preparing v7.6

sizeof(pointer) is hardcoded as 2 now

.gitignore

@@ -26,8 +26,9 @@ parser.out
 parsetab.py
 .pytest_cache/
 .attach_pid*
+compiler/lib/
 
 .gradle
 /prog8compiler.jar
 sd*.img
-
+*.d64

.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/codeStyles/Project.xml

@@ -0,0 +1,10 @@
+<component name="ProjectCodeStyleConfiguration">
+  <code_scheme name="Project" version="173">
+    <JetCodeStyleSettings>
+      <option name="CODE_STYLE_DEFAULTS" value="KOTLIN_OFFICIAL" />
+    </JetCodeStyleSettings>
+    <codeStyleSettings language="kotlin">
+      <option name="CODE_STYLE_DEFAULTS" value="KOTLIN_OFFICIAL" />
+    </codeStyleSettings>
+  </code_scheme>
+</component>

.idea/codeStyles/codeStyleConfig.xml

@@ -0,0 +1,5 @@
+<component name="ProjectCodeStyleConfiguration">
+  <state>
+    <option name="USE_PER_PROJECT_SETTINGS" value="true" />
+  </state>
+</component>

.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/inspectionProfiles/Project_Default.xml

@@ -3,14 +3,12 @@
     <option name="myName" value="Project Default" />
     <inspection_tool class="DuplicatedCode" enabled="true" level="WEAK WARNING" enabled_by_default="true">
       <Languages>
-        <language minSize="100" isEnabled="false" name="JavaScript" />
-        <language isEnabled="false" name="Groovy" />
-        <language isEnabled="false" name="Style Sheets" />
         <language minSize="70" name="Kotlin" />
-        <language isEnabled="false" name="TypeScript" />
-        <language isEnabled="false" name="ActionScript" />
+        <language isEnabled="false" name="Groovy" />
       </Languages>
     </inspection_tool>
+    <inspection_tool class="IncompleteDestructuring" enabled="true" level="WARNING" enabled_by_default="true" />
+    <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" />

.idea/kotlinc.xml

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <project version="4">
   <component name="Kotlin2JvmCompilerArguments">
-    <option name="jvmTarget" value="1.8" />
+    <option name="jvmTarget" value="11" />
   </component>
 </project>

.idea/libraries/antlr_4_9_complete.xml

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

.idea/libraries/antlr_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/antlr_runtime_4_9.xml

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

.idea/libraries/dbus_java_3_2_4.xml

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

.idea/libraries/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/io_kotest_assertions_core_jvm.xml

@@ -0,0 +1,22 @@
+<component name="libraryTable">
+  <library name="io.kotest.assertions.core.jvm" type="repository">
+    <properties maven-id="io.kotest:kotest-assertions-core-jvm:4.6.3" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/4.6.3/kotest-assertions-core-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.5.0/kotlin-stdlib-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.5.0/kotlin-stdlib-1.5.0.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-jdk7/1.5.0/kotlin-stdlib-jdk7-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/4.6.3/kotest-assertions-shared-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.9/java-diff-utils-4.9.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.5.0/kotlinx-coroutines-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.5.0/kotlin-reflect-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.5.0/kotlinx-coroutines-core-jvm-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.5.0/kotlin-stdlib-common-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/4.6.3/kotest-common-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/4.6.3/kotest-assertions-api-jvm-4.6.3.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/io_kotest_property_jvm.xml

@@ -0,0 +1,24 @@
+<component name="libraryTable">
+  <library name="io.kotest.property.jvm" type="repository">
+    <properties maven-id="io.kotest:kotest-property-jvm:4.6.3" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-property-jvm/4.6.3/kotest-property-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.5.0/kotlin-stdlib-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.5.0/kotlin-stdlib-1.5.0.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-jdk7/1.5.0/kotlin-stdlib-jdk7-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/4.6.3/kotest-common-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/4.6.3/kotest-assertions-shared-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/4.6.3/kotest-assertions-api-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.5.0/kotlinx-coroutines-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.9/java-diff-utils-4.9.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/mifmif/generex/1.0.2/generex-1.0.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/dk/brics/automaton/automaton/1.11-8/automaton-1.11-8.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.5.0/kotlin-reflect-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.5.0/kotlinx-coroutines-core-jvm-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.5.0/kotlin-stdlib-common-1.5.0.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/io_kotest_runner_junit5_jvm.xml

@@ -0,0 +1,56 @@
+<component name="libraryTable">
+  <library name="io.kotest.runner.junit5.jvm" type="repository">
+    <properties maven-id="io.kotest:kotest-runner-junit5-jvm:4.6.3" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-runner-junit5-jvm/4.6.3/kotest-runner-junit5-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-api-jvm/4.6.3/kotest-framework-api-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/4.6.3/kotest-assertions-shared-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.9/java-diff-utils-4.9.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/4.6.3/kotest-common-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-engine-jvm/4.6.3/kotest-framework-engine-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/github/classgraph/classgraph/4.8.105/classgraph-4.8.105.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/ajalt/mordant/1.2.1/mordant-1.2.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/github/ajalt/colormath/1.2.0/colormath-1.2.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-script-util/1.5.0/kotlin-script-util-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/intellij/deps/trove4j/1.0.20181211/trove4j-1.0.20181211.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-daemon-client/1.5.0/kotlin-daemon-client-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core/1.3.8/kotlinx-coroutines-core-1.3.8.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-scripting-jvm/1.5.0/kotlin-scripting-jvm-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-scripting-common/1.5.0/kotlin-scripting-common-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-discovery-jvm/4.6.3/kotest-framework-discovery-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/4.6.3/kotest-assertions-core-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.5.0/kotlinx-coroutines-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/4.6.3/kotest-assertions-api-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-extensions-jvm/4.6.3/kotest-extensions-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/commons-io/commons-io/2.6/commons-io-2.6.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk/1.9.3/mockk-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-common/1.9.3/mockk-common-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-dsl/1.9.3/mockk-dsl-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-dsl-jvm/1.9.3/mockk-dsl-jvm-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-agent-jvm/1.9.3/mockk-agent-jvm-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-agent-api/1.9.3/mockk-agent-api-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/mockk/mockk-agent-common/1.9.3/mockk-agent-common-1.9.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/objenesis/objenesis/3.0.1/objenesis-3.0.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy/1.9.10/byte-buddy-1.9.10.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy-agent/1.9.10/byte-buddy-agent-1.9.10.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-concurrency-jvm/4.6.3/kotest-framework-concurrency-jvm-4.6.3.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.5.0/kotlinx-coroutines-core-jvm-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.5.0/kotlin-stdlib-common-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-engine/1.6.2/junit-platform-engine-1.6.2.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/junit/platform/junit-platform-commons/1.6.2/junit-platform-commons-1.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-suite-api/1.6.2/junit-platform-suite-api-1.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-launcher/1.6.2/junit-platform-launcher-1.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/junit/jupiter/junit-jupiter-api/5.6.2/junit-jupiter-api-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.5.0/kotlin-stdlib-jdk8-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.5.0/kotlin-stdlib-1.5.0.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-jdk7/1.5.0/kotlin-stdlib-jdk7-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-script-runtime/1.5.0/kotlin-script-runtime-1.5.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.5.0/kotlin-reflect-1.5.0.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/javax_json_api_1_1_4.xml

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

.idea/libraries/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/kotlinx_cli_jvm.xml

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

.idea/libraries/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_api_1_7_30.xml

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

.idea/libraries/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/libraries/takes_http.xml

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

.idea/libraries/unittest_libs.xml

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

.idea/misc.xml

@@ -4,7 +4,7 @@
     <option name="perGrammarGenerationSettings">
       <list>
         <PerGrammarGenerationSettings>
-          <option name="fileName" value="$PROJECT_DIR$/parser/antlr/prog8.g4" />
+          <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="" />
@@ -16,7 +16,10 @@
       </list>
     </option>
   </component>
-  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" default="false" project-jdk-name="Kotlin SDK" project-jdk-type="KotlinSDK">
+  <component name="FrameworkDetectionExcludesConfiguration">
+    <type id="Python" />
+  </component>
+  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" project-jdk-name="11" project-jdk-type="JavaSDK">
     <output url="file://$PROJECT_DIR$/out" />
   </component>
 </project>

.idea/modules.xml

@@ -2,7 +2,11 @@
 <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" />

.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

.travis.yml

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

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

@@ -1,5 +1,3 @@
-[![saythanks](https://img.shields.io/badge/say-thanks-ff69b4.svg)](https://saythanks.io/to/irmen)
-[![Build Status](https://travis-ci.org/irmen/prog8.svg?branch=master)](https://travis-ci.org/irmen/prog8)
 [![Documentation](https://readthedocs.org/projects/prog8/badge/?version=latest)](https://prog8.readthedocs.io/)
 
 Prog8 - Structured Programming Language for 8-bit 6502/65c02 microprocessors
@@ -7,9 +5,6 @@ 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 6502/6510/65c02 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).
@@ -19,27 +14,37 @@ 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?
 ------------------------
 
-- big reduction of source code length over raw assembly
+- 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
 - various data types other than just bytes (16-bit words, floats, strings)
-- automatic variable allocations, automatic string and array variables and string sharing
-- subroutines with an input- and output parameter signature
-- no stack frame allocations because parameters and local variables are automatically allocated statically
-- constant folding in expressions and other high-level program optimizations
+- 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
-- floating point operations  (requires the C64 Basic ROM routines for this)
-- 'when' statement to provide a concise jump table alternative to if/elseif chains
-- structs to group together sets of variables and manipulate them at once
+- ``when`` statement to provide a concise jump table alternative to if/elseif chains
+- ``in`` expression for concise and efficient multi-value/containment check 
 - many built-in functions such as ``sin``, ``cos``, ``rnd``, ``abs``, ``min``, ``max``, ``sqrt``, ``msb``, ``rol``, ``ror``, ``swap``, ``sort`` and ``reverse``
 - various powerful built-in libraries to do I/O, number conversions, graphics and more  
 - convenience abstractions for low level aspects such as ZeroPage handling, program startup, explicit memory addresses
-- fast execution speed due to compilation to native assembly code
 - 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.
+- 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:*
 
@@ -48,11 +53,12 @@ What does Prog8 provide?
 - 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
 
-*Two supported compiler targets* (contributions to improve these or to add support for other machines are welcome!):
+*Multiple supported compiler targets* (contributions to improve these or to add support for other machines are welcome!):
 
-- "c64": Commodore-64  (6510 CPU = almost a 6502)
+- "c64": Commodore-64  (6502 like CPU)
+- "c128": Commodore-128  (6502 like CPU - the Z80 cpu mode is not supported)
 - "cx16": [CommanderX16](https://www.commanderx16.com)  (65c02 CPU)
-- If you only use standard kernel and prog8 library routines, it is possible to compile the *exact same program* for both machines (just change the compiler target flag)!
+- If you only use standard kernal and prog8 library routines, it is possible to compile the *exact same program* for different machines (just change the compiler target flag)
 
 
 

build.gradle

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

c128floatproblem.asm

@@ -0,0 +1,48 @@
+
+; this program adds 2 floats 1.2345 + 4.4444 and prints the result 5.67879
+; it works fine on C64 and CX16 but fails to add the right number on C128
+; I think it has something to do with the banking on the C128
+; because several rom fp routines mention "value in bank 1"
+; but I don't know what this means and why the program as given doesn't work
+
+.cpu  '6502'
+
+* = $1300       ; sys 4864
+
+P8ZP_SCRATCH_W1 = 251    ; free zp word
+CHROUT = $ffd2
+
+; c128:
+MOVFM = $af63
+FADD = $af18
+FOUT = $af06
+
+;cx16:
+;MOVFM = $fe42
+;FADD = $fe12
+;FOUT = $fe7b
+
+;c64:
+;MOVFM = $bba2
+;FADD = $b867
+;FOUT = $bddd
+
+            lda  #<f1
+            ldy  #>f1
+            jsr  MOVFM       ; load FAC1
+            lda  #<f2
+            ldy  #>f2
+            jsr  FADD        ; add mem float to FAC1     DOESN'T WORK ON C128!?
+            jsr  FOUT        ; convert to string
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+_printloop  lda  (P8ZP_SCRATCH_W1),y
+            beq  _done
+            jsr  CHROUT
+            iny
+            bne  _printloop
+_done       jmp  _done
+
+f1	.byte  $81, $1e, $04, $18, $93  ; float 1.2345
+f2	.byte  $83, $0e, $38, $86, $59  ; float 4.4444

codeGeneration/build.gradle

@@ -0,0 +1,46 @@
+
+plugins {
+    id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm"
+}
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+compileKotlin {
+    kotlinOptions {
+        jvmTarget = javaVersion
+    }
+}
+
+compileTestKotlin {
+    kotlinOptions {
+        jvmTarget = 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"
+
+}
+
+sourceSets {
+    main {
+        java {
+            srcDirs = ["${project.projectDir}/src"]
+        }
+        resources {
+            srcDirs = ["${project.projectDir}/res"]
+        }
+    }
+}
+
+// note: there are no unit tests in this module!

codeGeneration/codeGeneration.iml

@@ -0,0 +1,16 @@
+<?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" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
+    </content>
+    <orderEntry type="inheritedJdk" />
+    <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" />
+  </component>
+</module>

codeGeneration/readme-tests.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.

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

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

codeGeneration/src/prog8/codegen/target/C128Target.kt

@@ -0,0 +1,41 @@
+package prog8.codegen.target
+
+import com.github.michaelbull.result.fold
+import prog8.ast.base.*
+import prog8.ast.expressions.Expression
+import prog8.ast.statements.RegisterOrStatusflag
+import prog8.ast.statements.Subroutine
+import prog8.codegen.target.c128.C128MachineDefinition
+import prog8.codegen.target.cbm.Petscii
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsEvalOrder
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsHaveRegisterClobberRisk
+import prog8.compilerinterface.ICompilationTarget
+
+
+object C128Target: ICompilationTarget {
+    override val name = "c128"
+    override val machine = C128MachineDefinition()
+    override fun encodeString(str: String, altEncoding: Boolean): List<UByte> {
+        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<UByte>, altEncoding: Boolean) =
+        if (altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun asmsubArgsEvalOrder(sub: Subroutine): List<Int> =
+        asmsub6502ArgsEvalOrder(sub)
+    override fun asmsubArgsHaveRegisterClobberRisk(args: List<Expression>, paramRegisters: List<RegisterOrStatusflag>) =
+        asmsub6502ArgsHaveRegisterClobberRisk(args, paramRegisters)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes, in PassByReferenceDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            else -> Int.MIN_VALUE
+        }
+    }
+}

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

@@ -0,0 +1,41 @@
+package prog8.codegen.target
+
+import com.github.michaelbull.result.fold
+import prog8.ast.base.*
+import prog8.ast.expressions.Expression
+import prog8.ast.statements.RegisterOrStatusflag
+import prog8.ast.statements.Subroutine
+import prog8.codegen.target.c64.C64MachineDefinition
+import prog8.codegen.target.cbm.Petscii
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsEvalOrder
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsHaveRegisterClobberRisk
+import prog8.compilerinterface.ICompilationTarget
+
+
+object C64Target: ICompilationTarget {
+    override val name = "c64"
+    override val machine = C64MachineDefinition()
+    override fun encodeString(str: String, altEncoding: Boolean): List<UByte> {
+        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<UByte>, altEncoding: Boolean) =
+        if (altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun asmsubArgsEvalOrder(sub: Subroutine): List<Int> =
+        asmsub6502ArgsEvalOrder(sub)
+    override fun asmsubArgsHaveRegisterClobberRisk(args: List<Expression>, paramRegisters: List<RegisterOrStatusflag>) =
+        asmsub6502ArgsHaveRegisterClobberRisk(args, paramRegisters)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes, in PassByReferenceDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            else -> Int.MIN_VALUE
+        }
+    }
+}

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

@@ -0,0 +1,45 @@
+package prog8.codegen.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.ast.expressions.Expression
+import prog8.ast.statements.RegisterOrStatusflag
+import prog8.ast.statements.Subroutine
+import prog8.codegen.target.cbm.Petscii
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsEvalOrder
+import prog8.codegen.target.cpu6502.codegen.asmsub6502ArgsHaveRegisterClobberRisk
+import prog8.codegen.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<UByte> {
+        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<UByte>, altEncoding: Boolean) =
+        if (altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun asmsubArgsEvalOrder(sub: Subroutine): List<Int> =
+        asmsub6502ArgsEvalOrder(sub)
+    override fun asmsubArgsHaveRegisterClobberRisk(args: List<Expression>, paramRegisters: List<RegisterOrStatusflag>) =
+        asmsub6502ArgsHaveRegisterClobberRisk(args, paramRegisters)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes, in PassByReferenceDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            else -> Int.MIN_VALUE
+        }
+    }
+}
+

codeGeneration/src/prog8/codegen/target/c128/C128MachineDefinition.kt

@@ -0,0 +1,68 @@
+package prog8.codegen.target.c128
+
+import prog8.ast.base.DataType
+import prog8.codegen.target.c64.normal6502instructions
+import prog8.codegen.target.cbm.Mflpt5
+import prog8.codegen.target.cbm.viceMonListName
+import prog8.compilerinterface.*
+import java.io.IOException
+import java.nio.file.Path
+
+
+class C128MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU6502
+
+    override val FLOAT_MAX_POSITIVE = Mflpt5.FLOAT_MAX_POSITIVE
+    override val FLOAT_MAX_NEGATIVE = Mflpt5.FLOAT_MAX_NEGATIVE
+    override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
+    override val BASIC_LOAD_ADDRESS = 0x1c01u
+    override val RAW_LOAD_ADDRESS = 0x1300u
+
+    // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
+    override val ESTACK_LO = 0x1a00u     //  $1a00-$1aff inclusive
+    override val ESTACK_HI = 0x1b00u     //  $1b00-$1bff 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 c128 target only supports the main emulator (Vice).")
+            return
+        }
+
+        for(emulator in listOf("x128")) {
+            println("\nStarting C-128 emulator $emulator...")
+            val viceMonlist = viceMonListName(programNameWithPath.toString())
+            val cmdline = listOf(emulator, "-silent", "-moncommands", viceMonlist,
+                    "-autostartprgmode", "1", "-autostart-warp", "-autostart", "${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 isIOAddress(address: UInt): Boolean = address==0u || address==1u || address in 0xd000u..0xdfffu
+    override fun getPreallocatedZeropageVars(): Map<String, Pair<UInt, DataType>> = emptyMap()
+
+    override fun initializeZeropage(compilerOptions: CompilationOptions) {
+        zeropage = C128Zeropage(compilerOptions)
+    }
+
+    override val opcodeNames = normal6502instructions
+}

codeGeneration/src/prog8/codegen/target/c128/C128Zeropage.kt

@@ -0,0 +1,42 @@
+package prog8.codegen.target.c128
+
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.InternalCompilerException
+import prog8.compilerinterface.Zeropage
+import prog8.compilerinterface.ZeropageType
+
+class C128Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0x9bu      // temp storage for a single byte
+    override val SCRATCH_REG = 0x9cu     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0xfbu      // temp storage 1 for a word  $fb+$fc
+    override val SCRATCH_W2 = 0xfdu      // temp storage 2 for a word  $fd+$fe
+
+
+    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'")
+
+        when (options.zeropage) {
+            ZeropageType.FULL -> {
+                // TODO all c128 usable zero page locations, except the ones used by the system's IRQ routine
+                free.addAll(0x0au..0xffu)       // TODO c128 what about $02-$09?
+                // TODO c128  free.removeAll(setOf(0xa0u, 0xa1u, 0xa2u, 0x91u, 0xc0u, 0xc5u, 0xcbu, 0xf5u, 0xf6u))        // these are updated by IRQ
+            }
+            ZeropageType.KERNALSAFE,
+            ZeropageType.FLOATSAFE,
+            ZeropageType.BASICSAFE -> {
+                free.clear()   // TODO c128 usable zero page addresses
+            }
+            ZeropageType.DONTUSE -> {
+                free.clear()  // don't use zeropage at all
+            }
+        }
+
+        removeReservedFromFreePool()
+    }
+}

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

@@ -0,0 +1,80 @@
+package prog8.codegen.target.c64
+
+import prog8.ast.base.DataType
+import prog8.codegen.target.cbm.Mflpt5
+import prog8.codegen.target.cbm.viceMonListName
+import prog8.compilerinterface.*
+import java.io.IOException
+import java.nio.file.Path
+
+
+class C64MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU6502
+
+    override val FLOAT_MAX_POSITIVE = Mflpt5.FLOAT_MAX_POSITIVE
+    override val FLOAT_MAX_NEGATIVE = Mflpt5.FLOAT_MAX_NEGATIVE
+    override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
+    override val BASIC_LOAD_ADDRESS = 0x0801u
+    override val RAW_LOAD_ADDRESS = 0xc000u
+
+    // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
+    override val ESTACK_LO = 0xce00u     //  $ce00-$ceff inclusive
+    override val ESTACK_HI = 0xcf00u     //  $ce00-$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 $emulator...")
+            val viceMonlist = viceMonListName(programNameWithPath.toString())
+            val cmdline = listOf(emulator, "-silent", "-moncommands", viceMonlist,
+                    "-autostartprgmode", "1", "-autostart-warp", "-autostart", "${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 isIOAddress(address: UInt): Boolean = address==0u || address==1u || address in 0xd000u..0xdfffu
+    override fun getPreallocatedZeropageVars(): Map<String, Pair<UInt, DataType>> = emptyMap()
+
+    override fun initializeZeropage(compilerOptions: CompilationOptions) {
+        zeropage = C64Zeropage(compilerOptions)
+    }
+
+    override val opcodeNames = normal6502instructions
+}
+
+
+// 6502 opcodes (including aliases and illegal opcodes), these cannot be used as variable or label names
+internal val normal6502instructions = 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")

codeGeneration/src/prog8/codegen/target/c64/C64Zeropage.kt

@@ -0,0 +1,73 @@
+package prog8.codegen.target.c64
+
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.InternalCompilerException
+import prog8.compilerinterface.Zeropage
+import prog8.compilerinterface.ZeropageType
+
+class C64Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0x02u      // temp storage for a single byte
+    override val SCRATCH_REG = 0x03u     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0xfbu      // temp storage 1 for a word  $fb+$fc
+    override val SCRATCH_W2 = 0xfdu      // temp storage 2 for a word  $fd+$fe
+
+
+    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(0x02u..0xffu)
+            free.removeAll(setOf(SCRATCH_B1, SCRATCH_REG, SCRATCH_W1, SCRATCH_W1+1u, SCRATCH_W2, SCRATCH_W2+1u))
+            free.removeAll(setOf(0xa0u, 0xa1u, 0xa2u, 0x91u, 0xc0u, 0xc5u, 0xcbu, 0xf5u, 0xf6u))        // 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'
+                ).map{it.toUInt()})
+            }
+
+            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
+                ).map{it.toUInt()})
+            }
+
+            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).map{it.toUInt()})
+            } else {
+                // don't use the zeropage at all
+                free.clear()
+            }
+        }
+
+        removeReservedFromFreePool()
+    }
+}

