first (left)
+                // unless the other operand is the same in which case we don't swap (endless loop!)
+                if (!(binExpr.left isSameAs binExpr.right))
+                    return listOf(IAstModification.SwapOperands(binExpr))
+            }
+
+        }
+
+        return noModifications
+    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        if(assignment.target isSameAs assignment.value) {
+            // remove assignment to self
+            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+        }
+
+        val targetIDt = assignment.target.inferType(program)
+        if(!targetIDt.isKnown)
+            throw FatalAstException("can't infer type of assignment target")
+
+        // optimize binary expressions a bit
+        val targetDt = targetIDt.getOr(DataType.UNDEFINED)
+        val bexpr=assignment.value as? BinaryExpression
+        if(bexpr!=null) {
+            val rightCv = bexpr.right.constValue(program)?.number?.toDouble()
+            if (rightCv != null && assignment.target isSameAs bexpr.left) {
+                // assignments of the form:  X = X <operator> <expr>
+                // remove assignments that have no effect (such as X=X+0)
+                // optimize/rewrite some other expressions
+                val vardeclDt = (assignment.target.identifier?.targetVarDecl(program))?.type
+                when (bexpr.operator) {
+                    "+" -> {
+                        if (rightCv == 0.0) {
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                        } else if (targetDt in IntegerDatatypes && floor(rightCv) == rightCv) {
+                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..4.0) {
+                                // replace by several INCs if it's not a memory address (inc on a memory mapped register doesn't work very well)
+                                val incs = AnonymousScope(mutableListOf(), assignment.position)
+                                repeat(rightCv.toInt()) {
+                                    incs.statements.add(PostIncrDecr(assignment.target.copy(), "++", assignment.position))
+                                }
+                                return listOf(IAstModification.ReplaceNode(assignment, incs, parent))
+                            }
+                        }
+                    }
+                    "-" -> {
+                        if (rightCv == 0.0) {
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                        } else if (targetDt in IntegerDatatypes && floor(rightCv) == rightCv) {
+                            if (vardeclDt != VarDeclType.MEMORY && rightCv in 1.0..4.0) {
+                                // replace by several DECs if it's not a memory address (dec on a memory mapped register doesn't work very well)
+                                val decs = AnonymousScope(mutableListOf(), assignment.position)
+                                repeat(rightCv.toInt()) {
+                                    decs.statements.add(PostIncrDecr(assignment.target.copy(), "--", assignment.position))
+                                }
+                                return listOf(IAstModification.ReplaceNode(assignment, decs, parent))
+                            }
+                        }
+                    }
+                    "*" -> if (rightCv == 1.0) return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    "/" -> if (rightCv == 1.0) return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    "**" -> if (rightCv == 1.0) return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    "|" -> if (rightCv == 0.0) return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    "^" -> if (rightCv == 0.0) return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    "<<" -> {
+                        if (rightCv == 0.0)
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    }
+                    ">>" -> {
+                        if (rightCv == 0.0)
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
+                    }
+                }
+
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(returnStmt: Return, parent: Node): Iterable<IAstModification> {
+        fun returnViaIntermediaryVar(value: Expression): Iterable<IAstModification>? {
+            val subr = returnStmt.definingSubroutine!!
+            val returnDt = subr.returntypes.single()
+            if (returnDt in IntegerDatatypes) {
+                // first assign to intermediary variable, then return that
+                subsThatNeedReturnVariable.add(Triple(subr, returnDt, returnStmt.position))
+                val returnValueIntermediary1 = IdentifierReference(listOf(retvarName), returnStmt.position)
+                val returnValueIntermediary2 = IdentifierReference(listOf(retvarName), returnStmt.position)
+                val tgt = AssignTarget(returnValueIntermediary1, null, null, returnStmt.position)
+                val assign = Assignment(tgt, value, returnStmt.position)
+                val returnReplacement = Return(returnValueIntermediary2, returnStmt.position)
+                return listOf(
+                    IAstModification.InsertBefore(returnStmt, assign, parent as IStatementContainer),
+                    IAstModification.ReplaceNode(returnStmt, returnReplacement, parent)
+                )
+            }
+            return null
+        }
+
+        when(returnStmt.value) {
+            is PrefixExpression -> {
+                val mod = returnViaIntermediaryVar(returnStmt.value!!)