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

@@ -1,34 +1,55 @@
-package prog8.compiler.target.c64
+package prog8.codegen.target.cbm
 
-import prog8.compiler.CompilationOptions
-import prog8.compiler.OutputType
-import prog8.compiler.target.CompilationTarget
-import prog8.compiler.target.IAssemblyProgram
-import prog8.compiler.target.generatedLabelPrefix
+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.system.exitProcess
+import kotlin.io.path.Path
+import kotlin.io.path.isRegularFile
+
+
+internal fun viceMonListName(baseFilename: String) = "$baseFilename.vice-mon-list"
+
+class AssemblyProgram(
+        override val valid: Boolean,
+        override val name: String,
+        outputDir: Path,
+        private val compTarget: String) : IAssemblyProgram {
 
-class AssemblyProgram(override val name: String, outputDir: Path) : IAssemblyProgram {
     private val assemblyFile = outputDir.resolve("$name.asm")
     private val prgFile = outputDir.resolve("$name.prg")
     private val binFile = outputDir.resolve("$name.bin")
-    private val viceMonListFile = outputDir.resolve("$name.vice-mon-list")
+    private val viceMonListFile = outputDir.resolve(viceMonListName(name))
+    private val listFile = outputDir.resolve("$name.list")
 
-    override fun assemble(options: CompilationOptions) {
+    override fun assemble(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")
+                "--dump-labels", "--vice-labels", "--labels=$viceMonListFile", "--no-monitor"
+            )
+
+        if(options.asmQuiet)
+            command.add("--quiet")
+
+        if(options.asmListfile)
+            command.add("--list=$listFile")
 
         val outFile = when (options.output) {
             OutputType.PRG -> {
                 command.add("--cbm-prg")
-                println("\nCreating prg for target ${CompilationTarget.instance.name}.")
+                println("\nCreating prg for target $compTarget.")
                 prgFile
             }
             OutputType.RAW -> {
                 command.add("--nostart")
-                println("\nCreating raw binary for target ${CompilationTarget.instance.name}.")
+                println("\nCreating raw binary for target $compTarget.")
                 binFile
             }
         }
@@ -36,13 +57,11 @@ class AssemblyProgram(override val name: String, outputDir: Path) : IAssemblyPro
 
         val proc = ProcessBuilder(command).inheritIO().start()
         val result = proc.waitFor()
-        if (result != 0) {
-            System.err.println("assembler failed with returncode $result")
-            exitProcess(result)
+        if (result == 0) {
+            removeGeneratedLabelsFromMonlist()
+            generateBreakpointList()
         }
-
-        removeGeneratedLabelsFromMonlist()
-        generateBreakpointList()
+        return result
     }
 
     private fun removeGeneratedLabelsFromMonlist() {
@@ -72,3 +91,18 @@ class AssemblyProgram(override val name: String, outputDir: Path) : IAssemblyPro
         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/codegen/target/cbm/Mflpt5.kt

@@ -0,0 +1,78 @@
+package prog8.codegen.target.cbm
+
+import prog8.compilerinterface.IMachineFloat
+import prog8.compilerinterface.InternalCompilerException
+import kotlin.math.absoluteValue
+import kotlin.math.pow
+
+
+data class Mflpt5(val b0: UByte, val b1: UByte, val b2: UByte, val b3: UByte, val b4: UByte): IMachineFloat {
+
+    companion object {
+        const val FLOAT_MAX_POSITIVE = 1.7014118345e+38         // bytes: 255,127,255,255,255
+        const val FLOAT_MAX_NEGATIVE = -1.7014118345e+38        // bytes: 255,255,255,255,255
+        const val FLOAT_MEM_SIZE = 5
+
+        val zero = Mflpt5(0u, 0u, 0u, 0u, 0u)
+        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: $this")
+                exponent == 0 -> zero
+                else -> {
+                    val mantLong = mantissa.toLong()
+                    Mflpt5(
+                        exponent.toUByte(),
+                        (mantLong.and(0x7f000000L) ushr 24).or(sign).toUByte(),
+                        (mantLong.and(0x00ff0000L) ushr 16).toUByte(),
+                        (mantLong.and(0x0000ff00L) ushr 8).toUByte(),
+                        (mantLong.and(0x000000ffL)).toUByte()
+                    )
+                }
+            }
+        }
+    }
+
+    override fun toDouble(): Double {
+        if (this == zero) return 0.0
+        val exp = b0.toInt() - 128
+        val sign = (b1.toInt() and 0x80) > 0
+        val number = 0x80000000L.or(b1.toLong() shl 24).or(b2.toLong() shl 16).or(b3.toLong() shl 8).or(b4.toLong())
+        val result = number.toDouble() * (2.0).pow(exp) / 0x100000000
+        return if (sign) -result else result
+    }
+
+    override fun makeFloatFillAsm(): String {
+        val b0 = "$" + 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/codegen/target/cpu6502/codegen/AsmOptimizer.kt

@@ -0,0 +1,446 @@
+package prog8.codegen.target.cpu6502.codegen
+
+import prog8.ast.Program
+import prog8.ast.base.VarDeclType
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.VarDecl
+import prog8.compilerinterface.IMachineDefinition
+
+
+// note: see https://wiki.nesdev.org/w/index.php/6502_assembly_optimisations
+
+
+fun optimizeAssembly(lines: MutableList<String>, machine: IMachineDefinition, program: Program): 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, machine, program)
+    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, machine, program)
+    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>>>, machine: IMachineDefinition, program: Program): List<Modification> {
+
+    // Optimize sequential assignments of the same value to various targets (bytes, words, floats)
+    // the float one is the one that requires 2*7=14 lines of code to check...
+    // The better place to do this is in the Compiler instead and never create these types of assembly, but hey
+
+    val mods = mutableListOf<Modification>()
+    for (lines in linesByFourteen) {
+        val first = lines[0].value.trimStart()
+        val second = lines[1].value.trimStart()
+        val third = lines[2].value.trimStart()
+        val fourth = lines[3].value.trimStart()
+        val fifth = lines[4].value.trimStart()
+        val sixth = lines[5].value.trimStart()
+        val seventh = lines[6].value.trimStart()
+        val eighth = lines[7].value.trimStart()
+
+        if(first.startsWith("lda") && second.startsWith("ldy") && third.startsWith("sta") && fourth.startsWith("sty") &&
+                fifth.startsWith("lda") && sixth.startsWith("ldy") && seventh.startsWith("sta") && eighth.startsWith("sty")) {
+            val firstvalue = first.substring(4)
+            val secondvalue = second.substring(4)
+            val thirdvalue = fifth.substring(4)
+            val fourthvalue = sixth.substring(4)
+            if(firstvalue==thirdvalue && secondvalue==fourthvalue) {
+                // lda/ldy   sta/sty   twice the same word -->  remove second lda/ldy pair (fifth and sixth lines)
+                val address1 = getAddressArg(first, program)
+                val address2 = getAddressArg(second, program)
+                if(address1==null || address2==null || (!machine.isIOAddress(address1) && !machine.isIOAddress(address2))) {
+                    mods.add(Modification(lines[4].index, true, null))
+                    mods.add(Modification(lines[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 same-value / sta ?  -> remove second lda (third line)
+                val address = getAddressArg(first, program)
+                if(address==null || !machine.isIOAddress(address))
+                    mods.add(Modification(lines[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 = lines[8].value.trimStart()
+            val tenth = lines[9].value.trimStart()
+            val eleventh = lines[10].value.trimStart()
+            val twelveth = lines[11].value.trimStart()
+            val thirteenth = lines[12].value.trimStart()
+            val fourteenth = lines[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(lines[7].index, true, null))
+                    mods.add(Modification(lines[8].index, true, null))
+                    mods.add(Modification(lines[9].index, true, null))
+                    mods.add(Modification(lines[10].index, true, null))
+                }
+            }
+        }
+
+        var overlappingMods = false
+        /*
+        sta  prog8_lib.retval_intermX      ; remove
+        sty  prog8_lib.retval_intermY      ; remove
+        lda  prog8_lib.retval_intermX      ; remove
+        ldy  prog8_lib.retval_intermY      ; remove
+        sta  A1
+        sty  A2
+         */
+        if(first.startsWith("st") && second.startsWith("st")
+            && third.startsWith("ld") && fourth.startsWith("ld")
+            && fifth.startsWith("st") && sixth.startsWith("st")) {
+            val reg1 = first[2]
+            val reg2 = second[2]
+            val reg3 = third[2]
+            val reg4 = fourth[2]
+            val reg5 = fifth[2]
+            val reg6 = sixth[2]
+            if (reg1 == reg3 && reg1 == reg5 && reg2 == reg4 && reg2 == reg6) {
+                val firstvalue = first.substring(4)
+                val secondvalue = second.substring(4)
+                val thirdvalue = third.substring(4)
+                val fourthvalue = fourth.substring(4)
+                if(firstvalue.contains("prog8_lib.retval_interm") && secondvalue.contains("prog8_lib.retval_interm")
+                    && firstvalue==thirdvalue && secondvalue==fourthvalue) {
+                    mods.add(Modification(lines[0].index, true, null))
+                    mods.add(Modification(lines[1].index, true, null))
+                    mods.add(Modification(lines[2].index, true, null))
+                    mods.add(Modification(lines[3].index, true, null))
+                    overlappingMods = true
+                }
+            }
+        }
+
+        /*
+        sta  A1
+        sty  A2
+        lda  A1     ; can be removed
+        ldy  A2     ; can be removed if not followed by a branch instuction
+         */
+        if(!overlappingMods && first.startsWith("st") && second.startsWith("st")
+            && third.startsWith("ld") && fourth.startsWith("ld")) {
+            val reg1 = first[2]
+            val reg2 = second[2]
+            val reg3 = third[2]
+            val reg4 = fourth[2]
+            if(reg1==reg3 && reg2==reg4) {
+                val firstvalue = first.substring(4)
+                val secondvalue = second.substring(4)
+                val thirdvalue = third.substring(4)
+                val fourthvalue = fourth.substring(4)
+                if(firstvalue==thirdvalue && secondvalue == fourthvalue) {
+                    val address = getAddressArg(first, program)
+                    if(address==null || !machine.isIOAddress(address)) {
+                        overlappingMods = true
+                        mods.add(Modification(lines[2].index, true, null))
+                        if (!fifth.startsWith('b'))
+                            mods.add(Modification(lines[3].index, true, null))
+                    }
+                }
+            }
+        }
+        /*
+        sta  A1
+        sty  A2
+        lda  A1     ; can be removed if not followed by a branch instruction
+         */
+        if(!overlappingMods && first.startsWith("st") && second.startsWith("st")
+            && third.startsWith("ld") && !fourth.startsWith("b")) {
+            val reg1 = first[2]
+            val reg3 = third[2]
+            if(reg1==reg3) {
+                val firstvalue = first.substring(4)
+                val thirdvalue = third.substring(4)
+                if(firstvalue==thirdvalue) {
+                    val address = getAddressArg(first, program)
+                    if(address==null || !machine.isIOAddress(address)) {
+                        overlappingMods = true
+                        mods.add(Modification(lines[2].index, true, null))
+                    }
+                }
+            }
+        }
+
+        /*
+        sta  A1
+        ldy  A1     ; make tay
+        sta  A1     ; remove
+         */
+        if(!overlappingMods && first.startsWith("sta") && second.startsWith("ld")
+            && third.startsWith("sta") && second.length>4) {
+            val firstvalue = first.substring(4)
+            val secondvalue = second.substring(4)
+            val thirdvalue = third.substring(4)
+            if(firstvalue==secondvalue && firstvalue==thirdvalue) {
+                val address = getAddressArg(first, program)
+                if(address==null || !machine.isIOAddress(address)) {
+                    overlappingMods = true
+                    val reg2 = second[2]
+                    mods.add(Modification(lines[1].index, false, "  ta$reg2"))
+                    mods.add(Modification(lines[2].index, true, null))
+                }
+            }
+        }
+
+        /*
+        sta  A
+        sta  A
+         */
+        if(!overlappingMods && first.startsWith("st") && second.startsWith("st")) {
+            if(first[2]==second[2]) {
+                val firstvalue = first.substring(4)
+                val secondvalue = second.substring(4)
+                if(firstvalue==secondvalue) {
+                    val address = getAddressArg(first, program)
+                    if(address==null || !machine.isIOAddress(address)) {
+                        overlappingMods = true
+                        mods.add(Modification(lines[1].index, true, null))
+                    }
+                }
+            }
+        }
+    }
+
+    return mods
+}
+
+private fun optimizeStoreLoadSame(linesByFour: List<List<IndexedValue<String>>>, machine: IMachineDefinition, program: Program): List<Modification> {
+    // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can OFTEN be eliminated
+    val mods = mutableListOf<Modification>()
+    for (lines in linesByFour) {
+        val first = lines[1].value.trimStart()
+        val second = lines[2].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 = lines[3].value.trimStart()
+            val attemptRemove =
+                if(third.startsWith("b")) {
+                    // a branch instruction follows, we can only remove the load instruction if
+                    // another load instruction of the same register precedes the store instruction
+                    // (otherwise wrong cpu flags are used)
+                    val loadinstruction = second.substring(0, 3)
+                    lines[0].value.trimStart().startsWith(loadinstruction)
+                }
+                else {
+                    // no branch instruction follows, we can remove the load instruction
+                    val address = getAddressArg(lines[2].value, program)
+                    address==null || !machine.isIOAddress(address)
+                }
+
+            if(attemptRemove) {
+                val firstLoc = first.substring(4).trimStart()
+                val secondLoc = second.substring(4).trimStart()
+                if (firstLoc == secondLoc)
+                    mods.add(Modification(lines[2].index, true, null))
+            }
+        }
+        else if(first=="pha" && second=="pla" ||
+            first=="phx" && second=="plx" ||
+            first=="phy" && second=="ply" ||
+            first=="php" && second=="plp") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, true, null))
+        } else if(first=="pha" && second=="plx") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  tax"))
+        } else if(first=="pha" && second=="ply") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  tay"))
+        } else if(first=="phx" && second=="pla") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  txa"))
+        } else if(first=="phx" && second=="ply") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  txy"))
+        } else if(first=="phy" && second=="pla") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  tya"))
+        } else if(first=="phy" && second=="plx") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, false, "  tyx"))
+        }
+    }
+    return mods
+}
+
+private val identifierRegex = Regex("""^([a-zA-Z_$][a-zA-Z\d_\.$]*)""")
+
+private fun getAddressArg(line: String, program: Program): UInt? {
+    val loadArg = line.trimStart().substring(3).trim()
+    return when {
+        loadArg.startsWith('$') -> loadArg.substring(1).toUIntOrNull(16)
+        loadArg.startsWith('%') -> loadArg.substring(1).toUIntOrNull(2)
+        loadArg.startsWith('#') -> null
+        loadArg.startsWith('(') -> null
+        loadArg[0].isLetter() -> {
+            val identMatch = identifierRegex.find(loadArg)
+            if(identMatch!=null) {
+                val identifier = identMatch.value
+                val decl = program.toplevelModule.lookup(identifier.split(".")) as? VarDecl
+                if(decl!=null) {
+                    when(decl.type){
+                        VarDeclType.VAR -> null
+                        VarDeclType.CONST,
+                        VarDeclType.MEMORY -> (decl.value as NumericLiteralValue).number.toUInt()
+                    }
+                }
+                else null
+            } else null
+        }
+        else -> loadArg.substring(1).toUIntOrNull()
+    }
+}
+
+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 (lines in linesByFour) {
+        val first = lines[0].value
+        val second = lines[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)
+                || (" ina" in first || "\tina" in first) && (" dea" in second || "\tdea" in second)
+                || (" inc  a" in first || "\tinc  a" in first) && (" dec  a" in second || "\tdec  a" 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)
+                || (" dea" in first || "\tdea" in first) && (" ina" in second || "\tina" in second)
+                || (" dec  a" in first || "\tdec  a" in first) && (" inc  a" in second || "\tinc  a" in second)) {
+            mods.add(Modification(lines[0].index, true, null))
+            mods.add(Modification(lines[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 (lines in linesByFour) {
+        val first = lines[0].value
+        val second = lines[1].value
+        if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {
+            mods += Modification(lines[0].index, false, lines[0].value.replace("jsr", "jmp"))
+            mods += Modification(lines[1].index, true, null)
+        }
+    }
+    return mods
+}

codeGeneration/src/prog8/codegen/target/cpu6502/codegen/AsmsubHelpers.kt

@@ -0,0 +1,59 @@
+package prog8.codegen.target.cpu6502.codegen
+
+import prog8.ast.base.Cx16VirtualRegisters
+import prog8.ast.base.RegisterOrPair
+import prog8.ast.expressions.*
+import prog8.ast.statements.RegisterOrStatusflag
+import prog8.ast.statements.Subroutine
+
+
+internal fun asmsub6502ArgsEvalOrder(sub: Subroutine): List<Int> {
+    val order = mutableListOf<Int>()
+    // order is:
+    //  1) cx16 virtual word registers,
+    //  2) paired CPU registers,
+    //  3) single CPU registers (X last), except A,
+    //  4) CPU Carry status flag
+    //  5) the A register itself last   (so everything before it can use the accumulator without having to save its value)
+    val args = sub.parameters.zip(sub.asmParameterRegisters).withIndex()
+    val (cx16regs, args2) = args.partition { it.value.second.registerOrPair in Cx16VirtualRegisters }
+    val pairedRegisters = arrayOf(RegisterOrPair.AX, RegisterOrPair.AY, RegisterOrPair.XY)
+    val (pairedRegs , args3) = args2.partition { it.value.second.registerOrPair in pairedRegisters }
+    val (regsWithoutA, args4) = args3.partition { it.value.second.registerOrPair != RegisterOrPair.A }
+    val (regA, rest) = args4.partition { it.value.second.registerOrPair != null }
+
+    cx16regs.forEach { order += it.index }
+    pairedRegs.forEach { order += it.index }
+    regsWithoutA.forEach {
+        if(it.value.second.registerOrPair != RegisterOrPair.X)
+            order += it.index
+    }
+    regsWithoutA.firstOrNull { it.value.second.registerOrPair==RegisterOrPair.X } ?.let { order += it.index}
+    rest.forEach { order += it.index }
+    regA.forEach { order += it.index }
+    require(order.size==sub.parameters.size)
+    return order
+}
+
+internal fun asmsub6502ArgsHaveRegisterClobberRisk(args: List<Expression>,
+                                                   paramRegisters: List<RegisterOrStatusflag>): Boolean {
+    fun isClobberRisk(expr: Expression): Boolean {
+        when (expr) {
+            is ArrayIndexedExpression -> {
+                return paramRegisters.any {
+                    it.registerOrPair in listOf(RegisterOrPair.Y, RegisterOrPair.AY, RegisterOrPair.XY)
+                }
+            }
+            is FunctionCallExpr -> {
+                if (expr.target.nameInSource == listOf("lsb") || expr.target.nameInSource == listOf("msb"))
+                    return isClobberRisk(expr.args[0])
+                if (expr.target.nameInSource == listOf("mkword"))
+                    return isClobberRisk(expr.args[0]) && isClobberRisk(expr.args[1])
+                return !expr.isSimple
+            }
+            else -> return !expr.isSimple
+        }
+    }
+
+    return args.size>1 && args.any { isClobberRisk(it) }
+}

codeGeneration/src/prog8/codegen/target/cpu6502/codegen/ExpressionsAsmGen.kt

@@ -0,0 +1,792 @@
+package prog8.codegen.target.cpu6502.codegen
+
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.BuiltinFunctionPlaceholder
+import prog8.ast.statements.Subroutine
+import prog8.ast.toHex
+import prog8.codegen.target.AssemblyError
+import prog8.compilerinterface.BuiltinFunctions
+import prog8.compilerinterface.CpuType
+import kotlin.math.absoluteValue
+
+internal class ExpressionsAsmGen(private val program: Program, private val asmgen: AsmGen) {
+
+    @Deprecated("avoid calling this as it generates slow evalstack based code")
+    internal fun translateExpression(expression:Expression) {
+        if (this.asmgen.options.slowCodegenWarnings) {
+            asmgen.errors.warn("slow stack evaluation used for expression $expression", expression.position)
+        }
+        translateExpressionInternal(expression)
+    }
+
+
+    // the rest of the methods are all PRIVATE
+
+
+    private fun translateExpressionInternal(expression: Expression) {
+
+        when(expression) {
+            is PrefixExpression -> translateExpression(expression)
+            is BinaryExpression -> translateExpression(expression)
+            is ArrayIndexedExpression -> translateExpression(expression)
+            is TypecastExpression -> translateExpression(expression)
+            is AddressOf -> translateExpression(expression)
+            is DirectMemoryRead -> asmgen.translateDirectMemReadExpressionToRegAorStack(expression, true)
+            is NumericLiteralValue -> translateExpression(expression)
+            is IdentifierReference -> translateExpression(expression)
+            is FunctionCallExpr -> translateFunctionCallResultOntoStack(expression)
+            is ContainmentCheck -> throw AssemblyError("containment check as complex expression value is not supported")
+            is ArrayLiteralValue, is StringLiteralValue -> throw AssemblyError("no asm gen for string/array literal value assignment - should have been replaced by a variable")
+            is RangeExpr -> throw AssemblyError("range expression should have been changed into array values")
+            is CharLiteral -> throw AssemblyError("charliteral should have been replaced by ubyte using certain encoding")
+            else -> TODO("missing expression asmgen for $expression")
+        }
+    }
+
+    private fun translateFunctionCallResultOntoStack(call: FunctionCallExpr) {
+        // only for use in nested expression evaluation
+
+        val sub = call.target.targetStatement(program)
+        if(sub is BuiltinFunctionPlaceholder) {
+            val builtinFunc = BuiltinFunctions.getValue(sub.name)
+            asmgen.translateBuiltinFunctionCallExpression(call, builtinFunc, true, null)
+        } else {
+            sub as Subroutine
+            asmgen.saveXbeforeCall(call)
+            asmgen.translateFunctionCall(call, true)
+            if(sub.regXasResult()) {
+                // store the return value in X somewhere that we can acces again below
+                asmgen.out("  stx  P8ZP_SCRATCH_REG")
+            }
+            asmgen.restoreXafterCall(call)
+
+            val returns = sub.returntypes.zip(sub.asmReturnvaluesRegisters)
+            for ((_, reg) in returns) {
+                // result value is in cpu or status registers, put it on the stack instead (as we're evaluating an expression tree)
+                if (reg.registerOrPair != null) {
+                    when (reg.registerOrPair!!) {
+                        RegisterOrPair.A -> asmgen.out("  sta  P8ESTACK_LO,x |  dex")
+                        RegisterOrPair.Y -> asmgen.out("  tya |  sta  P8ESTACK_LO,x |  dex")
+                        RegisterOrPair.AY -> asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
+                        RegisterOrPair.X -> asmgen.out("  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_LO,x |  dex")
+                        RegisterOrPair.AX -> asmgen.out("  sta  P8ESTACK_LO,x |  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_HI,x |  dex")
+                        RegisterOrPair.XY -> asmgen.out("  tya |  sta  P8ESTACK_HI,x |  lda  P8ZP_SCRATCH_REG |  sta  P8ESTACK_LO,x |  dex")
+                        RegisterOrPair.FAC1 -> asmgen.out("  jsr  floats.push_fac1")
+                        RegisterOrPair.FAC2 -> asmgen.out("  jsr  floats.push_fac2")
+                        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 -> {
+                            asmgen.out(
+                                """
+                                lda  cx16.${reg.registerOrPair.toString().lowercase()}
+                                sta  P8ESTACK_LO,x
+                                lda  cx16.${reg.registerOrPair.toString().lowercase()}+1
+                                sta  P8ESTACK_HI,x
+                                dex
+                            """)
+                        }
+                    }
+                } else when(reg.statusflag) {
+                    Statusflag.Pc -> {
+                        asmgen.out("""
+                            lda  #0
+                            rol  a
+                            sta  P8ESTACK_LO,x
+                            dex""")
+                    }
+                    Statusflag.Pz -> {
+                        asmgen.out("""
+                            beq  +
+                            lda  #0
+                            beq  ++
++                           lda  #1
++                           sta  P8ESTACK_LO,x
+                            dex""")
+                    }
+                    Statusflag.Pv -> {
+                        asmgen.out("""
+                            bvs  +
+                            lda  #0
+                            beq  ++
++                           lda  #1
++                           sta  P8ESTACK_LO,x
+                            dex""")
+                    }
+                    Statusflag.Pn -> {
+                        asmgen.out("""
+                            bmi  +
+                            lda  #0
+                            beq  ++
++                           lda  #1
++                           sta  P8ESTACK_LO,x
+                            dex""")
+                    }
+                    null -> {}
+                }
+            }
+        }
+    }
+
+    private fun translateExpression(typecast: TypecastExpression) {
+        translateExpressionInternal(typecast.expression)
+        when(typecast.expression.inferType(program).getOr(DataType.UNDEFINED)) {
+            DataType.UBYTE -> {
+                when(typecast.type) {
+                    DataType.UBYTE, DataType.BYTE -> {}
+                    DataType.UWORD, DataType.WORD -> {
+                        if(asmgen.isTargetCpu(CpuType.CPU65c02))
+                            asmgen.out("  stz  P8ESTACK_HI+1,x")
+                        else
+                            asmgen.out("  lda  #0  |  sta  P8ESTACK_HI+1,x")
+                    }
+                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_ub2float")
+                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            DataType.BYTE -> {
+                when(typecast.type) {
+                    DataType.UBYTE, DataType.BYTE -> {}
+                    DataType.UWORD, DataType.WORD -> asmgen.signExtendStackLsb(DataType.BYTE)
+                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_b2float")
+                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            DataType.UWORD -> {
+                when(typecast.type) {
+                    DataType.BYTE, DataType.UBYTE -> {}
+                    DataType.WORD, DataType.UWORD -> {}
+                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_uw2float")
+                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            DataType.WORD -> {
+                when(typecast.type) {
+                    DataType.BYTE, DataType.UBYTE -> {}
+                    DataType.WORD, DataType.UWORD -> {}
+                    DataType.FLOAT -> asmgen.out(" jsr  floats.stack_w2float")
+                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            DataType.FLOAT -> {
+                when(typecast.type) {
+                    DataType.UBYTE -> asmgen.out(" jsr  floats.stack_float2uw")
+                    DataType.BYTE -> asmgen.out(" jsr  floats.stack_float2w")
+                    DataType.UWORD -> asmgen.out(" jsr  floats.stack_float2uw")
+                    DataType.WORD -> asmgen.out(" jsr  floats.stack_float2w")
+                    DataType.FLOAT -> {}
+                    in PassByReferenceDatatypes -> throw AssemblyError("cannot cast to a pass-by-reference datatype")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            DataType.STR -> {
+                if (typecast.type != DataType.UWORD && typecast.type == DataType.STR)
+                    throw AssemblyError("cannot typecast a string into another incompatitble type")
+            }
+            in PassByReferenceDatatypes -> throw AssemblyError("cannot cast pass-by-reference value into another type")
+            else -> throw AssemblyError("weird type")
+        }
+    }
+
+    private fun translateExpression(expr: AddressOf) {
+        val name = asmgen.asmVariableName(expr.identifier)
+        asmgen.out("  lda  #<$name |  sta  P8ESTACK_LO,x |  lda  #>$name  |  sta  P8ESTACK_HI,x  | dex")
+    }
+
+    private fun translateExpression(expr: NumericLiteralValue) {
+        when(expr.type) {
+            DataType.UBYTE, DataType.BYTE -> asmgen.out(" lda  #${expr.number.toHex()}  | sta  P8ESTACK_LO,x  | dex")
+            DataType.UWORD, DataType.WORD -> asmgen.out("""
+                lda  #<${expr.number.toHex()}
+                sta  P8ESTACK_LO,x
+                lda  #>${expr.number.toHex()}
+                sta  P8ESTACK_HI,x
+                dex
+            """)
+            DataType.FLOAT -> {
+                val floatConst = asmgen.getFloatAsmConst(expr.number)
+                asmgen.out(" lda  #<$floatConst |  ldy  #>$floatConst |  jsr  floats.push_float")
+            }
+            else -> throw AssemblyError("weird type")
+        }
+    }
+
+    private fun translateExpression(expr: IdentifierReference) {
+        val varname = asmgen.asmVariableName(expr)
+        when(expr.inferType(program).getOr(DataType.UNDEFINED)) {
+            DataType.UBYTE, DataType.BYTE -> {
+                asmgen.out("  lda  $varname  |  sta  P8ESTACK_LO,x  |  dex")
+            }
+            DataType.UWORD, DataType.WORD -> {
+                asmgen.out("  lda  $varname  |  sta  P8ESTACK_LO,x  |  lda  $varname+1 |  sta  P8ESTACK_HI,x |  dex")
+            }
+            DataType.FLOAT -> {
+                asmgen.out(" lda  #<$varname |  ldy  #>$varname|  jsr  floats.push_float")
+            }
+            in IterableDatatypes -> {
+                asmgen.out("  lda  #<$varname  |  sta  P8ESTACK_LO,x  |  lda  #>$varname |  sta  P8ESTACK_HI,x |  dex")
+            }
+            else -> throw AssemblyError("stack push weird variable type $expr")
+        }
+    }
+
+    private fun translateExpression(expr: BinaryExpression) {
+        // Uses evalstack to evaluate the given expression.
+        // TODO we're slowly reducing the number of places where this is called and instead replace that by more efficient assignment-form code (using temp var or register for instance).
+        val leftIDt = expr.left.inferType(program)
+        val rightIDt = expr.right.inferType(program)
+        if(!leftIDt.isKnown || !rightIDt.isKnown)
+            throw AssemblyError("can't infer type of both expression operands")
+
+        val leftDt = leftIDt.getOrElse { throw AssemblyError("unknown dt") }
+        val rightDt = rightIDt.getOrElse { throw AssemblyError("unknown dt") }
+        // see if we can apply some optimized routines
+        when(expr.operator) {
+            "+" -> {
+                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                    val leftVal = expr.left.constValue(program)?.number?.toInt()
+                    val rightVal = expr.right.constValue(program)?.number?.toInt()
+                    if (leftVal!=null && leftVal in -4..4) {
+                        translateExpressionInternal(expr.right)
+                        if(rightDt in ByteDatatypes) {
+                            val incdec = if(leftVal<0) "dec" else "inc"
+                            repeat(leftVal.absoluteValue) {
+                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
+                            }
+                        } else {
+                            // word
+                            if(leftVal<0) {
+                                repeat(leftVal.absoluteValue) {
+                                    asmgen.out("""
+                                        lda  P8ESTACK_LO+1,x
+                                        bne  +
+                                        dec  P8ESTACK_HI+1,x
++                                       dec  P8ESTACK_LO+1,x""")
+                                }
+                            } else {
+                                repeat(leftVal) {
+                                    asmgen.out("""
+                                        inc  P8ESTACK_LO+1,x
+                                        bne  +
+                                        inc  P8ESTACK_HI+1,x
++""")
+                                }
+                            }
+                        }
+                        return
+                    }
+                    else if (rightVal!=null && rightVal in -4..4)
+                    {
+                        translateExpressionInternal(expr.left)
+                        if(leftDt in ByteDatatypes) {
+                            val incdec = if(rightVal<0) "dec" else "inc"
+                            repeat(rightVal.absoluteValue) {
+                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
+                            }
+                        } else {
+                            // word
+                            if(rightVal<0) {
+                                repeat(rightVal.absoluteValue) {
+                                    asmgen.out("""
+                                        lda  P8ESTACK_LO+1,x
+                                        bne  +
+                                        dec  P8ESTACK_HI+1,x
++                                       dec  P8ESTACK_LO+1,x""")
+                                }
+                            } else {
+                                repeat(rightVal) {
+                                    asmgen.out("""
+                                        inc  P8ESTACK_LO+1,x
+                                        bne  +
+                                        inc  P8ESTACK_HI+1,x
++""")
+                                }
+                            }
+                        }
+                        return
+                    }
+                }
+            }
+            "-" -> {
+                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                    val rightVal = expr.right.constValue(program)?.number?.toInt()
+                    if (rightVal!=null && rightVal in -4..4)
+                    {
+                        translateExpressionInternal(expr.left)
+                        if(leftDt in ByteDatatypes) {
+                            val incdec = if(rightVal<0) "inc" else "dec"
+                            repeat(rightVal.absoluteValue) {
+                                asmgen.out("  $incdec  P8ESTACK_LO+1,x")
+                            }
+                        } else {
+                            // word
+                            if(rightVal>0) {
+                                repeat(rightVal.absoluteValue) {
+                                    asmgen.out("""
+                                        lda  P8ESTACK_LO+1,x
+                                        bne  +
+                                        dec  P8ESTACK_HI+1,x
++                                       dec  P8ESTACK_LO+1,x""")
+                                }
+                            } else {
+                                repeat(rightVal) {
+                                    asmgen.out("""
+                                        inc  P8ESTACK_LO+1,x
+                                        bne  +
+                                        inc  P8ESTACK_HI+1,x
++""")
+                                }
+                            }
+                        }
+                        return
+                    }
+                }
+            }
+            ">>" -> {
+                val amount = expr.right.constValue(program)?.number?.toInt()
+                if(amount!=null) {
+                    translateExpressionInternal(expr.left)
+                    when (leftDt) {
+                        DataType.UBYTE -> {
+                            if (amount <= 2)
+                                repeat(amount) { asmgen.out(" lsr  P8ESTACK_LO+1,x") }
+                            else {
+                                asmgen.out(" lda  P8ESTACK_LO+1,x")
+                                repeat(amount) { asmgen.out(" lsr  a") }
+                                asmgen.out(" sta  P8ESTACK_LO+1,x")
+                            }
+                        }
+                        DataType.BYTE -> {
+                            if (amount <= 2)
+                                repeat(amount) { asmgen.out(" lda  P8ESTACK_LO+1,x |  asl  a |  ror  P8ESTACK_LO+1,x") }
+                            else {
+                                asmgen.out(" lda  P8ESTACK_LO+1,x |  sta  P8ZP_SCRATCH_B1")
+                                repeat(amount) { asmgen.out(" asl  a |  ror  P8ZP_SCRATCH_B1 |  lda  P8ZP_SCRATCH_B1") }
+                                asmgen.out(" sta  P8ESTACK_LO+1,x")
+                            }
+                        }
+                        DataType.UWORD -> {
+                            if(amount>=16) {
+                                if(asmgen.isTargetCpu(CpuType.CPU65c02))
+                                    asmgen.out("  stz  P8ESTACK_LO+1,x |  stz  P8ESTACK_HI+1,x")
+                                else
+                                    asmgen.out("  lda  #0 |  sta  P8ESTACK_LO+1,x |  sta  P8ESTACK_HI+1,x")
+                                return
+                            }
+                            var left = amount
+                            while (left >= 7) {
+                                asmgen.out(" jsr  math.shift_right_uw_7")
+                                left -= 7
+                            }
+                            if (left in 0..2)
+                                repeat(left) { asmgen.out(" lsr  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x") }
+                            else
+                                asmgen.out(" jsr  math.shift_right_uw_$left")
+                        }
+                        DataType.WORD -> {
+                            if(amount>=16) {
+                                asmgen.out("""
+                                    lda  P8ESTACK_HI+1,x
+                                    bmi  +
+                                    lda  #0
+                                    sta  P8ESTACK_LO+1,x
+                                    sta  P8ESTACK_HI+1,x
+                                    beq  ++
++                                   lda  #255
+                                    sta  P8ESTACK_LO+1,x
+                                    sta  P8ESTACK_HI+1,x
++""")
+                                return
+                            }
+                            var left = amount
+                            while (left >= 7) {
+                                asmgen.out(" jsr  math.shift_right_w_7")
+                                left -= 7
+                            }
+                            if (left in 0..2)
+                                repeat(left) { asmgen.out(" lda  P8ESTACK_HI+1,x |  asl  a  |  ror  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x") }
+                            else
+                                asmgen.out(" jsr  math.shift_right_w_$left")
+                        }
+                        else -> throw AssemblyError("weird type")
+                    }
+                    return
+                }
+            }
+            "<<" -> {
+                val amount = expr.right.constValue(program)?.number?.toInt()
+                if(amount!=null) {
+                    translateExpressionInternal(expr.left)
+                    if (leftDt in ByteDatatypes) {
+                        if (amount <= 2)
+                            repeat(amount) { asmgen.out("  asl  P8ESTACK_LO+1,x") }
+                        else {
+                            asmgen.out("  lda  P8ESTACK_LO+1,x")
+                            repeat(amount) { asmgen.out("  asl  a") }
+                            asmgen.out("  sta  P8ESTACK_LO+1,x")
+                        }
+                    } else {
+                        var left = amount
+                        while (left >= 7) {
+                            asmgen.out(" jsr  math.shift_left_w_7")
+                            left -= 7
+                        }
+                        if (left in 0..2)
+                            repeat(left) { asmgen.out("  asl  P8ESTACK_LO+1,x |  rol  P8ESTACK_HI+1,x") }
+                        else
+                            asmgen.out(" jsr  math.shift_left_w_$left")
+                    }
+                    return
+                }
+            }
+            "*" -> {
+                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                    val leftVar = expr.left as? IdentifierReference
+                    val rightVar = expr.right as? IdentifierReference
+                    if(leftVar!=null && rightVar!=null && leftVar==rightVar)
+                        return translateSquared(leftVar, leftDt)
+                }
+
+                val value = expr.right.constValue(program)
+                if(value!=null) {
+                    if(rightDt in IntegerDatatypes) {
+                        val amount = value.number.toInt()
+                        if(amount==2) {
+                            // optimize x*2 common case
+                            translateExpressionInternal(expr.left)
+                            if(leftDt in ByteDatatypes) {
+                                asmgen.out("  asl  P8ESTACK_LO+1,x")
+                            } else {
+                                asmgen.out("  asl  P8ESTACK_LO+1,x |  rol  P8ESTACK_HI+1,x")
+                            }
+                            return
+                        }
+                        when(rightDt) {
+                            DataType.UBYTE -> {
+                                if(amount in asmgen.optimizedByteMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  math.stack_mul_byte_$amount")
+                                    return
+                                }
+                            }
+                            DataType.BYTE -> {
+                                if(amount in asmgen.optimizedByteMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  math.stack_mul_byte_$amount")
+                                    return
+                                }
+                                if(amount.absoluteValue in asmgen.optimizedByteMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  prog8_lib.neg_b |  jsr  math.stack_mul_byte_${amount.absoluteValue}")
+                                    return
+                                }
+                            }
+                            DataType.UWORD -> {
+                                if(amount in asmgen.optimizedWordMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  math.stack_mul_word_$amount")
+                                    return
+                                }
+                            }
+                            DataType.WORD -> {
+                                if(amount in asmgen.optimizedWordMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  math.stack_mul_word_$amount")
+                                    return
+                                }
+                                if(amount.absoluteValue in asmgen.optimizedWordMultiplications) {
+                                    translateExpressionInternal(expr.left)
+                                    asmgen.out(" jsr  prog8_lib.neg_w |  jsr  math.stack_mul_word_${amount.absoluteValue}")
+                                    return
+                                }
+                            }
+                            else -> {}
+                        }
+                    }
+                }
+            }
+            "/" -> {
+                if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                    val rightVal = expr.right.constValue(program)?.number?.toInt()
+                    if(rightVal!=null && rightVal==2) {
+                        translateExpressionInternal(expr.left)
+                        when(leftDt) {
+                            DataType.UBYTE -> asmgen.out("  lsr  P8ESTACK_LO+1,x")
+                            DataType.BYTE -> asmgen.out("  lda  P8ESTACK_LO+1,x |  asl  a |  ror  P8ESTACK_LO+1,x")
+                            DataType.UWORD -> asmgen.out("  lsr  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x")
+                            DataType.WORD -> asmgen.out("  lda  P8ESTACK_HI+1,x |  asl  a |  ror  P8ESTACK_HI+1,x |  ror  P8ESTACK_LO+1,x")
+                            else -> throw AssemblyError("wrong dt")
+                        }
+                        return
+                    }
+                }
+            }
+        }
+
+        if((leftDt in ByteDatatypes && rightDt !in ByteDatatypes)
+                || (leftDt in WordDatatypes && rightDt !in WordDatatypes))
+            throw AssemblyError("binary operator ${expr.operator} left/right dt not identical")
+
+        if(leftDt==DataType.STR && rightDt==DataType.STR && expr.operator in ComparisonOperators) {
+            translateCompareStrings(expr.left, expr.operator, expr.right)
+        }
+        else {
+            // the general, non-optimized cases  TODO optimize more cases.... (or one day just don't use the evalstack at all anymore)
+            translateExpressionInternal(expr.left)
+            translateExpressionInternal(expr.right)
+            when (leftDt) {
+                in ByteDatatypes -> translateBinaryOperatorBytes(expr.operator, leftDt)
+                in WordDatatypes -> translateBinaryOperatorWords(expr.operator, leftDt)
+                DataType.FLOAT -> translateBinaryOperatorFloats(expr.operator)
+                else -> throw AssemblyError("non-numerical datatype")
+            }
+        }
+    }
+
+    private fun translateSquared(variable: IdentifierReference, dt: DataType) {
+        val asmVar = asmgen.asmVariableName(variable)
+        when(dt) {
+            DataType.BYTE, DataType.UBYTE -> {
+                asmgen.out("  lda  $asmVar")
+                asmgen.signExtendAYlsb(dt)
+                asmgen.out("  jsr  math.square")
+            }
+            DataType.UWORD, DataType.WORD -> {
+                asmgen.out("  lda  $asmVar |  ldy  $asmVar+1 |  jsr  math.square")
+            }
+            else -> throw AssemblyError("require integer dt for square")
+        }
+        asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
+    }
+
+    private fun translateExpression(expr: PrefixExpression) {
+        translateExpressionInternal(expr.expression)
+        val itype = expr.inferType(program)
+        val type = itype.getOrElse { throw AssemblyError("unknown dt") }
+        when(expr.operator) {
+            "+" -> {}
+            "-" -> {
+                when(type) {
+                    in ByteDatatypes -> asmgen.out("  jsr  prog8_lib.neg_b")
+                    in WordDatatypes -> asmgen.out("  jsr  prog8_lib.neg_w")
+                    DataType.FLOAT -> asmgen.out("  jsr  floats.neg_f")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            "~" -> {
+                when(type) {
+                    in ByteDatatypes ->
+                        asmgen.out("""
+                            lda  P8ESTACK_LO+1,x
+                            eor  #255
+                            sta  P8ESTACK_LO+1,x
+                            """)
+                    in WordDatatypes -> asmgen.out("  jsr  prog8_lib.inv_word")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            "not" -> {
+                when(type) {
+                    in ByteDatatypes -> asmgen.out("  jsr  prog8_lib.not_byte")
+                    in WordDatatypes -> asmgen.out("  jsr  prog8_lib.not_word")
+                    else -> throw AssemblyError("weird type")
+                }
+            }
+            else -> throw AssemblyError("invalid prefix operator ${expr.operator}")
+        }
+    }
+
+    private fun translateExpression(arrayExpr: ArrayIndexedExpression) {
+        val elementIDt = arrayExpr.inferType(program)
+        if(!elementIDt.isKnown)
+            throw AssemblyError("unknown dt")
+        val elementDt = elementIDt.getOr(DataType.UNDEFINED)
+        val arrayVarName = asmgen.asmVariableName(arrayExpr.arrayvar)
+        val constIndexNum = arrayExpr.indexer.constIndex()
+        if(constIndexNum!=null) {
+            val indexValue = constIndexNum * program.memsizer.memorySize(elementDt)
+            when(elementDt) {
+                in ByteDatatypes -> {
+                    asmgen.out("  lda  $arrayVarName+$indexValue |  sta  P8ESTACK_LO,x |  dex")
+                }
+                in WordDatatypes -> {
+                    asmgen.out("  lda  $arrayVarName+$indexValue |  sta  P8ESTACK_LO,x |  lda  $arrayVarName+$indexValue+1 |  sta  P8ESTACK_HI,x |  dex")
+                }
+                DataType.FLOAT -> {
+                    asmgen.out("  lda  #<$arrayVarName+$indexValue |  ldy  #>$arrayVarName+$indexValue |  jsr  floats.push_float")
+                }
+                else -> throw AssemblyError("weird element type")
+            }
+        } else {
+            when(elementDt) {
+                in ByteDatatypes -> {
+                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.Y)
+                    asmgen.out(" lda  $arrayVarName,y |  sta  P8ESTACK_LO,x |  dex")
+                }
+                in WordDatatypes -> {
+                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.Y)
+                    asmgen.out(" lda  $arrayVarName,y |  sta  P8ESTACK_LO,x |  lda  $arrayVarName+1,y |  sta  P8ESTACK_HI,x |  dex")
+                }
+                DataType.FLOAT -> {
+                    asmgen.loadScaledArrayIndexIntoRegister(arrayExpr, elementDt, CpuRegister.A)
+                    asmgen.out("""
+                        ldy  #>$arrayVarName
+                        clc
+                        adc  #<$arrayVarName
+                        bcc  +
+                        iny
++                       jsr  floats.push_float""")
+                }
+                else -> throw AssemblyError("weird dt")
+            }
+
+        }
+    }
+
+    private fun translateBinaryOperatorBytes(operator: String, types: DataType) {
+        when(operator) {
+            "**" -> throw AssemblyError("** operator requires floats")
+            "*" -> asmgen.out("  jsr  prog8_lib.mul_byte")  //  the optimized routines should have been checked earlier
+            "/" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.idiv_ub" else "  jsr  prog8_lib.idiv_b")
+            "%" -> {
+                if(types==DataType.BYTE)
+                    throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
+                asmgen.out("  jsr prog8_lib.remainder_ub")
+            }
+            "+" -> asmgen.out("""
+                lda  P8ESTACK_LO+2,x
+                clc
+                adc  P8ESTACK_LO+1,x
+                inx
+                sta  P8ESTACK_LO+1,x
+                """)
+            "-" -> asmgen.out("""
+                lda  P8ESTACK_LO+2,x
+                sec
+                sbc  P8ESTACK_LO+1,x
+                inx
+                sta  P8ESTACK_LO+1,x
+                """)
+            "<<" -> asmgen.out("  jsr  prog8_lib.shiftleft_b")
+            ">>" -> asmgen.out("  jsr  prog8_lib.shiftright_b")
+            "<" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.less_ub" else "  jsr  prog8_lib.less_b")
+            ">" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.greater_ub" else "  jsr  prog8_lib.greater_b")
+            "<=" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.lesseq_ub" else "  jsr  prog8_lib.lesseq_b")
+            ">=" -> asmgen.out(if(types==DataType.UBYTE) "  jsr  prog8_lib.greatereq_ub" else "  jsr  prog8_lib.greatereq_b")
+            "==" -> asmgen.out("  jsr  prog8_lib.equal_b")
+            "!=" -> asmgen.out("  jsr  prog8_lib.notequal_b")
+            "&" -> asmgen.out("  jsr  prog8_lib.bitand_b")
+            "^" -> asmgen.out("  jsr  prog8_lib.bitxor_b")
+            "|" -> asmgen.out("  jsr  prog8_lib.bitor_b")
+            "and" -> asmgen.out("  jsr  prog8_lib.and_b")
+            "or" -> asmgen.out("  jsr  prog8_lib.or_b")
+            "xor" -> asmgen.out("  jsr  prog8_lib.xor_b")
+            else -> throw AssemblyError("invalid operator $operator")
+        }
+    }
+
+    private fun translateBinaryOperatorWords(operator: String, dt: DataType) {
+        when(operator) {
+            "**" -> throw AssemblyError("** operator requires floats")
+            "*" -> asmgen.out("  jsr  prog8_lib.mul_word")
+            "/" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.idiv_uw" else "  jsr  prog8_lib.idiv_w")
+            "%" -> {
+                if(dt==DataType.WORD)
+                    throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
+                asmgen.out("  jsr prog8_lib.remainder_uw")
+            }
+            "+" -> asmgen.out("  jsr  prog8_lib.add_w")
+            "-" -> asmgen.out("  jsr  prog8_lib.sub_w")
+            "<<" -> asmgen.out("  jsr  math.shift_left_w")
+            ">>" -> {
+                if(dt==DataType.UWORD)
+                    asmgen.out("  jsr  math.shift_right_uw")
+                else
+                    asmgen.out("  jsr  math.shift_right_w")
+            }
+            "<" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.less_uw" else "  jsr  prog8_lib.less_w")
+            ">" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.greater_uw" else "  jsr  prog8_lib.greater_w")
+            "<=" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.lesseq_uw" else "  jsr  prog8_lib.lesseq_w")
+            ">=" -> asmgen.out(if(dt==DataType.UWORD) "  jsr  prog8_lib.greatereq_uw" else "  jsr  prog8_lib.greatereq_w")
+            "==" -> asmgen.out("  jsr  prog8_lib.equal_w")
+            "!=" -> asmgen.out("  jsr  prog8_lib.notequal_w")            "&" -> asmgen.out("  jsr  prog8_lib.bitand_w")
+            "^" -> asmgen.out("  jsr  prog8_lib.bitxor_w")
+            "|" -> asmgen.out("  jsr  prog8_lib.bitor_w")
+            "and" -> asmgen.out("  jsr  prog8_lib.and_w")
+            "or" -> asmgen.out("  jsr  prog8_lib.or_w")
+            "xor" -> asmgen.out("  jsr  prog8_lib.xor_w")
+            else -> throw AssemblyError("invalid operator $operator")
+        }
+    }
+
+    private fun translateBinaryOperatorFloats(operator: String) {
+        when(operator) {
+            "**" -> asmgen.out(" jsr  floats.pow_f")
+            "*" -> asmgen.out("  jsr  floats.mul_f")
+            "/" -> asmgen.out("  jsr  floats.div_f")
+            "+" -> asmgen.out("  jsr  floats.add_f")
+            "-" -> asmgen.out("  jsr  floats.sub_f")
+            "<" -> asmgen.out("  jsr  floats.less_f")
+            ">" -> asmgen.out("  jsr  floats.greater_f")
+            "<=" -> asmgen.out("  jsr  floats.lesseq_f")
+            ">=" -> asmgen.out("  jsr  floats.greatereq_f")
+            "==" -> asmgen.out("  jsr  floats.equal_f")
+            "!=" -> asmgen.out("  jsr  floats.notequal_f")
+            "%", "<<", ">>", "&", "^", "|", "and", "or", "xor" -> throw AssemblyError("requires integer datatype")
+            else -> throw AssemblyError("invalid operator $operator")
+        }
+    }
+
+    private fun translateCompareStrings(s1: Expression, operator: String, s2: Expression) {
+        asmgen.assignExpressionToVariable(s1, "prog8_lib.strcmp_expression._arg_s1", DataType.UWORD, null)
+        asmgen.assignExpressionToVariable(s2, "prog8_lib.strcmp_expression._arg_s2", DataType.UWORD, null)
+        asmgen.out(" jsr  prog8_lib.strcmp_expression")    // result  of compare is in A
+        when(operator) {
+            "==" -> asmgen.out(" and  #1 |  eor  #1 |  sta  P8ESTACK_LO,x")
+            "!=" -> asmgen.out(" and  #1 |  sta  P8ESTACK_LO,x")
+            "<=" -> asmgen.out("""
+                bpl  +
+                lda  #1
+                bne  ++
++               lda  #0
++               sta  P8ESTACK_LO,x""")
+            ">=" -> asmgen.out("""
+                bmi  +
+                lda  #1
+                bne  ++
++               lda  #0
++               sta  P8ESTACK_LO,x""")
+            "<" -> asmgen.out("""
+                bmi  +
+                lda  #0
+                beq  ++
++               lda  #1
++               sta  P8ESTACK_LO,x""")
+            ">" -> asmgen.out("""
+                bpl  +
+                lda  #0
+                beq  ++
++               lda  #1
++               sta  P8ESTACK_LO,x""")
+        }
+        asmgen.out("  dex")
+    }
+}