+                if(mod!=null)
+                    return mod
+            }
+            is BinaryExpression -> {
+                val mod = returnViaIntermediaryVar(returnStmt.value!!)
+                if(mod!=null)
+                    return mod
+            }
+            else -> {}
+        }
+
+        return super.after(returnStmt, parent)
+    }
+
+    private fun hasBreak(scope: IStatementContainer): Boolean {
+
+        class Searcher: IAstVisitor
+        {
+            var count=0
+
+            override fun visit(breakStmt: Break) {
+                count++
+            }
+        }
+
+        val s=Searcher()
+        for(stmt in scope.statements) {
+            stmt.accept(s)
+            if(s.count>0)
+                return true
+        }
+        return s.count > 0
+    }
+
+}

codeOptimizers/src/prog8/optimizer/UnusedCodeRemover.kt

@@ -0,0 +1,144 @@
+package prog8.optimizer
+
+import prog8.ast.*
+import prog8.ast.base.VarDeclType
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.PrefixExpression
+import prog8.ast.expressions.TypecastExpression
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compilerinterface.CallGraph
+import prog8.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
+import prog8.compilerinterface.isInRegularRAMof
+
+
+class UnusedCodeRemover(private val program: Program,
+                        private val errors: IErrorReporter,
+                        private val compTarget: ICompilationTarget
+): AstWalker() {
+
+    private val callgraph = CallGraph(program)
+
+    override fun before(module: Module, parent: Node): Iterable<IAstModification> {
+        return if (!module.isLibrary && (module.containsNoCodeNorVars || callgraph.unused(module)))
+            listOf(IAstModification.Remove(module, parent as IStatementContainer))
+        else
+            noModifications
+    }
+
+    override fun before(breakStmt: Break, parent: Node): Iterable<IAstModification> {
+        reportUnreachable(breakStmt, parent as IStatementContainer)
+        return emptyList()
+    }
+
+    override fun before(jump: Jump, parent: Node): Iterable<IAstModification> {
+        reportUnreachable(jump, parent as IStatementContainer)
+        return emptyList()
+    }
+
+    override fun before(returnStmt: Return, parent: Node): Iterable<IAstModification> {
+        reportUnreachable(returnStmt, parent as IStatementContainer)
+        return emptyList()
+    }
+
+    override fun before(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        if(functionCallStatement.target.nameInSource.last() == "exit")
+            reportUnreachable(functionCallStatement, parent as IStatementContainer)
+        return emptyList()
+    }
+
+    private fun reportUnreachable(stmt: Statement, parent: IStatementContainer) {
+        when(val next = stmt.nextSibling()) {
+            null, is Label, is Directive, is VarDecl, is InlineAssembly, is Subroutine -> {}
+            else -> errors.warn("unreachable code", next.position)
+        }
+    }
+
+    override fun after(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
+        val removeDoubleAssignments = deduplicateAssignments(scope.statements)
+        return removeDoubleAssignments.map { IAstModification.Remove(it, scope) }
+    }
+
+    override fun after(block: Block, parent: Node): Iterable<IAstModification> {
+        if("force_output" !in block.options()) {
+            if (block.containsNoCodeNorVars) {
+                if(block.name != internedStringsModuleName)
+                    errors.warn("removing unused block '${block.name}'", block.position)
+                return listOf(IAstModification.Remove(block, parent as IStatementContainer))
+            }
+            if(callgraph.unused(block)) {
+                errors.warn("removing unused block '${block.name}'", block.position)