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

@@ -1,14 +1,14 @@
-package prog8.compiler.target.c64.codegen
+package prog8.codegen.target.cpu6502.codegen
 
 import prog8.ast.Program
-import prog8.ast.base.ArrayElementTypes
+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.compiler.AssemblyError
-import prog8.compiler.toHex
+import prog8.ast.toHex
+import prog8.codegen.target.AssemblyError
 import kotlin.math.absoluteValue
 
 internal class ForLoopsAsmGen(private val program: Program, private val asmgen: AsmGen) {
@@ -21,13 +21,13 @@ internal class ForLoopsAsmGen(private val program: Program, private val asmgen:
             is RangeExpr -> {
                 val range = (stmt.iterable as RangeExpr).toConstantIntegerRange()
                 if(range==null) {
-                    translateForOverNonconstRange(stmt, iterableDt.typeOrElse(DataType.STRUCT), stmt.iterable as RangeExpr)
+                    translateForOverNonconstRange(stmt, iterableDt.getOrElse { throw AssemblyError("unknown dt") }, stmt.iterable as RangeExpr)
                 } else {
-                    translateForOverConstRange(stmt, iterableDt.typeOrElse(DataType.STRUCT), range)
+                    translateForOverConstRange(stmt, iterableDt.getOrElse { throw AssemblyError("unknown dt") }, range)
                 }
             }
             is IdentifierReference -> {
-                translateForOverIterableVar(stmt, iterableDt.typeOrElse(DataType.STRUCT), stmt.iterable as IdentifierReference)
+                translateForOverIterableVar(stmt, iterableDt.getOrElse { throw AssemblyError("unknown dt") }, stmt.iterable as IdentifierReference)
             }
             else -> throw AssemblyError("can't iterate over ${stmt.iterable.javaClass} - should have been replaced by a variable")
         }
@@ -40,6 +40,13 @@ internal class ForLoopsAsmGen(private val program: Program, private val asmgen:
         val modifiedLabel2 = asmgen.makeLabel("for_modifiedb")
         asmgen.loopEndLabels.push(endLabel)
         val stepsize=range.step.constValue(program)!!.number.toInt()
+
+        if(stepsize < -1) {
+            val limit = range.to.constValue(program)?.number
+            if(limit==0.0)
+                throw AssemblyError("for unsigned loop variable it's not possible to count down with step != -1 from a non-const value to exactly zero due to value wrapping")
+        }
+
         when(iterableDt) {
             DataType.ARRAY_B, DataType.ARRAY_UB -> {
                 if (stepsize==1 || stepsize==-1) {
@@ -49,17 +56,17 @@ internal class ForLoopsAsmGen(private val program: Program, private val asmgen:
                     val incdec = if(stepsize==1) "inc" else "dec"
                     // loop over byte range via loopvar
                     val varname = asmgen.asmVariableName(stmt.loopVar)
-                    asmgen.assignExpressionToVariable(range.from, varname, ArrayElementTypes.getValue(iterableDt), null)
-                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.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
-                        jmp  $loopLabel
-$endLabel""")
+                        $incdec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
 
                 } else {
 
@@ -67,8 +74,8 @@ $endLabel""")
 
                     // loop over byte range via loopvar
                     val varname = asmgen.asmVariableName(stmt.loopVar)
-                    asmgen.assignExpressionToVariable(range.from, varname, ArrayElementTypes.getValue(iterableDt), null)
-                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt), null)
+                    asmgen.assignExpressionToVariable(range.to, "$modifiedLabel+1", ArrayToElementTypes.getValue(iterableDt), null)
                     asmgen.out(loopLabel)
                     asmgen.translate(stmt.body)
                     if(stepsize>0) {
@@ -117,16 +124,15 @@ $modifiedLabel2 cmp  #0    ; modified
                             asmgen.out("""
 +               inc  $varname
                 bne  $loopLabel
-                inc  $varname+1
-                jmp  $loopLabel
-                            """)
+                inc  $varname+1""")
+                            asmgen.jmp(loopLabel)
                         } else {
                             asmgen.out("""
 +               lda  $varname
                 bne  +
                 dec  $varname+1
-+               dec  $varname
-                jmp  $loopLabel""")
++               dec  $varname""")
+                            asmgen.jmp(loopLabel)
                         }
                         asmgen.out(endLabel)
                     }
@@ -239,7 +245,7 @@ $endLabel""")
         val endLabel = asmgen.makeLabel("for_end")
         asmgen.loopEndLabels.push(endLabel)
         val iterableName = asmgen.asmVariableName(ident)
-        val decl = ident.targetVarDecl(program.namespace)!!
+        val decl = ident.targetVarDecl(program)!!
         when(iterableDt) {
             DataType.STR -> {
                 asmgen.out("""
@@ -282,7 +288,7 @@ $loopLabel          sty  $indexVar
                         bne  $loopLabel
                         beq  $endLabel""")
                 }
-                if(length>=16 && asmgen.zeropage.available() > 0) {
+                if(length>=16 && asmgen.zeropage.hasByteAvailable()) {
                     // allocate index var on ZP
                     val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
                     asmgen.out("""$indexVar = $zpAddr  ; auto zp UBYTE""")
@@ -321,7 +327,7 @@ $loopLabel          sty  $indexVar
                         bne  $loopLabel
                         beq  $endLabel""")
                 }
-                if(length>=16 && asmgen.zeropage.available() > 0) {
+                if(length>=16 && asmgen.zeropage.hasByteAvailable()) {
                     // allocate index var on ZP
                     val zpAddr = asmgen.zeropage.allocate(indexVar, DataType.UBYTE, stmt.position, asmgen.errors)
                     asmgen.out("""$indexVar = $zpAddr  ; auto zp UBYTE""")
@@ -369,7 +375,7 @@ $loopLabel""")
                         throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $stmt")
                     }
                     2 -> {
-                        if(range.last==255) {
+                        if(range.last==255 || range.last==254) {
                             asmgen.out("""
                                 inc  $varname
                                 beq  $endLabel
@@ -377,34 +383,34 @@ $loopLabel""")
                                 bne  $loopLabel""")
                         } else {
                             asmgen.out("""
+                                inc  $varname
+                                inc  $varname
                                 lda  $varname
-                                cmp  #${range.last}
-                                beq  $endLabel
-                                inc  $varname
-                                inc  $varname
-                                jmp  $loopLabel""")
+                                cmp  #${range.last+2}
+                                bne  $loopLabel""")
                         }
                     }
                     -2 -> {
                         when (range.last) {
-                            0 -> asmgen.out("""
-                                lda  $varname
-                                beq  $endLabel
-                                dec  $varname
-                                dec  $varname
-                                jmp  $loopLabel""")
+                            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""")
+                                    dec  $varname
+                                    beq  $endLabel
+                                    dec  $varname
+                                    bne  $loopLabel""")
                             else -> asmgen.out("""
-                                lda  $varname
-                                cmp  #${range.last}
-                                beq  $endLabel
-                                dec  $varname
-                                dec  $varname
-                                jmp  $loopLabel""")
+                                    dec  $varname
+                                    dec  $varname
+                                    lda  $varname
+                                    cmp  #${range.last-2}
+                                    bne  $loopLabel""")
                         }
                     }
                     else -> {
@@ -415,8 +421,8 @@ $loopLabel""")
                             beq  $endLabel
                             clc
                             adc  #${range.step}
-                            sta  $varname
-                            jmp  $loopLabel""")
+                            sta  $varname""")
+                        asmgen.jmp(loopLabel)
                     }
                 }
                 asmgen.out(endLabel)
@@ -452,9 +458,9 @@ $loopLabel""")
                             sta  $varname
                             lda  $varname+1
                             adc  #>${range.step}
-                            sta  $varname+1
-                            jmp  $loopLabel
-$endLabel""")
+                            sta  $varname+1""")
+                        asmgen.jmp(loopLabel)
+                        asmgen.out(endLabel)
                     }
                 }
             }
@@ -480,11 +486,10 @@ $loopLabel""")
 $endLabel""")
         } else {
             asmgen.out("""
-                lda  $varname
-                cmp  #${range.last}
-                beq  $endLabel
                 inc  $varname
-                jmp  $loopLabel
+                lda  $varname
+                cmp  #${range.last+1}
+                bne  $loopLabel
 $endLabel""")
         }
         asmgen.loopEndLabels.pop()
@@ -505,23 +510,22 @@ $loopLabel""")
                 asmgen.out("""
                     lda  $varname
                     beq  $endLabel
-                    dec  $varname
-                    jmp  $loopLabel
-$endLabel""")
+                    dec  $varname""")
+                asmgen.jmp(loopLabel)
+                asmgen.out(endLabel)
             }
             1 -> {
                 asmgen.out("""
                     dec  $varname
-                    jmp  $loopLabel
+                    bne  $loopLabel
 $endLabel""")
             }
             else -> {
                 asmgen.out("""
-                    lda  $varname
-                    cmp  #${range.last}
-                    beq  $endLabel
                     dec  $varname
-                    jmp  $loopLabel
+                    lda  $varname
+                    cmp  #${range.last-1}
+                    bne  $loopLabel
 $endLabel""")
             }
         }
@@ -546,13 +550,12 @@ $loopLabel""")
             bne  +
             lda  $varname+1
             cmp  #>${range.last}
-            bne  +
             beq  $endLabel
 +           inc  $varname
             bne  $loopLabel
-            inc  $varname+1
-            jmp  $loopLabel
-$endLabel""")
+            inc  $varname+1""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
         asmgen.loopEndLabels.pop()
     }
 
@@ -574,17 +577,20 @@ $loopLabel""")
             bne  +
             lda  $varname+1
             cmp  #>${range.last}
-            bne  +
             beq  $endLabel
 +           lda  $varname
             bne  +
             dec  $varname+1
-+           dec  $varname
-            jmp  $loopLabel
-$endLabel""")
++           dec  $varname""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
         asmgen.loopEndLabels.pop()
     }
 
     private fun assignLoopvar(stmt: ForLoop, range: RangeExpr) =
-        asmgen.assignExpressionToVariable(range.from, asmgen.asmVariableName(stmt.loopVar), stmt.loopVarDt(program).typeOrElse(DataType.STRUCT), stmt.definingSubroutine())
+        asmgen.assignExpressionToVariable(
+            range.from,
+            asmgen.asmVariableName(stmt.loopVar),
+            stmt.loopVarDt(program).getOrElse { throw AssemblyError("unknown dt") },
+            stmt.definingSubroutine)
 }

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

@@ -0,0 +1,382 @@
+package prog8.codegen.target.cpu6502.codegen
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.codegen.target.AssemblyError
+import prog8.codegen.target.cpu6502.codegen.assignment.AsmAssignSource
+import prog8.codegen.target.cpu6502.codegen.assignment.AsmAssignTarget
+import prog8.codegen.target.cpu6502.codegen.assignment.AsmAssignment
+import prog8.codegen.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, false)
+        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 saveXbeforeCall(gosub: GoSub) {
+        val sub = gosub.identifier?.targetSubroutine(program)
+        if(sub?.shouldSaveX()==true) {
+            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, gosub.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 restoreXafterCall(gosub: GoSub) {
+        val sub = gosub.identifier?.targetSubroutine(program)
+        if(sub?.shouldSaveX()==true) {
+            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 optimizeIntArgsViaRegisters(sub: Subroutine) =
+        (sub.parameters.size==1 && sub.parameters[0].type in IntegerDatatypes)
+                || (sub.parameters.size==2 && sub.parameters[0].type in ByteDatatypes && sub.parameters[1].type in ByteDatatypes)
+
+    internal fun translateFunctionCall(call: IFunctionCall, isExpression: Boolean) {
+        // 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 = call.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${call.target}")
+        val subAsmName = asmgen.asmSymbolName(call.target)
+
+        if(!isExpression && !sub.isAsmSubroutine) {
+            if(!optimizeIntArgsViaRegisters(sub))
+                throw AssemblyError("functioncall statements to non-asmsub should have been replaced by GoSub $call")
+        }
+
+        if(sub.isAsmSubroutine) {
+            argumentsViaRegisters(sub, call)
+            if (sub.inline && asmgen.options.optimize) {
+                // 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)
+                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}")
+            } else {
+                asmgen.out("  jsr  $subAsmName")
+            }
+        }
+        else {
+            if(sub.inline)
+                throw AssemblyError("can only reliably inline asmsub routines at this time")
+
+            if(optimizeIntArgsViaRegisters(sub)) {
+                if(sub.parameters.size==1) {
+                    val register = if (sub.parameters[0].type in ByteDatatypes) RegisterOrPair.A else RegisterOrPair.AY
+                    argumentViaRegister(sub, IndexedValue(0, sub.parameters[0]), call.args[0], register)
+                } else {
+                    // 2 byte params, second in Y, first in A
+                    argumentViaRegister(sub, IndexedValue(0, sub.parameters[0]), call.args[0], RegisterOrPair.A)
+                    if(!call.args[1].isSimple)
+                        asmgen.out("  pha")
+                    argumentViaRegister(sub, IndexedValue(1, sub.parameters[1]), call.args[1], RegisterOrPair.Y)
+                    if(!call.args[1].isSimple)
+                        asmgen.out("  pla")
+                }
+            } else {
+                argumentsViaVariables(sub, call)
+            }
+            asmgen.out("  jsr  $subAsmName")
+        }
+
+        // remember: dealing with the X register and/or dealing with return values is the responsibility of the caller
+    }
+
+    private fun argumentsViaVariables(sub: Subroutine, call: IFunctionCall) {
+        for(arg in sub.parameters.withIndex().zip(call.args))
+            argumentViaVariable(sub, arg.first, arg.second)
+    }
+
+    private fun argumentsViaRegisters(sub: Subroutine, call: IFunctionCall) {
+        if(sub.parameters.size==1) {
+            argumentViaRegister(sub, IndexedValue(0, sub.parameters.single()), call.args[0])
+        } else {
+            if(asmgen.asmsubArgsHaveRegisterClobberRisk(call.args, sub.asmParameterRegisters)) {
+                registerArgsViaStackEvaluation(call, sub)
+            } else {
+                asmgen.asmsubArgsEvalOrder(sub).forEach {
+                    val param = sub.parameters[it]
+                    val arg = call.args[it]
+                    argumentViaRegister(sub, IndexedValue(it, param), arg)
+                }
+            }
+        }
+    }
+
+    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("""
+                clc
+                lda  P8ESTACK_LO$plusIdxStr,x
+                beq  +
+                sec
++               php""")             // push the status flags
+        }
+
+        if(argForAregister!=null) {
+            val plusIdxStr = if(argForAregister.index==0) "" else "+${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)
+        val valueDt = valueIDt.getOrElse { throw AssemblyError("unknown dt") }
+        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, registerOverride: RegisterOrPair? = null) {
+        // pass argument via a register parameter
+        val valueIDt = value.inferType(program)
+        val valueDt = valueIDt.getOrElse { throw AssemblyError("unknown dt") }
+        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
+            throw AssemblyError("argument type incompatible")
+
+        val paramRegister = if(registerOverride==null) sub.asmParameterRegisters[parameter.index] else RegisterOrStatusflag(registerOverride, null)
+        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
+                            clc
+                            lda  $sourceName
+                            beq  +
+                            sec  
++                           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 {
+                        val signed = parameter.value.type == DataType.BYTE || parameter.value.type == DataType.WORD
+                        AsmAssignTarget.fromRegisters(register, signed, 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/codegen/target/cpu6502/codegen/PostIncrDecrAsmGen.kt