+                return listOf(IAstModification.Remove(block, parent as IStatementContainer))
+            }
+        }
+
+        val removeDoubleAssignments = deduplicateAssignments(block.statements)
+        return removeDoubleAssignments.map { IAstModification.Remove(it, block) }
+    }
+
+    override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        val forceOutput = "force_output" in subroutine.definingBlock.options()
+        if (subroutine !== program.entrypoint && !forceOutput && !subroutine.inline && !subroutine.isAsmSubroutine) {
+            if(callgraph.unused(subroutine)) {
+                if(subroutine.containsNoCodeNorVars) {
+                    if(!subroutine.definingModule.isLibrary)
+                        errors.warn("removing empty subroutine '${subroutine.name}'", subroutine.position)
+                    val removals = mutableListOf(IAstModification.Remove(subroutine, parent as IStatementContainer))
+                    callgraph.calledBy[subroutine]?.let {
+                        for(node in it)
+                            removals.add(IAstModification.Remove(node, node.parent as IStatementContainer))
+                    }
+                    return removals
+                }
+                if(!subroutine.definingModule.isLibrary)
+                    errors.warn("removing unused subroutine '${subroutine.name}'", subroutine.position)
+                return listOf(IAstModification.Remove(subroutine, parent as IStatementContainer))
+            }
+        }
+
+        val removeDoubleAssignments = deduplicateAssignments(subroutine.statements)
+        return removeDoubleAssignments.map { IAstModification.Remove(it, subroutine) }
+    }
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        if(decl.type==VarDeclType.VAR) {
+            val forceOutput = "force_output" in decl.definingBlock.options()
+            if (!forceOutput && !decl.autogeneratedDontRemove && !decl.sharedWithAsm && !decl.definingBlock.isInLibrary) {
+                if (callgraph.unused(decl)) {
+                    errors.warn("removing unused variable '${decl.name}'", decl.position)
+                    return listOf(IAstModification.Remove(decl, parent as IStatementContainer))
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun deduplicateAssignments(statements: List<Statement>): List<Assignment> {
+        // removes 'duplicate' assignments that assign the same target directly after another
+        val linesToRemove = mutableListOf<Assignment>()
+
+        for (stmtPairs in statements.windowed(2, step = 1)) {
+            val assign1 = stmtPairs[0] as? Assignment
+            val assign2 = stmtPairs[1] as? Assignment
+            if (assign1 != null && assign2 != null && !assign2.isAugmentable) {
+                if (assign1.target.isSameAs(assign2.target, program) && assign1.target.isInRegularRAMof(compTarget.machine))  {
+                    if(assign2.target.identifier==null || !assign2.value.referencesIdentifier(*(assign2.target.identifier!!.nameInSource.toTypedArray())))
+                        // only remove the second assignment if its value is a simple expression!
+                        when(assign2.value) {
+                            is PrefixExpression,
+                            is BinaryExpression,
+                            is TypecastExpression,
+                            is FunctionCall -> { /* don't remove */ }
+                            else -> linesToRemove.add(assign1)
+                        }
+                }
+            }
+        }
+
+        return linesToRemove
+    }
+}

codeOptimizers/test/readme.txt

@@ -0,0 +1,2 @@
+Unittests for things in this module are located in the Compiler module instead,
+for convenience sake - and to not spread the test cases around too much.

compiler/build.gradle

@@ -0,0 +1,90 @@
+plugins {
+    id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm"
+    id 'com.github.johnrengelman.shadow' version '7.1.0'
+}
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+def prog8version = rootProject.file('compiler/res/version.txt').text.trim()
+
+dependencies {
+    implementation project(':compilerInterfaces')
+    implementation project(':codeOptimizers')
+    implementation project(':compilerAst')
+    implementation project(':codeGeneration')
+    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
+    // implementation "org.jetbrains.kotlin:kotlin-reflect"
+    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.3'
+    implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.12"
+
+    testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
+    testImplementation 'org.junit.jupiter:junit-jupiter-api:5.7.2'
+    testImplementation 'org.hamcrest:hamcrest:2.2'
+    testRuntimeOnly 'org.junit.jupiter:junit-jupiter-engine:5.7.2'
+}
+
+configurations.all {
+    exclude group: 'com.ibm.icu', module: 'icu4j'
+    exclude group: "org.antlr", module: "antlr4"
+}
+
+configurations {
+    // strange antlr plugin issue, see https://github.com/gradle/gradle/issues/820
+    // this avoids linking in the complete antlr binary jar
+    compile {
+        extendsFrom = extendsFrom.findAll { it != configurations.antlr }
+    }
+}
+
+
+sourceSets {
+    main {
+        java {
+            srcDirs = ["${project.projectDir}/src"]
+        }
+        resources {
+            srcDirs = ["${project.projectDir}/res"]
+        }
+    }
+    test {
+        java {
+            srcDir "${project.projectDir}/test"
+            srcDir "${project(':compilerAst').projectDir}/test/helpers"
+        }
+    }
+}
+
+startScripts.enabled = true
+
+application {
+    mainClass = 'prog8.CompilerMainKt'
+    applicationName = 'p8compile'
+}
+
+shadowJar {
+    archiveBaseName = 'prog8compiler'
+    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"
+    }
+}
+
+build.finalizedBy installDist, installShadowDist

compiler/compiler.iml

@@ -1,17 +1,29 @@
 <?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="openjdk-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="module" module-name="compilerAst" />
+    <orderEntry type="library" name="Python 3.9 interpreter library" level="application" />
+    <orderEntry type="library" name="hamcrest" level="project" />
+    <orderEntry type="library" name="jetbrains.kotlinx.cli.jvm" level="project" />
+    <orderEntry type="library" name="junit.jupiter" level="project" />
+    <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
+    <orderEntry type="module" module-name="codeOptimizers" />
+    <orderEntry type="module" module-name="compilerInterfaces" />
+    <orderEntry type="module" module-name="codeGeneration" />
   </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
+
+%option enable_floats
+
+floats {
+	; ---- this block contains C-64 floating point related functions ----
+
+        const float  PI     = 3.141592653589793
+        const float  TWOPI  = 6.283185307179586
+
+        ubyte[5] tempvar_swap_float     ; used for some swap() operations
+
+; ---- 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 (fac1 = -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 +1,369 @@
+%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 +1,735 @@
+; 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 {
+        &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 (A=lo,X=mid,Y=high)
+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 in AY 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 initial power-on 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)
+        ;     note: the system irq handler has to be active for this to work as it depends on the system jiffy clock
+        repeat jiffies {
+            ubyte jiff = lsb(c64.RDTIM16())
+            while jiff==lsb(c64.RDTIM16()) {
+                ; wait until 1 jiffy has passed
+            }
+        }
+    }
+
+    asmsub waitvsync() clobbers(A) {
+        ; --- busy wait till the next vsync has occurred (approximately), without depending on custom irq handling.
+        ;     note: a more accurate way to wait for vsync is to set up a vsync irq handler instead.
+        %asm {{
+-           bit  c64.SCROLY
+            bpl  -
+-           bit  c64.SCROLY
+            bmi  -
+            rts
+        }}
+    }
+
+    inline asmsub waitrastborder() {
+        ; --- busy wait till the raster position has reached the bottom screen border (approximately)
+        ;     note: a more accurate way to do this is by using a raster irq handler instead.