@@ -1,12 +1,12 @@
-package prog8.compiler.target.c64.codegen
+package prog8.codegen.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.compiler.AssemblyError
-import prog8.compiler.toHex
+import prog8.ast.toHex
+import prog8.codegen.target.AssemblyError
 
 
 internal class PostIncrDecrAsmGen(private val program: Program, private val asmgen: AsmGen) {
@@ -15,11 +15,11 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
         val targetIdent = stmt.target.identifier
         val targetMemory = stmt.target.memoryAddress
         val targetArrayIdx = stmt.target.arrayindexed
-        val scope = stmt.definingSubroutine()
+        val scope = stmt.definingSubroutine
         when {
             targetIdent!=null -> {
                 val what = asmgen.asmVariableName(targetIdent)
-                when (stmt.target.inferType(program).typeOrElse(DataType.STRUCT)) {
+                when (stmt.target.inferType(program).getOr(DataType.UNDEFINED)) {
                     in ByteDatatypes -> asmgen.out(if (incr) "  inc  $what" else "  dec  $what")
                     in WordDatatypes -> {
                         if(incr)
@@ -65,9 +65,10 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
             }
             targetArrayIdx!=null -> {
                 val asmArrayvarname = asmgen.asmVariableName(targetArrayIdx.arrayvar)
-                val elementDt = targetArrayIdx.inferType(program).typeOrElse(DataType.STRUCT)
-                if(targetArrayIdx.indexer.indexNum!=null) {
-                    val indexValue = targetArrayIdx.indexer.constIndex()!! * elementDt.memorySize()
+                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 -> {

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

@@ -1,11 +1,12 @@
-package prog8.compiler.target.c64.codegen.assignment
+package prog8.codegen.target.cpu6502.codegen.assignment
 
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
 import prog8.ast.statements.*
-import prog8.compiler.AssemblyError
-import prog8.compiler.target.c64.codegen.AsmGen
+import prog8.compilerinterface.IMemSizer
+import prog8.codegen.target.AssemblyError
+import prog8.codegen.target.cpu6502.codegen.AsmGen
 
 
 internal enum class TargetStorageKind {
@@ -38,13 +39,14 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
                                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
-        get() = if(array==null)
+    val constMemoryAddress by lazy { memory?.addressExpression?.constValue(program)?.number?.toUInt() ?: 0u}
+    val constArrayIndexValue by lazy { array?.indexer?.constIndex()?.toUInt() }
+    val asmVarname: String by lazy {
+        if (array == null)
             variableAsmName!!
         else
             asmgen.asmVariableName(array.arrayvar)
+    }
 
     lateinit var origAssign: AsmAssignment
 
@@ -54,27 +56,40 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
     }
 
     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.typeOrElse(DataType.STRUCT)
-            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 fromAstAssignment(assign: Assignment, program: Program, asmgen: AsmGen): AsmAssignTarget {
+            with(assign.target) {
+                val idt = inferType(program)
+                val dt = idt.getOrElse { throw AssemblyError("unknown dt") }
+                when {
+                    identifier != null -> {
+                        val parameter = identifier!!.targetVarDecl(program)?.subroutineParameter
+                        if (parameter!=null) {
+                            val sub = parameter.definingSubroutine!!
+                            if (sub.isAsmSubroutine) {
+                                val reg = sub.asmParameterRegisters[sub.parameters.indexOf(parameter)]
+                                if(reg.statusflag!=null)
+                                    throw AssemblyError("can't assign value to processor statusflag directly")
+                                else
+                                    return AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, dt, assign.definingSubroutine, register=reg.registerOrPair, origAstTarget = this)
+                            }
+                        }
+                        return AsmAssignTarget(TargetStorageKind.VARIABLE, program, asmgen, dt, assign.definingSubroutine, variableAsmName = asmgen.asmVariableName(identifier!!), origAstTarget =  this)
+                    }
+                    arrayindexed != null -> return AsmAssignTarget(TargetStorageKind.ARRAY, program, asmgen, dt, assign.definingSubroutine, array = arrayindexed, origAstTarget =  this)
+                    memoryAddress != null -> return AsmAssignTarget(TargetStorageKind.MEMORY, program, asmgen, dt, assign.definingSubroutine, memory =  memoryAddress, origAstTarget =  this)
+                    else -> throw AssemblyError("weird target")
+                }
             }
         }
 
-        fun fromRegisters(registers: RegisterOrPair, scope: Subroutine?, program: Program, asmgen: AsmGen): AsmAssignTarget =
+        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, DataType.UBYTE, scope, register = registers)
+                    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, DataType.UWORD, scope, register = registers)
+                    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,
@@ -92,7 +107,7 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
                     RegisterOrPair.R12,
                     RegisterOrPair.R13,
                     RegisterOrPair.R14,
-                    RegisterOrPair.R15 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, DataType.UWORD, scope, register = registers)
+                    RegisterOrPair.R15 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, if(signed) DataType.WORD else DataType.UWORD, scope, register = registers)
                 }
     }
 }
@@ -104,13 +119,13 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                                private val variableAsmName: String? = null,
                                val array: ArrayIndexedExpression? = null,
                                val memory: DirectMemoryRead? = null,
-                               val register: CpuRegister? = 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 constMemoryAddress by lazy { memory?.addressExpression?.constValue(program)?.number?.toUInt() ?: 0u}
+    val constArrayIndexValue by lazy { array?.indexer?.constIndex()?.toUInt() }
 
     val asmVarname: String
         get() = if(array==null)
@@ -119,13 +134,7 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
             asmgen.asmVariableName(array.arrayvar)
 
     companion object {
-        fun fromAstSource(indexer: ArrayIndex, program: Program, asmgen: AsmGen): AsmAssignSource {
-            return when {
-                indexer.indexNum!=null -> fromAstSource(indexer.indexNum!!, program, asmgen)
-                indexer.indexVar!=null -> fromAstSource(indexer.indexVar!!, program, asmgen)
-                else -> throw AssemblyError("weird indexer")
-            }
-        }
+        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)
@@ -133,33 +142,42 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                 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 NumericLiteralValue -> throw AssemblyError("should have been constant value")
                 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).typeOrElse(DataType.STRUCT)
-                    AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, dt, variableAsmName = asmgen.asmVariableName(value))
+                    val parameter = value.targetVarDecl(program)?.subroutineParameter
+                    if(parameter!=null && parameter.definingSubroutine!!.isAsmSubroutine)
+                        throw AssemblyError("can't assign from a asmsub register parameter $value ${value.position}")
+                    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).typeOrElse(DataType.STRUCT)
+                    val dt = value.inferType(program).getOrElse { throw AssemblyError("unknown dt") }
                     AsmAssignSource(SourceStorageKind.ARRAY, program, asmgen, dt, array = value)
                 }
-                is FunctionCall -> {
-                    when (val sub = value.target.targetStatement(program.namespace)) {
+                is FunctionCallExpr -> {
+                    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 -> {
+                        is BuiltinFunctionPlaceholder -> {
                             val returnType = value.inferType(program)
-                            if(!returnType.isKnown)
-                                throw AssemblyError("unknown dt")
-                            AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, returnType.typeOrElse(DataType.STRUCT), expression = value)
+                            AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, returnType.getOrElse { throw AssemblyError("unknown dt") }, expression = value)
                         }
                         else -> {
                             throw AssemblyError("weird call")
@@ -168,9 +186,7 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                 }
                 else -> {
                     val dt = value.inferType(program)
-                    if(!dt.isKnown)
-                        throw AssemblyError("unknown dt")
-                    AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, dt.typeOrElse(DataType.STRUCT), expression = value)
+                    AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, dt.getOrElse { throw AssemblyError("unknown dt") }, expression = value)
                 }
             }
         }
@@ -198,13 +214,14 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
 internal class AsmAssignment(val source: AsmAssignSource,
                              val target: AsmAssignTarget,
                              val isAugmentable: Boolean,
+                             memsizer: IMemSizer,
                              val position: Position) {
 
     init {
-        if(target.register !in setOf(RegisterOrPair.XY, RegisterOrPair.AX, RegisterOrPair.AY))
-            require(source.datatype != DataType.STRUCT) { "must not be placeholder datatype" }
-            require(source.datatype.memorySize() <= target.datatype.memorySize()) {
-                "source storage size must be less or equal to target datatype storage size"
+        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 dt size must be less or equal to target dt size at $position"
             }
     }
 }

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

@@ -0,0 +1,101 @@
+package prog8.codegen.target.cx16
+
+import prog8.ast.base.DataType
+import prog8.codegen.target.cbm.Mflpt5
+import prog8.codegen.target.cbm.viceMonListName
+import prog8.compilerinterface.*
+import java.io.IOException
+import java.nio.file.Path
+
+
+class CX16MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU65c02
+
+    override val FLOAT_MAX_POSITIVE = Mflpt5.FLOAT_MAX_POSITIVE
+    override val FLOAT_MAX_NEGATIVE = Mflpt5.FLOAT_MAX_NEGATIVE
+    override val FLOAT_MEM_SIZE = Mflpt5.FLOAT_MEM_SIZE
+    override val BASIC_LOAD_ADDRESS = 0x0801u
+    override val RAW_LOAD_ADDRESS = 0x8000u
+
+    // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
+    override val ESTACK_LO = 0x0400u        //  $0400-$04ff inclusive
+    override val ESTACK_HI = 0x0500u        //  $0500-$05ff 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) {
+        val emulatorName: String
+        val extraArgs: List<String>
+
+        when(selectedEmulator) {
+            1 -> {
+                emulatorName = "x16emu"
+                extraArgs = emptyList()
+            }
+            2 -> {
+                emulatorName = "box16"
+                extraArgs = listOf("-sym", viceMonListName(programNameWithPath.toString()))
+            }
+            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 isIOAddress(address: UInt): Boolean = address==0u || address==1u || address in 0x9f00u..0x9fffu
+    override fun getPreallocatedZeropageVars(): Map<String, Pair<UInt, DataType>> {
+        val vars = mutableMapOf<String, Pair<UInt, DataType>>()
+        for(reg in 0..15) {
+            vars["cx16.r${reg}"] = (2+reg*2).toUInt() to DataType.UWORD        // cx16.r0 .. cx16.r15
+            vars["cx16.r${reg}s"] = (2+reg*2).toUInt() to DataType.WORD        // cx16.r0s .. cx16.r15s
+            vars["cx16.r${reg}L"] = (2+reg*2).toUInt() to DataType.UBYTE       // cx16.r0L .. cx16.r15L
+            vars["cx16.r${reg}H"] = (3+reg*2).toUInt() to DataType.UBYTE       // cx16.r0H .. cx16.r15H
+            vars["cx16.r${reg}sL"] = (2+reg*2).toUInt() to DataType.BYTE       // cx16.r0sL .. cx16.r15sL
+            vars["cx16.r${reg}sH"] = (3+reg*2).toUInt() to DataType.BYTE       // cx16.r0sH .. cx16.r15sH
+        }
+        return vars
+    }
+
+    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")
+
+
+}

codeGeneration/src/prog8/codegen/target/cx16/CX16Zeropage.kt

@@ -0,0 +1,41 @@
+package prog8.codegen.target.cx16
+
+import prog8.compilerinterface.CompilationOptions
+import prog8.compilerinterface.InternalCompilerException
+import prog8.compilerinterface.Zeropage
+import prog8.compilerinterface.ZeropageType
+
+class CX16Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0x7au      // temp storage for a single byte
+    override val SCRATCH_REG = 0x7bu     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0x7cu      // temp storage 1 for a word  $7c+$7d
+    override val SCRATCH_W2 = 0x7eu      // 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(0x22u..0xffu)
+            }
+            ZeropageType.KERNALSAFE -> {
+                free.addAll(0x22u..0x7fu)
+                free.addAll(0xa9u..0xffu)
+            }
+            ZeropageType.BASICSAFE -> {
+                free.addAll(0x22u..0x7fu)
+            }
+            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()
+    }
+}

codeOptimizers/build.gradle

@@ -0,0 +1,44 @@
+
+plugins {
+    id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm"
+}
+
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
+}
+
+compileKotlin {
+    kotlinOptions {
+        jvmTarget = javaVersion
+    }
+}
+
+compileTestKotlin {
+    kotlinOptions {
+        jvmTarget = javaVersion
+    }
+}
+
+dependencies {
+    implementation project(':compilerInterfaces')
+    implementation project(':compilerAst')
+    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
+    // implementation "org.jetbrains.kotlin:kotlin-reflect"
+}
+
+sourceSets {
+    main {
+        java {
+            srcDirs = ["${project.projectDir}/src"]
+        }
+        resources {
+            srcDirs = ["${project.projectDir}/res"]
+        }
+    }
+}
+
+// note: there are no unit tests in this module!

codeOptimizers/codeOptimizers.iml

@@ -0,0 +1,15 @@
+<?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" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
+    </content>
+    <orderEntry type="inheritedJdk" />
+    <orderEntry type="sourceFolder" forTests="false" />
+    <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
+    <orderEntry type="module" module-name="compilerInterfaces" />
+    <orderEntry type="module" module-name="compilerAst" />
+  </component>
+</module>

codeOptimizers/readme-tests.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.

codeOptimizers/src/prog8/optimizer/BinExprSplitter.kt

@@ -0,0 +1,137 @@
+package prog8.optimizer
+
+import prog8.ast.IStatementContainer
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.base.FatalAstException
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.TypecastExpression
+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.isIOAddress
+
+
+class BinExprSplitter(private val program: Program, private val options: CompilationOptions, private val compTarget: ICompilationTarget) : AstWalker() {
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+
+        if(assignment.value.inferType(program) istype DataType.FLOAT && !options.optimizeFloatExpressions)
+            return noModifications
+
+        val binExpr = assignment.value as? BinaryExpression
+        if (binExpr != null) {
+
+/*
+
+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 + listOf("==", "!=") && isSimpleTarget(assignment.target)) {
+                if(assignment.target isSameAs binExpr.right)
+                    return noModifications
+                if(assignment.target isSameAs binExpr.left) {
+                    if(binExpr.right.isSimple)
+                        return noModifications
+                    val leftBx = binExpr.left as? BinaryExpression
+                    if(leftBx!=null && (!leftBx.left.isSimple || !leftBx.right.isSimple))
+                        return noModifications
+                    val rightBx = binExpr.right as? BinaryExpression
+                    if(rightBx!=null && (!rightBx.left.isSimple || !rightBx.right.isSimple))
+                        return noModifications
+
+                      // TODO below attempts to remove stack-based evaluated expressions, but often the resulting code is BIGGER, and SLOWER.
+//                    val dt = assignment.target.inferType(program)
+//                    if(!dt.isInteger)
+//                        return noModifications
+//                    val tempVar = IdentifierReference(getTempVarName(dt), binExpr.right.position)
+//                    val assignTempVar = Assignment(
+//                        AssignTarget(tempVar, null, null, binExpr.right.position),
+//                        binExpr.right, binExpr.right.position
+//                    )
+//                    return listOf(
+//                        IAstModification.InsertBefore(assignment, assignTempVar, assignment.parent as IStatementContainer),
+//                        IAstModification.ReplaceNode(binExpr.right, tempVar.copy(), binExpr)
+//                    )
+                }
+
+                if(binExpr.operator == "==" || binExpr.operator == "!=") {
+                    // don't split if the operand(s) don't fit the type of the resulting variable
+                    val targetDt = assignment.target.inferType(program)
+                    val leftDt = binExpr.left.inferType(program)
+                    val rightDt = binExpr.right.inferType(program)
+                    if(leftDt isNotAssignableTo targetDt || rightDt isNotAssignableTo targetDt)
+                        return noModifications
+                }
+
+                if(binExpr.right.isSimple) {
+                    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 work on
+            // the expressiontrees that are not used in an assignment statement...
+        }
+
+        val typecast = assignment.value as? TypecastExpression
+        if(typecast!=null) {
+            val origExpr = typecast.expression as? BinaryExpression
+            if(origExpr!=null) {
+                // it's a typecast of a binary expression.
+                // we can see if we can unwrap the binary expression by working on a new temporary variable
+                // (that has the type of the expression), and then finally doing the typecast.
+                // Once it's outside the typecast, the regular splitting can commence.
+                val tempVar = when(val tempDt = origExpr.inferType(program).getOr(DataType.UNDEFINED)) {
+                    DataType.UBYTE -> listOf("prog8_lib", "retval_interm_ub")
+                    DataType.BYTE -> listOf("prog8_lib", "retval_interm_b")
+                    DataType.UWORD -> listOf("prog8_lib", "retval_interm_uw")
+                    DataType.WORD -> listOf("prog8_lib", "retval_interm_w")
+                    DataType.FLOAT -> listOf("floats", "tempvar_swap_float")
+                    else -> throw FatalAstException("invalid dt $tempDt")
+                }
+                val assignTempVar = Assignment(
+                    AssignTarget(IdentifierReference(tempVar, typecast.position), null, null, typecast.position),
+                    typecast.expression, typecast.position
+                )
+                return listOf(
+                    IAstModification.InsertBefore(assignment, assignTempVar, parent as IStatementContainer),
+                    IAstModification.ReplaceNode(typecast.expression, IdentifierReference(tempVar, typecast.position), typecast)
+                )
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun isSimpleTarget(target: AssignTarget) =
+            if (target.identifier!=null || target.memoryAddress!=null)
+                !target.isIOAddress(compTarget.machine)
+            else
+                false
+
+}

codeOptimizers/src/prog8/optimizer/ConstExprEvaluator.kt

@@ -9,28 +9,32 @@ import kotlin.math.pow
 class ConstExprEvaluator {
 
     fun evaluate(left: NumericLiteralValue, operator: String, right: NumericLiteralValue): Expression {
-        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")
+        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)
         }
     }
 
@@ -42,14 +46,14 @@ class ConstExprEvaluator {
                     left.number.toInt().ushr(amount.number.toInt())
                 else
                     left.number.toInt().shr(amount.number.toInt())
-        return NumericLiteralValue(left.type, result, left.position)
+        return NumericLiteralValue(left.type, result.toDouble(), 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)
+        return NumericLiteralValue(left.type, result.toDouble(), left.position)
     }
 
     private fun logicalxor(left: NumericLiteralValue, right: NumericLiteralValue): NumericLiteralValue {
@@ -57,12 +61,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean((left.number.toInt() != 0) xor (right.number != 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)
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean((left.number != 0.0) xor (right.number.toInt() != 0), left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean((left.number != 0.0) xor (right.number != 0.0), left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -74,12 +78,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 || right.number != 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)
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number != 0.0 || right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number != 0.0 || right.number != 0.0, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -91,12 +95,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number.toInt() != 0 && right.number != 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)
+                in IntegerDatatypes -> NumericLiteralValue.fromBoolean(left.number != 0.0 && right.number.toInt() != 0, left.position)
+                DataType.FLOAT -> NumericLiteralValue.fromBoolean(left.number != 0.0 && right.number != 0.0, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -106,11 +110,11 @@ class ConstExprEvaluator {
     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)
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() xor (right.number.toInt() and 255)).toDouble(), 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)
+                return NumericLiteralValue(DataType.UWORD, (left.number.toInt() xor right.number.toInt()).toDouble(), left.position)
             }
         }
         throw ExpressionError("cannot calculate $left ^ $right", left.position)
@@ -119,11 +123,11 @@ class ConstExprEvaluator {
     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)
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() or (right.number.toInt() and 255)).toDouble(), 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)
+                return NumericLiteralValue(DataType.UWORD, (left.number.toInt() or right.number.toInt()).toDouble(), left.position)
             }
         }
         throw ExpressionError("cannot calculate $left | $right", left.position)
@@ -132,11 +136,11 @@ class ConstExprEvaluator {
     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)
+                return NumericLiteralValue(DataType.UBYTE, (left.number.toInt() and (right.number.toInt() and 255)).toDouble(), 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)
+                return NumericLiteralValue(DataType.UWORD, (left.number.toInt() and right.number.toInt()).toDouble(), left.position)
             }
         }
         throw ExpressionError("cannot calculate $left & $right", left.position)
@@ -147,12 +151,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt().toDouble().pow(right.number), 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)
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number.pow(right.number.toInt()), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.pow(right.number), left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -164,12 +168,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() + right.number, 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)
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number + right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number + right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -181,12 +185,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() - right.number, 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)
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number - right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number - right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -198,12 +202,12 @@ class ConstExprEvaluator {
         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)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number.toInt() * right.number, 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)
+                in IntegerDatatypes -> NumericLiteralValue(DataType.FLOAT, left.number * right.number.toInt(), left.position)
+                DataType.FLOAT -> NumericLiteralValue(DataType.FLOAT, left.number * right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
             else -> throw ExpressionError(error, left.position)
@@ -223,19 +227,19 @@ class ConstExprEvaluator {
                     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)
+                    if(right.number==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toInt() / right.number, 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)
+                    NumericLiteralValue(DataType.FLOAT, left.number / 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)
+                    if(right.number ==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number / right.number, left.position)
                 }
                 else -> throw ExpressionError(error, left.position)
             }
@@ -252,19 +256,19 @@ class ConstExprEvaluator {
                     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)
+                    if(right.number ==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number.toInt() % right.number, 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)
+                    NumericLiteralValue(DataType.FLOAT, left.number % 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)
+                    if(right.number ==0.0) divideByZeroError(right.position)
+                    NumericLiteralValue(DataType.FLOAT, left.number % right.number, left.position)
                 }
                 else -> throw ExpressionError(error, left.position)
             }

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -4,14 +4,15 @@ import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
-import prog8.ast.statements.*
+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
 
 
-internal class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
-    private val noModifications = emptyList<IAstModification>()
+class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
 
     override fun before(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
         // @( &thing )  -->  thing
@@ -22,6 +23,13 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
             noModifications
     }
 