+        %asm {{
+-           bit  c64.SCROLY
+            bpl  -
+        }}
+    }
+
+    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
+
+    &ubyte r0L  = $cf00
+    &ubyte r1L  = $cf02
+    &ubyte r2L  = $cf04
+    &ubyte r3L  = $cf06
+    &ubyte r4L  = $cf08
+    &ubyte r5L  = $cf0a
+    &ubyte r6L  = $cf0c
+    &ubyte r7L  = $cf0e
+    &ubyte r8L  = $cf10
+    &ubyte r9L  = $cf12
+    &ubyte r10L = $cf14
+    &ubyte r11L = $cf16
+    &ubyte r12L = $cf18
+    &ubyte r13L = $cf1a
+    &ubyte r14L = $cf1c
+    &ubyte r15L = $cf1e
+
+    &ubyte r0H  = $cf01
+    &ubyte r1H  = $cf03
+    &ubyte r2H  = $cf05
+    &ubyte r3H  = $cf07
+    &ubyte r4H  = $cf09
+    &ubyte r5H  = $cf0b
+    &ubyte r6H  = $cf0d
+    &ubyte r7H  = $cf0f
+    &ubyte r8H  = $cf11
+    &ubyte r9H  = $cf13
+    &ubyte r10H = $cf15
+    &ubyte r11H = $cf17
+    &ubyte r12H = $cf19
+    &ubyte r13H = $cf1b
+    &ubyte r14H = $cf1d
+    &ubyte r15H = $cf1f
+}

compiler/res/prog8lib/c64/textio.p8

@@ -0,0 +1,617 @@
+; 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
+
+%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
+
+%option enable_floats
+
+floats {
+	; ---- this block contains C-64 compatible floating point related functions ----
+	;      the addresses are from cx16 V39 emulator and roms! they won't work on older versions.
+
+        const float  PI     = 3.141592653589793
+        const float  TWOPI  = 6.283185307179586
+
+        ubyte[5] tempvar_swap_float     ; used for some swap() operations
+
+
+; ---- 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 $fe30 = CONUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory  in A/Y into fac2
+romsub $fe33 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
+romsub $fe36 = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive!
+romsub $fe39 = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
+romsub $fe3c = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
+
+romsub $fe42 = MOVFM(uword mflpt @ AY) clobbers(A,X,Y)      ; load mflpt value from memory  in A/Y into fac1
+romsub $fe45 = MOVMF(uword mflpt @ XY) clobbers(A,X,Y)      ; store fac1 to memory  X/Y as 5-byte mflpt
+romsub $fe48 = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+romsub $fe4b = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded)
+romsub $fe4e = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+romsub $fe54 = SIGN() clobbers(X,Y) -> ubyte @ A            ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+romsub $fe57 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
+romsub $fe5a = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
+romsub $fe66 = ABS() clobbers(A,X,Y)                        ; fac1 = ABS(fac1)
+romsub $fe69 = 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 $fe72 = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
+romsub $fe78 = FINLOG(byte value @A) clobbers (A, X, Y)           ; fac1 += signed byte in A
+romsub $fe7b = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
+romsub $fe81 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+romsub $fe84 = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+romsub $fe8a = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
+romsub $fe8d = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+romsub $fe96 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
+romsub $fe99 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+romsub $fe9c = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+romsub $fe9f = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+romsub $fea2 = 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 +1,989 @@
+; 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) {
+        when active_mode {
+            1, 5 -> {
+                ; monochrome, lo-res
+                cx16.r15L = gfx2.plot.bits[x as ubyte & 7]           ; bitmask
+                if color {
+                    if monochrome_dont_stipple_flag {
+                        ; draw continuous line.
+                        position2(x,y,true)
+                        if active_mode==1
+                            set_both_strides(11)    ; 40 increment = 1 line in 320 px monochrome
+                        else
+                            set_both_strides(12)    ; 80 increment = 1 line in 640 px monochrome
+                        repeat height {
+                            %asm {{
+                                lda  cx16.VERA_DATA0
+                                ora  cx16.r15L
+                                sta  cx16.VERA_DATA1
+                            }}
+                        }
+                    } else {
+                        ; draw stippled line.
+                        if x&1 {
+                            y++
+                            height--
+                        }
+                        position2(x,y,true)
+                        if active_mode==1
+                            set_both_strides(12)    ; 80 increment = 2 line in 320 px monochrome
+                        else
+                            set_both_strides(13)    ; 160 increment = 2 line in 640 px monochrome
+                        repeat height/2 {
+                            %asm {{
+                                lda  cx16.VERA_DATA0
+                                ora  cx16.r15L
+                                sta  cx16.VERA_DATA1
+                            }}
+                        }
+                    }
+                } else {
+                    position2(x,y,true)
+                    cx16.r15 = ~cx16.r15    ; erase pixels
+                    if active_mode==1
+                        set_both_strides(11)    ; 40 increment = 1 line in 320 px monochrome
+                    else
+                        set_both_strides(12)    ; 80 increment = 1 line in 640 px monochrome
+                    repeat height {
+                        %asm {{
+                            lda  cx16.VERA_DATA0
+                            and  cx16.r15L
+                            sta  cx16.VERA_DATA1
+                        }}
+                    }
+                }
+            }
+            4 -> {
+                ; lores 256c
+                ; set vera auto-increment to 320 pixel increment (=next line)
+                position(x,y)
+                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
+                ; use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment is possible
+                if height==0
+                    return
+                position2(x,y,true)
+                set_both_strides(13)    ; 160 increment = 1 line in 640 px 4c mode
+                color &= 3
+                color <<= gfx2.plot.shift4c[lsb(x) & 3]
+                ubyte mask = gfx2.plot.mask4c[lsb(x) & 3]
+                repeat height {
+                    %asm {{
+                        lda  cx16.VERA_DATA0
+                        and  mask
+                        ora  color
+                        sta  cx16.VERA_DATA1
+                    }}
+                }
+            }
+        }
+
+        sub set_both_strides(ubyte stride) {
+            stride <<= 4
+            cx16.VERA_CTRL = 0
+            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
+            cx16.VERA_CTRL = 1
+            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
+        }
+
+    }
+
+    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]
+
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    cx16.r0L = lsb(x) & 7       ; xbits
+                    x /= 8
+                    x += y*(320/8)
+                    %asm {{
+                        stz  cx16.VERA_CTRL
+                        stz  cx16.VERA_ADDR_H
+                        lda  x+1
+                        sta  cx16.VERA_ADDR_M
+                        lda  x
+                        sta  cx16.VERA_ADDR_L
+                        ldy  cx16.r0L       ; xbits
+                        lda  bits,y
+                        ldy  color
+                        beq  +
+                        tsb  cx16.VERA_DATA0
+                        bra  ++
++                       trb  cx16.VERA_DATA0
++
+                    }}
+                }
+            }
+            ; TODO mode 2,3
+            4 -> {
+                ; lores 256c
+                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                %asm {{
+                    stz  cx16.VERA_CTRL
+                    lda  cx16.r1
+                    ora  #%00010000     ; enable auto-increment so next_pixel() can be used after this
+                    sta  cx16.VERA_ADDR_H
+                    lda  cx16.r0+1
+                    sta  cx16.VERA_ADDR_M
+                    lda  cx16.r0
+                    sta  cx16.VERA_ADDR_L
+                    lda  color
+                    sta  cx16.VERA_DATA0
+                }}
+            }
+            5 -> {
+                ; highres monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    cx16.r0L = lsb(x) & 7       ; xbits
+                    x /= 8
+                    x += y*(640/8)
+                    %asm {{
+                        stz  cx16.VERA_CTRL
+                        stz  cx16.VERA_ADDR_H
+                        lda  x+1