+    override fun after(containment: ContainmentCheck, parent: Node): Iterable<IAstModification> {
+        val result = containment.constValue(program)
+        if(result!=null)
+            return listOf(IAstModification.ReplaceNode(containment, result, parent))
+        return 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.
@@ -39,7 +47,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
                     }
                     DataType.FLOAT -> {
                         listOf(IAstModification.ReplaceNode(expr,
-                                NumericLiteralValue(DataType.FLOAT, -subexpr.number.toDouble(), subexpr.position),
+                                NumericLiteralValue(DataType.FLOAT, -subexpr.number, subexpr.position),
                                 parent))
                     }
                     else -> throw ExpressionError("can only take negative of int or float", subexpr.position)
@@ -47,29 +55,29 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
                 "~" -> when (subexpr.type) {
                     DataType.BYTE -> {
                         listOf(IAstModification.ReplaceNode(expr,
-                                NumericLiteralValue(DataType.BYTE, subexpr.number.toInt().inv(), subexpr.position),
+                                NumericLiteralValue(DataType.BYTE, subexpr.number.toInt().inv().toDouble(), subexpr.position),
                                 parent))
                     }
                     DataType.UBYTE -> {
                         listOf(IAstModification.ReplaceNode(expr,
-                                NumericLiteralValue(DataType.UBYTE, subexpr.number.toInt().inv() and 255, subexpr.position),
+                                NumericLiteralValue(DataType.UBYTE, (subexpr.number.toInt().inv() and 255).toDouble(), subexpr.position),
                                 parent))
                     }
                     DataType.WORD -> {
                         listOf(IAstModification.ReplaceNode(expr,
-                                NumericLiteralValue(DataType.WORD, subexpr.number.toInt().inv(), subexpr.position),
+                                NumericLiteralValue(DataType.WORD, subexpr.number.toInt().inv().toDouble(), subexpr.position),
                                 parent))
                     }
                     DataType.UWORD -> {
                         listOf(IAstModification.ReplaceNode(expr,
-                                NumericLiteralValue(DataType.UWORD, subexpr.number.toInt().inv() and 65535, subexpr.position),
+                                NumericLiteralValue(DataType.UWORD, (subexpr.number.toInt().inv() and 65535).toDouble(), 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),
+                            NumericLiteralValue.fromBoolean(subexpr.number == 0.0, subexpr.position),
                             parent))
                 }
                 else -> throw ExpressionError(expr.operator, subexpr.position)
@@ -78,7 +86,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         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.
@@ -100,34 +108,61 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         val rightconst = expr.right.constValue(program)
         val modifications = mutableListOf<IAstModification>()
 
+        if(expr.operator=="==" && rightconst!=null) {
+            val leftExpr = expr.left as? BinaryExpression
+            if(leftExpr!=null) {
+                val leftRightConst = leftExpr.right.constValue(program)
+                if(leftRightConst!=null) {
+                    when (leftExpr.operator) {
+                        "+" -> {
+                            // X + С1 == C2  -->  X == C2 - C1
+                            val newRightConst = NumericLiteralValue(rightconst.type, rightconst.number - leftRightConst.number, rightconst.position)
+                            return listOf(
+                                IAstModification.ReplaceNode(leftExpr, leftExpr.left, expr),
+                                IAstModification.ReplaceNode(expr.right, newRightConst, expr)
+                            )
+                        }
+                        "-" -> {
+                            // X - С1 == C2  -->  X == C2 + C1
+                            val newRightConst = NumericLiteralValue(rightconst.type, rightconst.number + leftRightConst.number, rightconst.position)
+                            return listOf(
+                                IAstModification.ReplaceNode(leftExpr, leftExpr.left, expr),
+                                IAstModification.ReplaceNode(expr.right, newRightConst, expr)
+                            )
+                        }
+                    }
+                }
+            }
+        }
+
         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).typeOrElse(DataType.STRUCT)
-            when (leftconst.number.toDouble()) {
+            val leftDt = leftconst.inferType(program).getOr(DataType.UNDEFINED)
+            when (leftconst.number) {
                 0.0 -> {
-                    val value = NumericLiteralValue(leftDt, 0, expr.position)
+                    val value = NumericLiteralValue(leftDt, 0.0, expr.position)
                     modifications += IAstModification.ReplaceNode(expr, value, parent)
                 }
                 1.0 -> {
-                    val value = NumericLiteralValue(leftDt, 1, expr.position)
+                    val value = NumericLiteralValue(leftDt, 1.0, 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)
+                        val value = NumericLiteralValue(leftDt, 2.0.pow(rightconst.number), expr.position)
                         modifications += IAstModification.ReplaceNode(expr, value, parent)
                     } else {
-                        val rightDt = expr.right.inferType(program).typeOrElse(DataType.STRUCT)
+                        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).typeOrElse(DataType.STRUCT)
+                                    is Assignment -> parent.target.inferType(program).getOr(DataType.UNDEFINED)
                                     is VarDecl -> parent.datatype
                                     else -> leftDt
                                 }
-                            val one = NumericLiteralValue(targetDt, 1, expr.position)
+                            val one = NumericLiteralValue(targetDt, 1.0, expr.position)
                             val shift = BinaryExpression(one, "<<", expr.right, expr.position)
                             modifications += IAstModification.ReplaceNode(expr, shift, parent)
                         }
@@ -136,7 +171,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
             }
         }
 
-        if(expr.inferType(program).istype(DataType.FLOAT)) {
+        if(expr.inferType(program) istype DataType.FLOAT) {
             val subExpr: BinaryExpression? = when {
                 leftconst != null -> expr.right as? BinaryExpression
                 rightconst != null -> expr.left as? BinaryExpression
@@ -158,13 +193,94 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
             }
         }
 
+        val evaluator = ConstExprEvaluator()
+
         // 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)
         }
 
+
+        val leftBinExpr = expr.left as? BinaryExpression
+        val rightBinExpr = expr.right as? BinaryExpression
+
+        if(leftBinExpr!=null && rightconst!=null) {
+            if(expr.operator=="+" || expr.operator=="-") {
+                if(leftBinExpr.operator in listOf("+", "-")) {
+                    val c2 = leftBinExpr.right.constValue(program)
+                    if(c2!=null) {
+                        // (X + C2) +/- rightConst  -->  X + (C2 +/- rightConst)
+                        // (X - C2) +/- rightConst  -->  X - (C2 +/- rightConst)   mind the flipped right operator
+                        val operator = if(leftBinExpr.operator=="+") expr.operator else if(expr.operator=="-") "+" else "-"
+                        val constants = BinaryExpression(c2, operator, rightconst, c2.position)
+                        val newExpr = BinaryExpression(leftBinExpr.left, leftBinExpr.operator, constants, expr.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    }
+                }
+            }
+            else if(expr.operator=="*") {
+                val c2 = leftBinExpr.right.constValue(program)
+                if(c2!=null) {
+                    if(leftBinExpr.operator=="*") {
+                        // (X * C2) * rightConst  -->  X * (rightConst*C2)
+                        val constants = BinaryExpression(rightconst, "*", c2, c2.position)
+                        val newExpr = BinaryExpression(leftBinExpr.left, "*", constants, expr.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    } else if (leftBinExpr.operator=="/") {
+                        //  (X / C2) * rightConst   -->  X * (rightConst/C2)
+                        val constants = BinaryExpression(rightconst, "/", c2, c2.position)
+                        val newExpr = BinaryExpression(leftBinExpr.left, "*", constants, expr.position)
+                        return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                    }
+                }
+            }
+            else if(expr.operator=="/") {
+                val c2 = leftBinExpr.right.constValue(program)
+                if(c2!=null && leftBinExpr.operator=="/") {
+                    // (X / C1) / C2   -->  X / (C1*C2)
+                    // NOTE: do not optimize  (X * C1) / C2   -->  X * (C1/C2)  because this causes precision loss on integers
+                    val constants = BinaryExpression(c2, "*", rightconst, c2.position)
+                    val newExpr = BinaryExpression(leftBinExpr.left, "/", constants, expr.position)
+                    return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
+                }
+            }
+        }
+
+        if(expr.operator=="+" || expr.operator=="-") {
+            if(leftBinExpr!=null && rightBinExpr!=null) {
+                val c1 = leftBinExpr.right.constValue(program)
+                val c2 = rightBinExpr.right.constValue(program)
+                if(leftBinExpr.operator=="+" && rightBinExpr.operator=="+") {
+                    if (c1 != null && c2 != null) {
+                        // (X + C1) <plusmin> (Y + C2)  =>  (X <plusmin> Y) + (C1 <plusmin> C2)
+                        val c3 = evaluator.evaluate(c1, expr.operator, c2)
+                        val xwithy = BinaryExpression(leftBinExpr.left, expr.operator, rightBinExpr.left, expr.position)
+                        val newExpr = BinaryExpression(xwithy, "+", c3, expr.position)
+                        modifications += IAstModification.ReplaceNode(expr, newExpr, parent)
+                    }
+                }
+                else if(leftBinExpr.operator=="-" && rightBinExpr.operator=="-") {
+                    if (c1 != null && c2 != null) {
+                        // (X - C1) <plusmin> (Y - C2)  =>  (X <plusmin> Y) - (C1 <plusmin> C2)
+                        val c3 = evaluator.evaluate(c1, expr.operator, c2)
+                        val xwithy = BinaryExpression(leftBinExpr.left, expr.operator, rightBinExpr.left, expr.position)
+                        val newExpr = BinaryExpression(xwithy, "-", c3, expr.position)
+                        modifications += IAstModification.ReplaceNode(expr, newExpr, parent)
+                    }
+                }
+                else if(leftBinExpr.operator=="*" && rightBinExpr.operator=="*"){
+                    if (c1 != null && c2 != null && c1==c2) {
+                        //(X * C) <plusmin> (Y * C)  =>  (X <plusmin> Y) * C
+                        val xwithy = BinaryExpression(leftBinExpr.left, expr.operator, rightBinExpr.left, expr.position)
+                        val newExpr = BinaryExpression(xwithy, "*", c1, expr.position)
+                        modifications += IAstModification.ReplaceNode(expr, newExpr, parent)
+                    }
+                }
+            }
+        }
+
+
         return modifications
     }
 
@@ -185,7 +301,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         } else {
             val arrayDt = array.guessDatatype(program)
             if (arrayDt.isKnown) {
-                val newArray = array.cast(arrayDt.typeOrElse(DataType.STRUCT))
+                val newArray = array.cast(arrayDt.getOr(DataType.UNDEFINED))
                 if (newArray != null && newArray != array)
                     return listOf(IAstModification.ReplaceNode(array, newArray, parent))
             }
@@ -194,11 +310,11 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         return noModifications
     }
 
-    override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+    override fun after(functionCallExpr: FunctionCallExpr, parent: Node): Iterable<IAstModification> {
         // the args of a fuction are constfolded via recursion already.
-        val constvalue = functionCall.constValue(program)
+        val constvalue = functionCallExpr.constValue(program)
         return if(constvalue!=null)
-            listOf(IAstModification.ReplaceNode(functionCall, constvalue, parent))
+            listOf(IAstModification.ReplaceNode(functionCallExpr, constvalue, parent))
         else
             noModifications
     }
@@ -230,7 +346,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         val rangeTo = iterableRange.to as? NumericLiteralValue
         if(rangeFrom==null || rangeTo==null) return noModifications
 
-        val loopvar = forLoop.loopVar.targetVarDecl(program.namespace) ?: throw UndefinedSymbolError(forLoop.loopVar)
+        val loopvar = forLoop.loopVar.targetVarDecl(program) ?: throw UndefinedSymbolError(forLoop.loopVar)
 
         val stepLiteral = iterableRange.step as? NumericLiteralValue
         when(loopvar.datatype) {
@@ -276,7 +392,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         val numval = decl.value as? NumericLiteralValue
         if(decl.type== VarDeclType.CONST && numval!=null) {
             val valueDt = numval.inferType(program)
-            if(!valueDt.istype(decl.datatype)) {
+            if(valueDt isnot decl.datatype) {
                 val cast = numval.cast(decl.datatype)
                 if(cast.isValid)
                     return listOf(IAstModification.ReplaceNode(numval, cast.valueOrZero(), decl))
@@ -311,12 +427,12 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
     {
         // NOTE: THIS IS ONLY VALID ON FLOATING POINT CONSTANTS
 
-        // todo: this implements only a small set of possible reorderings at this time
+        // todo: this implements only a small set of possible reorderings at this time, we could think of more
         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) {
+            if(expr.operator in AssociativeOperators) {
                 return if(leftIsConst) {
                     if(subleftIsConst)
                         ShuffleOperands(expr, null, subExpr, subExpr.right, null, null, expr.left)

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -4,27 +4,34 @@ import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
-import prog8.ast.statements.ArrayIndex
-import prog8.ast.statements.AssignTarget
-import prog8.ast.statements.ForLoop
-import prog8.ast.statements.VarDecl
-import prog8.compiler.target.CompilationTarget
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
 
 // Fix up the literal value's type to match that of the vardecl
-internal class VarConstantValueTypeAdjuster(private val program: Program, private val errors: ErrorReporter) : AstWalker() {
-    private val noModifications = emptyList<IAstModification>()
+//   (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).istype(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))
+                && declConstValue.type != decl.datatype) {
+                // avoid silent float roundings
+                if(decl.datatype in IntegerDatatypes && declConstValue.type==DataType.FLOAT) {
+                    errors.err("refused silent rounding of float to avoid loss of precision", decl.value!!.position)
+                } else {
+                    // 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)
@@ -32,14 +39,42 @@ internal class VarConstantValueTypeAdjuster(private val program: Program, privat
 
         return noModifications
     }
+
+    override fun after(range: RangeExpr, parent: Node): Iterable<IAstModification> {
+        val from = range.from.constValue(program)?.number
+        val to = range.to.constValue(program)?.number
+        val step = range.step.constValue(program)?.number
+
+        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: ErrorReporter) : AstWalker() {
-    private val noModifications = emptyList<IAstModification>()
+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
@@ -73,8 +108,7 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
 
     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?.indexVar?.referencesIdentifier(decl.name) == true) {
+        if(decl.value?.referencesIdentifier(listOf(decl.name)) == true || decl.arraysize?.indexExpr?.referencesIdentifier(listOf(decl.name)) == true) {
             errors.err("recursive var declaration", decl.position)
             return noModifications
         }
@@ -92,19 +126,6 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                                 decl
                         ))
                     }
-                } else if(arraysize.constIndex()==null) {
-                    // see if we can calculate the size from other fields
-                    try {
-                        val cval =  arraysize.indexVar?.constValue(program) ?: arraysize.origExpression?.constValue(program)
-                        if (cval != null) {
-                            arraysize.indexVar = null
-                            arraysize.origExpression = null
-                            arraysize.indexNum = cval
-                        }
-                    } catch (x: UndefinedSymbolError) {
-                        errors.err(x.message, x.position)
-                        return noModifications
-                    }
                 }
             }
 
@@ -113,33 +134,33 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                     // 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)
+                        val newValue = NumericLiteralValue(DataType.FLOAT, litval.number, litval.position)
                         return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
                     }
                 }
                 DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W -> {
-                    val numericLv = decl.value as? NumericLiteralValue
                     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!!)
+                            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 eltType = rangeExpr.inferType(program).typeOrElse(DataType.UBYTE)
+                            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(),
+                                        constRange.map { NumericLiteralValue(eltType, it.toDouble(), decl.value!!.position) }.toTypedArray(),
                                         position = decl.value!!.position)
                             } else {
                                 ArrayLiteralValue(InferredTypes.InferredType.known(decl.datatype),
-                                        constRange.map { NumericLiteralValue(eltType, it, decl.value!!.position) }.toTypedArray(),
+                                        constRange.map { NumericLiteralValue(eltType, it.toDouble(), 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
@@ -166,20 +187,18 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                             else -> {}
                         }
                         // create the array itself, filled with the fillvalue.
-                        val array = Array(size) {fillvalue}.map { NumericLiteralValue(ArrayElementTypes.getValue(decl.datatype), it, numericLv.position) }.toTypedArray<Expression>()
+                        val array = Array(size) {fillvalue}.map { NumericLiteralValue(ArrayToElementTypes.getValue(decl.datatype), it.toDouble(), 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 size = decl.arraysize?.constIndex() ?: return noModifications
-                    val litval = decl.value as? NumericLiteralValue
                     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 doesn't match declared array size", decl.value?.position!!)
+                            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),
@@ -188,22 +207,24 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                             return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
                         }
                     }
-                    if(rangeExpr==null && litval!=null) {
+
+                    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 = litval.number.toDouble()
-                        if (fillvalue < CompilationTarget.instance.machine.FLOAT_MAX_NEGATIVE || fillvalue > CompilationTarget.instance.machine.FLOAT_MAX_POSITIVE)
-                            errors.err("float value overflow", litval.position)
+                        val fillvalue = numericLv.number
+                        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, litval.position) }.toTypedArray<Expression>()
-                            val refValue = ArrayLiteralValue(InferredTypes.InferredType.known(DataType.ARRAY_F), array, position = litval.position)
+                            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
-                    // this includes strings and structs
                 }
             }
         }

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -1,28 +1,34 @@
 package prog8.optimizer
 
+import prog8.ast.IStatementContainer
 import prog8.ast.Node
 import prog8.ast.Program
-import prog8.ast.base.*
+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.processing.AstWalker
-import prog8.ast.processing.IAstModification
+import prog8.ast.statements.AnonymousScope
 import prog8.ast.statements.Assignment
+import prog8.ast.statements.IfElse
+import prog8.ast.statements.Jump
+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 expression optimizations
+    todo add more peephole expression optimizations
 
     Investigate what optimizations binaryen has, also see  https://egorbo.com/peephole-optimizations.html
 
  */
 
 
-internal class ExpressionSimplifier(private val program: Program) : AstWalker() {
+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()
-    private val noModifications = emptyList<IAstModification>()
 
     override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
         val mods = mutableListOf<IAstModification>()
@@ -43,7 +49,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                 mods += IAstModification.ReplaceNode(typecast.expression, subTypecast.expression, typecast)
             }
         } else {
-            if (typecast.expression.inferType(program).istype(typecast.type)) {
+            if (typecast.expression.inferType(program) istype typecast.type) {
                 // remove duplicate cast
                 mods += IAstModification.ReplaceNode(typecast, typecast.expression, parent)
             }
@@ -52,35 +58,26 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         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))
+    override fun after(ifElse: IfElse, parent: Node): Iterable<IAstModification> {
+        val truepart = ifElse.truepart
+        val elsepart = ifElse.elsepart
+        if(truepart.isNotEmpty() && elsepart.isNotEmpty()) {
+            if(truepart.statements.singleOrNull() is Jump) {
+                return listOf(
+                    IAstModification.InsertAfter(ifElse, elsepart, parent as IStatementContainer),
+                    IAstModification.ReplaceNode(elsepart, AnonymousScope(mutableListOf(), elsepart.position), ifElse)
+                )
+            }
+            if(elsepart.statements.singleOrNull() is Jump) {
+                val invertedCondition = invertCondition(ifElse.condition)
+                if(invertedCondition!=null) {
+                    return listOf(
+                        IAstModification.ReplaceNode(ifElse.condition, invertedCondition, ifElse),
+                        IAstModification.InsertAfter(ifElse, truepart, parent as IStatementContainer),
+                        IAstModification.ReplaceNode(elsepart, AnonymousScope(mutableListOf(), elsepart.position), ifElse),
+                        IAstModification.ReplaceNode(truepart, elsepart, ifElse)
+                    )
                 }
-                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
@@ -96,11 +93,11 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             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)
+        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 (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))
         }
@@ -132,8 +129,8 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             ))
         }
 
-        val leftDt = leftIDt.typeOrElse(DataType.STRUCT)
-        val rightDt = rightIDt.typeOrElse(DataType.STRUCT)
+        val leftDt = leftIDt.getOr(DataType.UNDEFINED)
+        val rightDt = rightIDt.getOr(DataType.UNDEFINED)
 
         if (expr.operator == "+" || expr.operator == "-"
                 && leftVal == null && rightVal == null
@@ -147,7 +144,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     val x = expr.right
                     val y = determineY(x, leftBinExpr)
                     if (y != null) {
-                        val yPlus1 = BinaryExpression(y, "+", NumericLiteralValue(leftDt, 1, y.position), y.position)
+                        val yPlus1 = BinaryExpression(y, "+", NumericLiteralValue(leftDt, 1.0, y.position), y.position)
                         val newExpr = BinaryExpression(x, "*", yPlus1, x.position)
                         return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
                     }
@@ -157,7 +154,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     val x = expr.right
                     val y = determineY(x, leftBinExpr)
                     if (y != null) {
-                        val yMinus1 = BinaryExpression(y, "-", NumericLiteralValue(leftDt, 1, y.position), y.position)
+                        val yMinus1 = BinaryExpression(y, "-", NumericLiteralValue(leftDt, 1.0, y.position), y.position)
                         val newExpr = BinaryExpression(x, "*", yMinus1, x.position)
                         return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
                     }
@@ -177,14 +174,26 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             }
         }
 