+                        sta  cx16.VERA_ADDR_M
+                        lda  x
+                        sta  cx16.VERA_ADDR_L
+                        ldy  cx16.r0L           ; xbits
+                        lda  bits,y
+                        ldy  color
+                        beq  +
+                        tsb  cx16.VERA_DATA0
+                        bra  ++
++                       trb  cx16.VERA_DATA0
++
+                    }}
+                }
+            }
+            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)
+                cx16.r2L = lsb(x) & 3       ; xbits
+                color &= 3
+                color <<= shift4c[cx16.r2L]
+                %asm {{
+                    stz  cx16.VERA_CTRL
+                    lda  cx16.r1L
+                    sta  cx16.VERA_ADDR_H
+                    lda  cx16.r0H
+                    sta  cx16.VERA_ADDR_M
+                    lda  cx16.r0L
+                    sta  cx16.VERA_ADDR_L
+                    ldy  cx16.r2L           ; xbits
+                    lda  mask4c,y
+                    and  cx16.VERA_DATA0
+                    ora  color
+                    sta  cx16.VERA_DATA0
+                }}
+            }
+        }
+    }
+
+    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)
+            }
+        }
+    }
+
+    sub position2(uword @zp x, uword y, ubyte also_port_1) {
+        position(x, y)
+        if also_port_1 {
+            when active_mode {
+                1, 5 -> cx16.vaddr(0, cx16.r0, 1, 1)
+                ; TODO modes 2, 3
+                4, 6 -> {
+                    ubyte bank = lsb(cx16.r1)
+                    cx16.vaddr(bank, cx16.r0, 1, 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 multiples 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 +1,137 @@
+%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 +1,189 @@
+; 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 as uword * 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_all_black() {
+        set_monochrome($000, $000)
+    }
+
+    sub set_all_white() {
+        set_monochrome($fff, $fff)
+    }
+
+    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 +1,880 @@
+; 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
+
+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 cx16.numbanks()
+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 (A=lo,X=mid,Y=high)
+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 in AY for convenience
+    %asm {{
+        phx
+        jsr  c64.RDTIM
+        pha
+        txa
+        tay
+        pla
+        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
+
+    &ubyte r0L  = $0002
+    &ubyte r1L  = $0004
+    &ubyte r2L  = $0006
+    &ubyte r3L  = $0008
+    &ubyte r4L  = $000a
+    &ubyte r5L  = $000c
+    &ubyte r6L  = $000e
+    &ubyte r7L  = $0010
+    &ubyte r8L  = $0012
+    &ubyte r9L  = $0014
+    &ubyte r10L = $0016
+    &ubyte r11L = $0018
+    &ubyte r12L = $001a
+    &ubyte r13L = $001c
+    &ubyte r14L = $001e
+    &ubyte r15L = $0020
+
+    &ubyte r0H  = $0003
+    &ubyte r1H  = $0005
+    &ubyte r2H  = $0007
+    &ubyte r3H  = $0009
+    &ubyte r4H  = $000b
+    &ubyte r5H  = $000d
+    &ubyte r6H  = $000f
+    &ubyte r7H  = $0011
+    &ubyte r8H  = $0013
+    &ubyte r9H  = $0015
+    &ubyte r10H = $0017
+    &ubyte r11H = $0019
+    &ubyte r12H = $001b
+    &ubyte r13H = $001d
+    &ubyte r14H = $001f
+    &ubyte r15H = $0021
+
+
+; 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   = $9f00                  ;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   = $9f10                  ;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
+
+    &ubyte  ym2151adr	= $9f40
+    &ubyte  ym2151dat	= $9f41
+
+    const uword  extdev	= $9f60
+
+
+; ---- 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 (v39+, used to be cx16.d1prb $9f60 in v38)
+    }}
+}
+
+inline asmsub rambank(ubyte rambank @A) {
+    ; -- set the ram bank
+    %asm {{
+        sta  $00            ; ram bank register (v39+, used to be cx16.d1pra $9f61 in v38)
+    }}
+}
+
+asmsub numbanks() -> ubyte @A {
+    ; -- uses MEMTOP's cx16 extension to query the number of available RAM banks. (each is 8 Kb)
+    %asm {{
+        phx
+        sec
+        jsr  c64.MEMTOP
+        plx
+        rts
+    }}
+}
+
+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
+    }}
+}
+
+inline asmsub joystick_get2(ubyte joynr @A) clobbers(Y) -> uword @AX  {
+    ; convenience routine to get the joystick state without requiring inline assembly that deals with the multiple return values.