-        if(expr.operator == ">=" && rightVal?.number == 0) {
+        if(leftDt!=DataType.FLOAT && expr.operator == ">=" && rightVal?.number == 1.0) {
+            // for integers: x >= 1  -->  x > 0
+            expr.operator = ">"
+            return listOf(IAstModification.ReplaceNode(expr.right, NumericLiteralValue.optimalInteger(0, expr.right.position), expr))
+        }
+
+        if(expr.operator == ">=" && rightVal?.number == 0.0) {
             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.FLOAT && expr.operator == "<" && rightVal?.number == 1.0) {
+            // for integers: x < 1  -->  x <= 0
+            expr.operator = "<="
+            return listOf(IAstModification.ReplaceNode(expr.right, NumericLiteralValue.optimalInteger(0, expr.right.position), expr))
+        }
+
+        if(expr.operator == "<" && rightVal?.number == 0.0) {
             if (leftDt == DataType.UBYTE || leftDt == DataType.UWORD) {
                 // unsigned < 0 --> false
                 return listOf(IAstModification.ReplaceNode(expr, NumericLiteralValue.fromBoolean(false, expr.position), parent))
@@ -202,7 +211,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     // signedw < 0 -->  msb(signedw) & $80
                     return listOf(IAstModification.ReplaceNode(
                             expr,
-                            BinaryExpression(FunctionCall(IdentifierReference(listOf("msb"), expr.position),
+                            BinaryExpression(FunctionCallExpr(IdentifierReference(listOf("msb"), expr.position),
                                     mutableListOf(expr.left),
                                     expr.position
                             ), "&", NumericLiteralValue.optimalInteger(0x80, expr.position), expr.position),
@@ -218,52 +227,62 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         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
+                when {
+                    leftVal != null && leftVal.asBooleanValue || rightVal != null && rightVal.asBooleanValue -> constTrue
+                    leftVal != null && !leftVal.asBooleanValue -> expr.right
+                    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
+                when {
+                    leftVal != null && !leftVal.asBooleanValue || rightVal != null && !rightVal.asBooleanValue -> constFalse
+                    leftVal != null && leftVal.asBooleanValue -> expr.right
+                    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
+                when {
+                    leftVal != null && !leftVal.asBooleanValue -> expr.right
+                    rightVal != null && !rightVal.asBooleanValue -> expr.left
+                    leftVal != null && leftVal.asBooleanValue -> PrefixExpression("not", expr.right, expr.right.position)
+                    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
+            "|" -> {
+                when {
+                    leftVal?.number==0.0 -> expr.right
+                    rightVal?.number==0.0 -> expr.left
+                    rightIDt.isBytes && rightVal?.number==255.0 -> NumericLiteralValue(DataType.UBYTE, 255.0, rightVal.position)
+                    rightIDt.isWords && rightVal?.number==65535.0 -> NumericLiteralValue(DataType.UWORD, 65535.0, rightVal.position)
+                    leftIDt.isBytes && leftVal?.number==255.0 -> NumericLiteralValue(DataType.UBYTE, 255.0, leftVal.position)
+                    leftIDt.isWords && leftVal?.number==65535.0 -> NumericLiteralValue(DataType.UWORD, 65535.0, leftVal.position)
+                    else -> null
+                }
+            }
+            "^" -> {
+                when {
+                    leftVal?.number==0.0 -> expr.right
+                    rightVal?.number==0.0 -> expr.left
+                    rightIDt.isBytes && rightVal?.number==255.0 -> PrefixExpression("~", expr.left, expr.left.position)
+                    rightIDt.isWords && rightVal?.number==65535.0 -> PrefixExpression("~", expr.left, expr.left.position)
+                    leftIDt.isBytes && leftVal?.number==255.0 -> PrefixExpression("~", expr.right, expr.right.position)
+                    leftIDt.isWords && leftVal?.number==65535.0 -> PrefixExpression("~", expr.right, expr.right.position)
+                    else -> null
+                }
             }
             "&" -> {
-                if (leftVal != null && !leftVal.asBooleanValue)
-                    constFalse
-                else if (rightVal != null && !rightVal.asBooleanValue)
-                    constFalse
-                else
-                    null
+                when {
+                    leftVal?.number==0.0  -> constFalse
+                    rightVal?.number==0.0 -> constFalse
+                    rightIDt.isBytes && rightVal?.number==255.0 -> expr.left
+                    rightIDt.isWords && rightVal?.number==65535.0 -> expr.left
+                    leftIDt.isBytes && leftVal?.number==255.0 -> expr.right
+                    leftIDt.isWords && leftVal?.number==65535.0 -> expr.right
+                    else -> null
+                }
             }
             "*" -> optimizeMultiplication(expr, leftVal, rightVal)
             "/" -> optimizeDivision(expr, leftVal, rightVal)
@@ -282,41 +301,41 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         return noModifications
     }
 
-    override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
-        if(functionCall.target.nameInSource == listOf("lsb")) {
-            val arg = functionCall.args[0]
+    override fun after(functionCallExpr: FunctionCallExpr, parent: Node): Iterable<IAstModification> {
+        if(functionCallExpr.target.nameInSource == listOf("lsb")) {
+            val arg = functionCallExpr.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 casted to word
-                    return listOf(IAstModification.ReplaceNode(functionCall, arg.expression, parent))
+                if (valueDt istype DataType.BYTE || valueDt istype DataType.UBYTE) {
+                    // useless lsb() of byte value that was typecasted to word
+                    return listOf(IAstModification.ReplaceNode(functionCallExpr, arg.expression, parent))
                 }
             } else {
                 val argDt = arg.inferType(program)
-                if (argDt.istype(DataType.BYTE) || argDt.istype(DataType.UBYTE)) {
+                if (argDt istype DataType.BYTE || argDt istype DataType.UBYTE) {
                     // useless lsb() of byte value
-                    return listOf(IAstModification.ReplaceNode(functionCall, arg, parent))
+                    return listOf(IAstModification.ReplaceNode(functionCallExpr, arg, parent))
                 }
             }
         }
-        else if(functionCall.target.nameInSource == listOf("msb")) {
-            val arg = functionCall.args[0]
+        else if(functionCallExpr.target.nameInSource == listOf("msb")) {
+            val arg = functionCallExpr.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 casted to word, replace with 0
+                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.typeOrElse(DataType.UBYTE), 0, arg.expression.position),
+                            functionCallExpr,
+                            NumericLiteralValue(valueDt.getOr(DataType.UBYTE), 0.0, arg.expression.position),
                             parent))
                 }
             } else {
                 val argDt = arg.inferType(program)
-                if (argDt.istype(DataType.BYTE) || argDt.istype(DataType.UBYTE)) {
+                if (argDt istype DataType.BYTE || argDt istype DataType.UBYTE) {
                     // useless msb() of byte value, replace with 0
                     return listOf(IAstModification.ReplaceNode(
-                            functionCall,
-                            NumericLiteralValue(argDt.typeOrElse(DataType.UBYTE), 0, arg.position),
+                            functionCallExpr,
+                            NumericLiteralValue(argDt.getOr(DataType.UBYTE), 0.0, arg.position),
                             parent))
                 }
             }
@@ -325,6 +344,25 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         return noModifications
     }
 
+    override fun after(containment: ContainmentCheck, parent: Node): Iterable<IAstModification> {
+        val range = containment.iterable as? RangeExpr
+        if(range!=null && range.step.constValue(program)?.number==1.0) {
+            val from = range.from.constValue(program)
+            val to = range.to.constValue(program)
+            val value = containment.element
+            if(from!=null && to!=null && value.isSimple) {
+                if(to.number-from.number>6.0) {
+                    // replace containment test with X>=from and X<=to
+                    val left = BinaryExpression(value, ">=", from, containment.position)
+                    val right = BinaryExpression(value.copy(), "<=", to, containment.position)
+                    val comparison = BinaryExpression(left, "and", right, containment.position)
+                    return listOf(IAstModification.ReplaceNode(containment, comparison, parent))
+                }
+            }
+        }
+        return noModifications
+    }
+
     private fun determineY(x: Expression, subBinExpr: BinaryExpression): Expression? {
         return when {
             subBinExpr.left isSameAs x -> subBinExpr.right
@@ -349,7 +387,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (rightVal2 != null) {
             // right value is a constant, see if we can optimize
             val rightConst: NumericLiteralValue = rightVal2
-            when (rightConst.number.toDouble()) {
+            when (rightConst.number) {
                 0.0 -> {
                     // left
                     return expr2.left
@@ -358,7 +396,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         }
         // no need to check for left val constant (because of associativity)
 
-        val rnum = rightVal?.number?.toDouble()
+        val rnum = rightVal?.number
         if(rnum!=null && rnum<0.0) {
             expr.operator = "-"
             expr.right = NumericLiteralValue(rightVal.type, -rnum, rightVal.position)
@@ -379,7 +417,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
 
         if (rightVal != null) {
             // right value is a constant, see if we can optimize
-            val rnum = rightVal.number.toDouble()
+            val rnum = rightVal.number
             if (rnum == 0.0) {
                 // left
                 return expr.left
@@ -393,7 +431,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         }
         if (leftVal != null) {
             // left value is a constant, see if we can optimize
-            when (leftVal.number.toDouble()) {
+            when (leftVal.number) {
                 0.0 -> {
                     // -right
                     return PrefixExpression("-", expr.right, expr.position)
@@ -412,7 +450,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (rightVal != null) {
             // right value is a constant, see if we can optimize
             val rightConst: NumericLiteralValue = rightVal
-            when (rightConst.number.toDouble()) {
+            when (rightConst.number) {
                 -3.0 -> {
                     // -1/(left*left*left)
                     return BinaryExpression(NumericLiteralValue(DataType.FLOAT, -1.0, expr.position), "/",
@@ -432,11 +470,11 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                 }
                 0.0 -> {
                     // 1
-                    return NumericLiteralValue(rightConst.type, 1, expr.position)
+                    return NumericLiteralValue(rightConst.type, 1.0, expr.position)
                 }
                 0.5 -> {
                     // sqrt(left)
-                    return FunctionCall(IdentifierReference(listOf("sqrt"), expr.position), mutableListOf(expr.left), expr.position)
+                    return FunctionCallExpr(IdentifierReference(listOf("sqrt"), expr.position), mutableListOf(expr.left), expr.position)
                 }
                 1.0 -> {
                     // left
@@ -454,18 +492,18 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         }
         if (leftVal != null) {
             // left value is a constant, see if we can optimize
-            when (leftVal.number.toDouble()) {
+            when (leftVal.number) {
                 -1.0 -> {
                     // -1
                     return NumericLiteralValue(DataType.FLOAT, -1.0, expr.position)
                 }
                 0.0 -> {
                     // 0
-                    return NumericLiteralValue(leftVal.type, 0, expr.position)
+                    return NumericLiteralValue(leftVal.type, 0.0, expr.position)
                 }
                 1.0 -> {
                     //1
-                    return NumericLiteralValue(leftVal.type, 1, expr.position)
+                    return NumericLiteralValue(leftVal.type, 1.0, expr.position)
                 }
 
             }
@@ -487,10 +525,10 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     val idt = expr.inferType(program)
                     if(!idt.isKnown)
                         throw FatalAstException("unknown dt")
-                    return NumericLiteralValue(idt.typeOrElse(DataType.STRUCT), 0, expr.position)
-                } else if (cv == 2.0) {
+                    return NumericLiteralValue(idt.getOr(DataType.UNDEFINED), 0.0, expr.position)
+                } else if (cv in powersOfTwo) {
                     expr.operator = "&"
-                    expr.right = NumericLiteralValue.optimalInteger(1, expr.position)
+                    expr.right = NumericLiteralValue.optimalInteger(cv!!.toInt()-1, expr.position)
                     return null
                 }
             }
@@ -507,11 +545,11 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (rightVal != null) {
             // right value is a constant, see if we can optimize
             val rightConst: NumericLiteralValue = rightVal
-            val cv = rightConst.number.toDouble()
+            val cv = rightConst.number
             val leftIDt = expr.left.inferType(program)
             if (!leftIDt.isKnown)
                 return null
-            val leftDt = leftIDt.typeOrElse(DataType.STRUCT)
+            val leftDt = leftIDt.getOr(DataType.UNDEFINED)
             when (cv) {
                 -1.0 -> {
                     //  '/' -> -left
@@ -542,22 +580,22 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             }
 
             if (leftDt == DataType.UBYTE) {
-                if (abs(rightConst.number.toDouble()) >= 256.0) {
-                    return NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                if (abs(rightConst.number) >= 256.0) {
+                    return NumericLiteralValue(DataType.UBYTE, 0.0, expr.position)
                 }
             } else if (leftDt == DataType.UWORD) {
-                if (abs(rightConst.number.toDouble()) >= 65536.0) {
-                    return NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                if (abs(rightConst.number) >= 65536.0) {
+                    return NumericLiteralValue(DataType.UBYTE, 0.0, expr.position)
                 }
             }
         }
 
         if (leftVal != null) {
             // left value is a constant, see if we can optimize
-            when (leftVal.number.toDouble()) {
+            when (leftVal.number) {
                 0.0 -> {
                     // 0
-                    return NumericLiteralValue(leftVal.type, 0, expr.position)
+                    return NumericLiteralValue(leftVal.type, 0.0, expr.position)
                 }
             }
         }
@@ -574,28 +612,28 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             // 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()) {
+            when (val cv = rightConst.number) {
                 -1.0 -> {
                     // -left
                     return PrefixExpression("-", leftValue, expr.position)
                 }
                 0.0 -> {
                     // 0
-                    return NumericLiteralValue(rightConst.type, 0, expr.position)
+                    return NumericLiteralValue(rightConst.type, 0.0, expr.position)
                 }
                 1.0 -> {
                     // left
                     return expr2.left
                 }
                 in powersOfTwo -> {
-                    if (leftValue.inferType(program).typeOrElse(DataType.STRUCT) in IntegerDatatypes) {
+                    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).typeOrElse(DataType.STRUCT) in IntegerDatatypes) {
+                    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)
@@ -619,22 +657,22 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         val targetIDt = expr.left.inferType(program)
         if(!targetIDt.isKnown)
             throw FatalAstException("unknown dt")
-        when (val targetDt = targetIDt.typeOrElse(DataType.STRUCT)) {
+        when (val targetDt = targetIDt.getOr(DataType.UNDEFINED)) {
             DataType.UBYTE, DataType.BYTE -> {
                 if (amount >= 8) {
-                    return NumericLiteralValue(targetDt, 0, expr.position)
+                    return NumericLiteralValue(targetDt, 0.0, expr.position)
                 }
             }
             DataType.UWORD, DataType.WORD -> {
                 if (amount >= 16) {
-                    return NumericLiteralValue(targetDt, 0, expr.position)
+                    return NumericLiteralValue(targetDt, 0.0, expr.position)
                 } else if (amount >= 8) {
-                    val lsb = FunctionCall(IdentifierReference(listOf("lsb"), expr.position), mutableListOf(expr.left), expr.position)
+                    val lsb = FunctionCallExpr(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)
+                        return FunctionCallExpr(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)
+                    return FunctionCallExpr(IdentifierReference(listOf("mkword"), expr.position), mutableListOf(shifted, NumericLiteralValue.optimalInteger(0, expr.position)), expr.position)
                 }
             }
             else -> {
@@ -654,7 +692,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         val idt = expr.left.inferType(program)
         if(!idt.isKnown)
             throw FatalAstException("unknown dt")
-        when (idt.typeOrElse(DataType.STRUCT)) {
+        when (idt.getOr(DataType.UNDEFINED)) {
             DataType.UBYTE -> {
                 if (amount >= 8) {
                     return NumericLiteralValue.optimalInteger(0, expr.position)
@@ -671,11 +709,11 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
                     return NumericLiteralValue.optimalInteger(0, expr.position)
                 }
                 else if (amount >= 8) {
-                    val msb = FunctionCall(IdentifierReference(listOf("msb"), expr.position), mutableListOf(expr.left), expr.position)
+                    val msb = FunctionCallExpr(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)
+                        val zero = NumericLiteralValue(DataType.UBYTE, 0.0, expr.position)
+                        return FunctionCallExpr(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)
                 }
@@ -693,7 +731,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
     }
 
     private fun reorderAssociativeWithConstant(expr: BinaryExpression, leftVal: NumericLiteralValue?): BinExprWithConstants {
-        if (expr.operator in associativeOperators && leftVal != null) {
+        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
@@ -705,4 +743,4 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
 
     private data class BinExprWithConstants(val expr: BinaryExpression, val leftVal: NumericLiteralValue?, val rightVal: NumericLiteralValue?)
 
-}
+}

codeOptimizers/src/prog8/optimizer/Extensions.kt

@@ -0,0 +1,82 @@
+package prog8.optimizer
+
+import prog8.ast.IBuiltinFunctions
+import prog8.ast.Program
+import prog8.ast.base.DataType
+import prog8.ast.base.FatalAstException
+import prog8.ast.expressions.InferredTypes
+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()
+}
+
+fun getTempVarName(dt: InferredTypes.InferredType): List<String> {
+    return when {
+        // TODO assume (hope) cx16.r9 isn't used for anything else during the use of this temporary variable...
+        dt istype DataType.UBYTE -> listOf("cx16", "r9L")
+        dt istype DataType.BYTE -> listOf("cx16", "r9sL")
+        dt istype DataType.UWORD -> listOf("cx16", "r9")
+        dt istype DataType.WORD -> listOf("cx16", "r9s")
+        dt.isPassByReference -> listOf("cx16", "r9")
+        else -> throw FatalAstException("invalid dt $dt")
+    }
+}

codeOptimizers/src/prog8/optimizer/StatementOptimizer.kt

@@ -1,179 +1,164 @@
 package prog8.optimizer
 
-import prog8.ast.INameScope
-import prog8.ast.Node
-import prog8.ast.Program
+import prog8.ast.*
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
-import prog8.ast.processing.IAstVisitor
 import prog8.ast.statements.*
-import prog8.compiler.target.CompilationTarget
-import prog8.functions.BuiltinFunctions
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compilerinterface.ICompilationTarget
+import prog8.compilerinterface.IErrorReporter
 import kotlin.math.floor
 
 
-internal class StatementOptimizer(private val program: Program,
-                                  private val errors: ErrorReporter) : AstWalker() {
+class StatementOptimizer(private val program: Program,
+                                  private val errors: IErrorReporter,
+                                  private val functions: IBuiltinFunctions,
+                                  private val compTarget: ICompilationTarget
+) : AstWalker() {
 
-    private val noModifications = emptyList<IAstModification>()
-    private val callgraph = CallGraph(program)
-    private val pureBuiltinFunctions = BuiltinFunctions.filter { it.value.pure }
+    override fun before(functionCallExpr: FunctionCallExpr, 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.
 
-    override fun after(block: Block, parent: Node): Iterable<IAstModification> {
-        if("force_output" !in block.options()) {
-            if (block.containsNoCodeNorVars()) {
-                errors.warn("removing empty block '${block.name}'", block.position)
-                return listOf(IAstModification.Remove(block, parent as INameScope))
-            }
+        fun scopePrefix(variable: IdentifierReference): IdentifierReference {
+            val target = variable.targetStatement(program) as INamedStatement
+            return IdentifierReference(target.scopedName, variable.position)
+        }
 
-            if (block !in callgraph.usedSymbols) {
-                errors.warn("removing unused block '${block.name}'", block.position)
-                return listOf(IAstModification.Remove(block, parent as INameScope))
+        val subroutine = functionCallExpr.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)
+                        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)
+                    is NumericLiteralValue -> orig.copy()
+                    is StringLiteralValue -> orig.copy()
+                    else -> return noModifications
+                }
+                return listOf(IAstModification.ReplaceNode(functionCallExpr, copy, parent))
             }
         }
         return noModifications
     }
 
-    override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
-        val forceOutput = "force_output" in subroutine.definingBlock().options()
-        if(subroutine.asmAddress==null && !forceOutput) {
-            if(subroutine.containsNoCodeNorVars() && !subroutine.inline) {
-                errors.warn("removing empty subroutine '${subroutine.name}'", subroutine.position)
-                val removals = callgraph.calledBy.getValue(subroutine).map {
-                    IAstModification.Remove(it, it.definingScope())
-                }.toMutableList()
-                removals += IAstModification.Remove(subroutine, subroutine.definingScope())
-                return removals
-            }
-        }
 
-        if(subroutine !in callgraph.usedSymbols && !forceOutput) {
-            errors.warn("removing unused subroutine '${subroutine.name}'", subroutine.position)
-            return listOf(IAstModification.Remove(subroutine, subroutine.definingScope()))
-        }
-
-        return noModifications
-    }
-
-    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
-        val forceOutput = "force_output" in decl.definingBlock().options()
-        if(decl !in callgraph.usedSymbols && !forceOutput) {
-            if(decl.type == VarDeclType.VAR)
-                errors.warn("removing unused variable '${decl.name}'", decl.position)
-
-            return listOf(IAstModification.Remove(decl, decl.definingScope()))
-        }
-
-        return noModifications
-    }
 
     override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
-        if(functionCallStatement.target.nameInSource.size==1 && functionCallStatement.target.nameInSource[0] in BuiltinFunctions) {
+        if(functionCallStatement.target.targetStatement(program) is BuiltinFunctionPlaceholder) {
             val functionName = functionCallStatement.target.nameInSource[0]
-            if (functionName in pureBuiltinFunctions) {
+            if (functionName in functions.purefunctionNames) {
                 errors.warn("statement has no effect (function return value is discarded)", functionCallStatement.position)
-                return listOf(IAstModification.Remove(functionCallStatement, functionCallStatement.definingScope()))
+                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
-        // this is a C-64 specific optimization
-        if(functionCallStatement.target.nameInSource==listOf("c64scr", "print")) {
+        // 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?
-            stringVar = if(arg is AddressOf) {
+            val stringVar: IdentifierReference? = if(arg is AddressOf) {
                 arg.identifier
             } else {
                 arg as? IdentifierReference
             }
-            if(stringVar!=null) {
-                val vardecl = stringVar.targetVarDecl(program.namespace)!!
-                val string = vardecl.value as? StringLiteralValue
+            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 = CompilationTarget.instance.encodeString(string.value, string.altEncoding)[0]
+                        val firstCharEncoded = compTarget.encodeString(string.value, string.altEncoding)[0]
                         val chrout = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
-                                mutableListOf(NumericLiteralValue(DataType.UBYTE, firstCharEncoded.toInt(), pos)),
-                                functionCallStatement.void, pos
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstCharEncoded.toDouble(), pos)),
+                            functionCallStatement.void, pos
                         )
                         return listOf(IAstModification.ReplaceNode(functionCallStatement, chrout, parent))
                     } else if (string.value.length == 2) {
-                        val firstTwoCharsEncoded = CompilationTarget.instance.encodeString(string.value.take(2), string.altEncoding)
+                        val firstTwoCharsEncoded = compTarget.encodeString(string.value.take(2), string.altEncoding)
                         val chrout1 = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
-                                mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[0].toInt(), pos)),
-                                functionCallStatement.void, pos
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[0].toDouble(), pos)),
+                            functionCallStatement.void, pos
                         )
                         val chrout2 = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
-                                mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[1].toInt(), pos)),
-                                functionCallStatement.void, pos
+                            IdentifierReference(listOf("txt", "chrout"), pos),
+                            mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[1].toDouble(), pos)),
+                            functionCallStatement.void, pos
+                        )
+                        return listOf(
+                            IAstModification.InsertBefore(functionCallStatement, chrout1, parent as IStatementContainer),
+                            IAstModification.ReplaceNode(functionCallStatement, chrout2, parent)
                         )
-                        val anonscope = AnonymousScope(mutableListOf(), pos)
-                        anonscope.statements.add(chrout1)
-                        anonscope.statements.add(chrout2)
-                        return listOf(IAstModification.ReplaceNode(functionCallStatement, anonscope, parent))
                     }
                 }
             }
         }
 