+    ; Also disables interrupts to avoid the IRQ race condition mentioned here: https://github.com/commanderx16/x16-rom/issues/203
+    ; TODO once that issue is resolved, this routine can be redefined as:  romsub $ff56 = joystick_get2(ubyte joynr @A) clobbers(Y) -> uword @AX
+    %asm {{
+        sei
+        jsr  cx16.joystick_get
+        cli
+    }}
+}
+
+
+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
+        lda  #$80
+        sta  VERA_CTRL
+        stz  $01        ; select rom bank 0 (enable kernal)
+        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 initial power-on Basic prompt.
+        %asm {{
+            sei
+            stz  $01            ; bank the kernal in
+            jmp  (cx16.RESET_VEC)
+        }}
+    }
+
+    asmsub wait(uword jiffies @AY) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds) (N or N+1)
+        ;     note: regular system vsync irq handler must be running, and no nother irqs
+        %asm {{
+-           wai             ; wait for irq (assume it was vsync)
+            cmp  #0
+            bne  +
+            dey
++           dec  a
+            bne  -
+            cpy  #0
+            bne  -
+            rts
+        }}
+    }
+
+    inline asmsub waitvsync()  {
+        ; --- suspend execution until the next vsync has occurred, without depending on custom irq handling.
+        ;     note: system vsync irq handler has to be active for this routine to work (and no other irqs-- which is the default).
+        ;     note: a more accurate way to wait for vsync is to set up a vsync irq handler instead.
+        %asm {{
+            wai
+        }}
+    }
+
+    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 +1,748 @@
+; 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
+
+%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 +1,492 @@
+; 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
+        }
+    }
+
+
+    ; ----- iterative file saver functions (uses io channel 14) -----
+
+    sub f_open_w(ubyte drivenumber, uword filenameptr) -> ubyte {
+        ; -- open a file for iterative writing with f_write
+        f_close_w()
+
+        c64.SETNAM(string.length(filenameptr), filenameptr)
+        c64.SETLFS(14, drivenumber, 1)
+        void c64.OPEN()          ; open 14,8,1,"filename"
+        if_cc {
+            void c64.CHKOUT(14)        ; use #14 as input channel
+            return not c64.READST()
+        }
+        f_close_w()
+        return false
+    }
+
+    sub f_write(uword bufferpointer, uword num_bytes) -> ubyte {
+        ; -- write the given umber of bytes to the currently open file
+        if num_bytes!=0 {
+            void c64.CHKOUT(14)        ; use #14 as input channel again
+            repeat num_bytes {
+                c64.CHROUT(@(bufferpointer))
+                bufferpointer++
+            }
+            return not c64.READST()
+        }
+        return true
+    }
+
+    sub f_close_w() {
+        ; -- end an iterative file writing session (close channels).
+        c64.CLRCHN()
+        c64.CLOSE(14)
+    }
+
+
+    ; ---- other functions ----
+
+    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++
+        }
+
+        @(messageptr) = 0
+done:
+        c64.CLRCHN()        ; restore default i/o devices
+        c64.CLOSE(15)
+        return filename
+
+io_error:
+        filename = "?disk error"
+        goto done
+    }
+
+    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
+        first_byte = 0      ; result var reuse
+
+        %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
+        }}
+
+        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 {
-	%asminclude "library:math.asm", ""
+math {
+	%asminclude "library:math.asm"
 }

compiler/res/prog8lib/prog8_lib.p8

@@ -0,0 +1,86 @@
+; 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"
+
+        ; TODO these retval variables are no longer used???
+	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)
+	; NOTE: these variables are checked in the StatementReorderer (in fun after(decl: VarDecl)), for these exact names!
+
+	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 depending 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) -> ubyte @Y {
+        ; Lowercases the petscii string in-place. Returns length of the string.
+        ; (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) -> ubyte @Y {
+        ; Uppercases the petscii string in-place. Returns length of the string.
+        %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 @@
-1.0 (beta)
+7.2