         // if the first instruction in the called subroutine is a return statement, remove the jump altogeter
-        val subroutine = functionCallStatement.target.targetSubroutine(program.namespace)
+        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, functionCallStatement.definingScope()))
+                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.namespace)
-        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))
+        // see if we can optimize any complex arguments
+        // TODO for now, only works for single-argument functions because we use just 1 temp var: R9
+        if(functionCallStatement.target.nameInSource !in listOf(listOf("pop"), listOf("popw")) && functionCallStatement.args.size==1) {
+            val arg = functionCallStatement.args[0]
+            if(!arg.isSimple && arg !is TypecastExpression && arg !is IFunctionCall) {
+                val name = getTempVarName(arg.inferType(program))
+                val tempvar = IdentifierReference(name, functionCallStatement.position)
+                val assignTempvar = Assignment(AssignTarget(tempvar.copy(), null, null, functionCallStatement.position), arg, functionCallStatement.position)
+                return listOf(
+                    IAstModification.InsertBefore(functionCallStatement, assignTempvar, parent as IStatementContainer),
+                    IAstModification.ReplaceNode(arg, tempvar, functionCallStatement)
+                )
             }
         }
+
+
         return noModifications
     }
 
-    override fun after(ifStatement: IfStatement, parent: Node): Iterable<IAstModification> {
+    override fun after(ifElse: IfElse, parent: Node): Iterable<IAstModification> {
         // remove empty if statements
-        if(ifStatement.truepart.containsNoCodeNorVars() && ifStatement.elsepart.containsNoCodeNorVars())
-            return listOf(IAstModification.Remove(ifStatement, ifStatement.definingScope()))
+        if(ifElse.truepart.isEmpty() && ifElse.elsepart.isEmpty())
+            return listOf(IAstModification.Remove(ifElse, parent as IStatementContainer))
 
         // empty true part? switch with the else part
-        if(ifStatement.truepart.containsNoCodeNorVars() && ifStatement.elsepart.containsCodeOrVars()) {
-            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)
+        if(ifElse.truepart.isEmpty() && ifElse.elsepart.isNotEmpty()) {
+            val invertedCondition = PrefixExpression("not", ifElse.condition, ifElse.condition.position)
+            val emptyscope = AnonymousScope(mutableListOf(), ifElse.elsepart.position)
+            val truepart = AnonymousScope(ifElse.elsepart.statements, ifElse.truepart.position)
             return listOf(
-                    IAstModification.ReplaceNode(ifStatement.condition, invertedCondition, ifStatement),
-                    IAstModification.ReplaceNode(ifStatement.truepart, truepart, ifStatement),
-                    IAstModification.ReplaceNode(ifStatement.elsepart, emptyscope, ifStatement)
+                    IAstModification.ReplaceNode(ifElse.condition, invertedCondition, ifElse),
+                    IAstModification.ReplaceNode(ifElse.truepart, truepart, ifElse),
+                    IAstModification.ReplaceNode(ifElse.elsepart, emptyscope, ifElse)
             )
         }
 
-        val constvalue = ifStatement.condition.constValue(program)
+        val constvalue = ifElse.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))
+                errors.warn("condition is always true", ifElse.condition.position)
+                listOf(IAstModification.ReplaceNode(ifElse, ifElse.truepart, parent))
             } else {
                 // always false -> keep only else-part
-                errors.warn("condition is always false", ifStatement.position)
-                listOf(IAstModification.ReplaceNode(ifStatement, ifStatement.elsepart, parent))
+                errors.warn("condition is always false", ifElse.condition.position)
+                listOf(IAstModification.ReplaceNode(ifElse, ifElse.elsepart, parent))
             }
         }
 
@@ -181,20 +166,20 @@ internal class StatementOptimizer(private val program: Program,
     }
 
     override fun after(forLoop: ForLoop, parent: Node): Iterable<IAstModification> {
-        if(forLoop.body.containsNoCodeNorVars()) {
+        if(forLoop.body.isEmpty()) {
             errors.warn("removing empty for loop", forLoop.position)
-            return listOf(IAstModification.Remove(forLoop, forLoop.definingScope()))
+            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, forLoop.definingScope()))
+                return listOf(IAstModification.Remove(forLoop, parent as IStatementContainer))
             }
         }
 
         val range = forLoop.iterable as? RangeExpr
         if(range!=null) {
-            if(range.size()==1) {
+            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)
@@ -203,15 +188,15 @@ internal class StatementOptimizer(private val program: Program,
                 return listOf(IAstModification.ReplaceNode(forLoop, scope, parent))
             }
         }
-        val iterable = (forLoop.iterable as? IdentifierReference)?.targetVarDecl(program.namespace)
+        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 = CompilationTarget.instance.encodeString(sv.value, sv.altEncoding)[0]
-                    val byte = NumericLiteralValue(DataType.UBYTE, character, iterable.position)
+                    val character = compTarget.encodeString(sv.value, sv.altEncoding)[0]
+                    val byte = NumericLiteralValue(DataType.UBYTE, character.toDouble(), 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)
@@ -241,15 +226,14 @@ internal class StatementOptimizer(private val program: Program,
     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)
+            return if(constvalue.asBooleanValue) {
+                // always true -> keep only the statement block
                 errors.warn("condition is always true", untilLoop.condition.position)
-                if(!hasBreak(untilLoop.body))
-                    return listOf(IAstModification.ReplaceNode(untilLoop, untilLoop.body, parent))
+                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))
+                listOf(IAstModification.ReplaceNode(untilLoop, forever, parent))
             }
         }
         return noModifications
@@ -265,7 +249,7 @@ internal class StatementOptimizer(private val program: Program,
             } else {
                 // always false -> remove the while statement altogether
                 errors.warn("condition is always false", whileLoop.condition.position)
-                listOf(IAstModification.Remove(whileLoop, whileLoop.definingScope()))
+                listOf(IAstModification.Remove(whileLoop, parent as IStatementContainer))
             }
         }
         return noModifications
@@ -274,14 +258,14 @@ internal class StatementOptimizer(private val program: Program,
     override fun after(repeatLoop: RepeatLoop, parent: Node): Iterable<IAstModification> {
         val iter = repeatLoop.iterations
         if(iter!=null) {
-            if(repeatLoop.body.containsNoCodeNorVars()) {
+            if(repeatLoop.body.isEmpty()) {
                 errors.warn("empty loop removed", repeatLoop.position)
-                return listOf(IAstModification.Remove(repeatLoop, repeatLoop.definingScope()))
+                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, repeatLoop.definingScope()))
+                return listOf(IAstModification.Remove(repeatLoop, parent as IStatementContainer))
             }
             if (iterations == 1) {
                 errors.warn("iterations is always 1", iter.position)
@@ -291,25 +275,27 @@ internal class StatementOptimizer(private val program: Program,
         return noModifications
     }
 
-    override fun after(whenStatement: WhenStatement, parent: Node): Iterable<IAstModification> {
-        // remove empty choices
-        class ChoiceRemover(val choice: WhenChoice) : IAstModification {
-            override fun perform() {
-                whenStatement.choices.remove(choice)
-            }
-        }
-        return whenStatement.choices
-                .filter { !it.statements.containsCodeOrVars() }
-                .map { ChoiceRemover(it) }
-    }
-
     override fun after(jump: Jump, parent: Node): Iterable<IAstModification> {
         // if the jump is to the next statement, remove the jump
-        val scope = jump.definingScope()
-        val label = jump.identifier?.targetStatement(scope)
-        if(label!=null && scope.statements.indexOf(label) == scope.statements.indexOf(jump)+1)
-            return listOf(IAstModification.Remove(jump, jump.definingScope()))
+        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 after(gosub: GoSub, parent: Node): Iterable<IAstModification> {
+        // if the next statement is return with no returnvalue, change into a regular jump if there are no parameters as well.
+        val subroutineParams = gosub.identifier?.targetSubroutine(program)?.parameters
+        if(subroutineParams!=null && subroutineParams.isEmpty()) {
+            val returnstmt = gosub.nextSibling() as? Return
+            if(returnstmt!=null && returnstmt.value==null) {
+                return listOf(
+                    IAstModification.Remove(returnstmt, parent as IStatementContainer),
+                    IAstModification.ReplaceNode(gosub, Jump(gosub.address, gosub.identifier, gosub.generatedLabel, gosub.position), parent)
+                )
+            }
+        }
         return noModifications
     }
 
@@ -323,7 +309,7 @@ internal class StatementOptimizer(private val program: Program,
                     val op1 = binExpr.operator
                     val op2 = rExpr.operator
 
-                    if(rExpr.left is NumericLiteralValue && op2 in associativeOperators) {
+                    if(rExpr.left is NumericLiteralValue && op2 in AssociativeOperators) {
                         // associative operator, make sure the constant numeric value is second (right)
                         return listOf(IAstModification.SwapOperands(rExpr))
                     }
@@ -332,34 +318,34 @@ internal class StatementOptimizer(private val program: Program,
                     if(rNum!=null) {
                         if (op1 == "+" || op1 == "-") {
                             if (op2 == "+") {
-                                // A = A +/- B + N
+                                // A = A +/- B + N  --->  A = A +/- B  ;  A = A + 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.target.copy(),
+                                        BinaryExpression(binExpr.left.copy(), "+", rExpr.right, rExpr.position),
                                         assignment.position
                                 )
                                 return listOf(
                                         IAstModification.ReplaceNode(binExpr, expr2, binExpr.parent),
-                                        IAstModification.InsertAfter(assignment, addConstant, assignment.definingScope()))
+                                        IAstModification.InsertAfter(assignment, addConstant, parent as IStatementContainer))
                             } else if (op2 == "-") {
-                                // A = A +/- B - N
+                                // A = A +/- B - N  --->  A = A +/- B  ;  A = A - 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.target.copy(),
+                                        BinaryExpression(binExpr.left.copy(), "-", rExpr.right, rExpr.position),
                                         assignment.position
                                 )
                                 return listOf(
                                         IAstModification.ReplaceNode(binExpr, expr2, binExpr.parent),
-                                        IAstModification.InsertAfter(assignment, subConstant, assignment.definingScope()))
+                                        IAstModification.InsertAfter(assignment, subConstant, parent as IStatementContainer))
                             }
                         }
                     }
                 }
             }
 
-            if(binExpr.operator in associativeOperators && binExpr.right isSameAs assignment.target) {
+            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))
@@ -374,64 +360,76 @@ internal class StatementOptimizer(private val program: Program,
     override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
         if(assignment.target isSameAs assignment.value) {
             // remove assignment to self
-            return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
+            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")
+            return noModifications
 
         // optimize binary expressions a bit
-        val targetDt = targetIDt.typeOrElse(DataType.STRUCT)
         val bexpr=assignment.value as? BinaryExpression
         if(bexpr!=null) {
-            val rightCv = bexpr.right.constValue(program)?.number?.toDouble()
+            val rightCv = bexpr.right.constValue(program)?.number
+            if(bexpr.operator=="-" && rightCv==null) {
+                if(bexpr.right isSameAs assignment.target) {
+                    // X = value - X  -->  X = -X ; X += value  (to avoid need of stack-evaluation)
+                    val negation = PrefixExpression("-", bexpr.right.copy(), bexpr.position)
+                    val addValue = Assignment(assignment.target.copy(), BinaryExpression(bexpr.right, "+", bexpr.left, bexpr.position), assignment.position)
+                    return listOf(
+                        IAstModification.ReplaceNode(bexpr, negation, assignment),
+                        IAstModification.InsertAfter(assignment, addValue, parent as IStatementContainer)
+                    )
+                }
+            }
+
             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.namespace))?.type
+                val targetDt = targetIDt.getOr(DataType.UNDEFINED)
+                val vardeclDt = (assignment.target.identifier?.targetVarDecl(program))?.type
                 when (bexpr.operator) {
                     "+" -> {
                         if (rightCv == 0.0) {
-                            return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
+                            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, "++", assignment.position))
+                                    incs.statements.add(PostIncrDecr(assignment.target.copy(), "++", assignment.position))
                                 }
-                                return listOf(IAstModification.ReplaceNode(assignment, incs, parent))
+                                listOf(IAstModification.ReplaceNode(assignment, if(incs.statements.size==1) incs.statements[0] else incs, parent))
                             }
                         }
                     }
                     "-" -> {
                         if (rightCv == 0.0) {
-                            return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
+                            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, "--", assignment.position))
+                                    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, assignment.definingScope()))
-                    "/" -> if (rightCv == 1.0) return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
-                    "**" -> if (rightCv == 1.0) return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
-                    "|" -> if (rightCv == 0.0) return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
-                    "^" -> if (rightCv == 0.0) return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
+                    "*" -> 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, assignment.definingScope()))
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
                     }
                     ">>" -> {
                         if (rightCv == 0.0)
-                            return listOf(IAstModification.Remove(assignment, assignment.definingScope()))
+                            return listOf(IAstModification.Remove(assignment, parent as IStatementContainer))
                     }
                 }
 
@@ -442,64 +440,36 @@ internal class StatementOptimizer(private val program: Program,
     }
 
     override fun after(returnStmt: Return, parent: Node): Iterable<IAstModification> {
-        fun returnViaIntermediary(value: Expression): Iterable<IAstModification>? {
-            val returnDt = returnStmt.definingSubroutine()!!.returntypes.single()
+        fun returnViaIntermediaryVar(value: Expression): Iterable<IAstModification>? {
+            val subr = returnStmt.definingSubroutine!!
+            val returnDt = subr.returntypes.single()
             if (returnDt in IntegerDatatypes) {
-                // first assign to intermediary, then return that register
-                val returnValueIntermediary =
-                    when(returnDt) {
-                        DataType.UBYTE -> IdentifierReference(listOf("prog8_lib", "retval_interm_ub"), returnStmt.position)
-                        DataType.BYTE -> IdentifierReference(listOf("prog8_lib", "retval_interm_b"), returnStmt.position)
-                        DataType.UWORD -> IdentifierReference(listOf("prog8_lib", "retval_interm_uw"), returnStmt.position)
-                        DataType.WORD -> IdentifierReference(listOf("prog8_lib", "retval_interm_w"), returnStmt.position)
-                        else -> throw FatalAstException("weird return dt")
-                    }
+                // first assign to intermediary variable, then return that
+                val returnVarName = "retval_interm_" + when(returnDt) {
+                    DataType.UBYTE -> "ub"
+                    DataType.BYTE -> "b"
+                    DataType.UWORD -> "uw"
+                    DataType.WORD -> "w"
+                    else -> "<undefined>"
+                }
+                val returnValueIntermediary = IdentifierReference(listOf("prog8_lib", returnVarName), returnStmt.position)
                 val tgt = AssignTarget(returnValueIntermediary, null, null, returnStmt.position)
                 val assign = Assignment(tgt, value, returnStmt.position)
-                val returnReplacement = Return(returnValueIntermediary, returnStmt.position)
+                val returnReplacement = Return(returnValueIntermediary.copy(), returnStmt.position)
                 return listOf(
-                    IAstModification.InsertBefore(returnStmt, assign, parent as INameScope),
+                    IAstModification.InsertBefore(returnStmt, assign, parent as IStatementContainer),
                     IAstModification.ReplaceNode(returnStmt, returnReplacement, parent)
                 )
             }
             return null
         }
 
-        when(returnStmt.value) {
-            is PrefixExpression -> {
-                val mod = returnViaIntermediary(returnStmt.value!!)
-                if(mod!=null)
-                    return mod
-            }
-            is BinaryExpression -> {
-                val mod = returnViaIntermediary(returnStmt.value!!)
-                if(mod!=null)
-                    return mod
-            }
-            else -> {}
+        if(returnStmt.value is BinaryExpression) {
+            val mod = returnViaIntermediaryVar(returnStmt.value!!)
+            if(mod!=null)
+                return mod
         }
 
-        return super.after(returnStmt, parent)
+        return noModifications
     }
-
-    private fun hasBreak(scope: INameScope): 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 +1,247 @@
+package prog8.optimizer
+
+import prog8.ast.*
+import prog8.ast.base.DataType
+import prog8.ast.base.VarDeclType
+import prog8.ast.expressions.*
+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.isIOAddress
+
+
+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)
+        return emptyList()
+    }
+
+    override fun before(jump: Jump, parent: Node): Iterable<IAstModification> {
+        reportUnreachable(jump)
+        return emptyList()
+    }
+
+    override fun before(returnStmt: Return, parent: Node): Iterable<IAstModification> {
+        reportUnreachable(returnStmt)
+        return emptyList()
+    }
+
+    override fun before(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        if(functionCallStatement.target.nameInSource.last() == "exit")
+            reportUnreachable(functionCallStatement)
+        return emptyList()
+    }
+
+    private fun reportUnreachable(stmt: Statement) {
+        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> {
+        return deduplicateAssignments(scope.statements, 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 '${block.name}'", block.position)
+                return listOf(IAstModification.Remove(block, parent as IStatementContainer))
+            }
+            if(callgraph.unused(block)) {
+                if(block.statements.any{ it !is VarDecl || it.type==VarDeclType.VAR})
+                    errors.warn("removing unused block '${block.name}'", block.position)
+                return listOf(IAstModification.Remove(block, parent as IStatementContainer))
+            }
+        }
+
+        return deduplicateAssignments(block.statements, 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))
+            }
+        }
+
+        return deduplicateAssignments(subroutine.statements, 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) {
+                val usages = callgraph.usages(decl)
+                if (usages.isEmpty()) {
+                    errors.warn("removing unused variable '${decl.name}'", decl.position)
+                    return listOf(IAstModification.Remove(decl, parent as IStatementContainer))
+                } else {
+                    // if all usages are just an assignment to this vardecl,
+                    // and it is in regular RAM, then remove the var as well including all assignments
+                    val assignTargets = usages.mapNotNull {
+                        it.parent as? AssignTarget
+                    }.filter {
+                        !it.isIOAddress(compTarget.machine)
+                    }
+                    if(assignTargets.size==usages.size) {
+                        errors.warn("removing unused variable '${decl.name}'", decl.position)
+                        val assignmentsToRemove = assignTargets.map { it.parent to it.parent.parent as IStatementContainer}.toSet()
+                        return assignmentsToRemove.map {
+                            IAstModification.Remove(it.first, it.second)
+                        } + listOf(
+                            IAstModification.Remove(decl, parent as IStatementContainer)
+                        )
+                    }
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun deduplicateAssignments(statements: List<Statement>, scope: IStatementContainer): List<IAstModification> {
+        // removes 'duplicate' assignments that assign the same target directly after another
+        val linesToRemove = mutableListOf<Assignment>()
+        val modifications = mutableListOf<IAstModification>()
+
+        fun substituteZeroInBinexpr(expr: BinaryExpression, zero: NumericLiteralValue, assign1: Assignment, assign2: Assignment) {
+            if(expr.left isSameAs assign2.target) {
+                // X = X <oper> Right
+                linesToRemove.add(assign1)
+                modifications.add(IAstModification.ReplaceNode(
+                    expr.left, zero, expr
+                ))
+            }
+            if(expr.right isSameAs assign2.target) {
+                // X = Left <oper> X
+                linesToRemove.add(assign1)
+                modifications.add(IAstModification.ReplaceNode(
+                    expr.right, zero, expr
+                ))
+            }
+            val leftBinExpr = expr.left as? BinaryExpression
+            val rightBinExpr = expr.right as? BinaryExpression
+            if(leftBinExpr!=null && rightBinExpr==null) {
+                if(leftBinExpr.left isSameAs assign2.target) {
+                    // X = (X <oper> Right) <oper> Something
+                    linesToRemove.add(assign1)
+                    modifications.add(IAstModification.ReplaceNode(
+                        leftBinExpr.left, zero, leftBinExpr
+                    ))
+                }
+                if(leftBinExpr.right isSameAs assign2.target) {
+                    // X = (Left <oper> X) <oper> Something
+                    linesToRemove.add(assign1)
+                    modifications.add(IAstModification.ReplaceNode(
+                        leftBinExpr.right, zero, leftBinExpr
+                    ))
+                }
+            }
+            if(leftBinExpr==null && rightBinExpr!=null) {
+                if(rightBinExpr.left isSameAs assign2.target) {
+                    // X = Something <oper> (X <oper> Right)
+                    linesToRemove.add(assign1)
+                    modifications.add(IAstModification.ReplaceNode(
+                        rightBinExpr.left, zero, rightBinExpr
+                    ))
+                }
+                if(rightBinExpr.right isSameAs assign2.target) {
+                    // X = Something <oper> (Left <oper> X)
+                    linesToRemove.add(assign1)
+                    modifications.add(IAstModification.ReplaceNode(
+                        rightBinExpr.right, zero, rightBinExpr
+                    ))
+                }
+            }
+        }
+
+        fun substituteZeroInPrefixexpr(expr: PrefixExpression, zero: NumericLiteralValue, assign1: Assignment, assign2: Assignment) {
+            if(expr.expression isSameAs assign2.target) {
+                linesToRemove.add(assign1)
+                modifications.add(IAstModification.ReplaceNode(
+                    expr.expression, zero, expr
+                ))
+            }
+        }
+
+        fun substituteZeroInTypecast(expr: TypecastExpression, zero: NumericLiteralValue, assign1: Assignment, assign2: Assignment) {
+            if(expr.expression isSameAs assign2.target) {
+                linesToRemove.add(assign1)
+                modifications.add(IAstModification.ReplaceNode(
+                    expr.expression, zero, expr
+                ))
+            }
+            val subCast = expr.expression as? TypecastExpression
+            if(subCast!=null && subCast.expression isSameAs assign2.target) {
+                linesToRemove.add(assign1)
+                modifications.add(IAstModification.ReplaceNode(
+                    subCast.expression, zero, subCast
+                ))
+            }
+        }
+
+        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) {
+                val cvalue1 = assign1.value.constValue(program)
+                if(cvalue1!=null && cvalue1.number==0.0 && assign2.target.isSameAs(assign1.target, program) && assign2.isAugmentable) {
+                    val value2 = assign2.value
+                    val zero = VarDecl.defaultZero(value2.inferType(program).getOr(DataType.UNDEFINED), value2.position)
+                    when(value2) {
+                        is BinaryExpression -> substituteZeroInBinexpr(value2, zero, assign1, assign2)
+                        is PrefixExpression -> substituteZeroInPrefixexpr(value2, zero, assign1, assign2)
+                        is TypecastExpression -> substituteZeroInTypecast(value2, zero, assign1, assign2)
+                        else -> {}
+                    }
+                } else {
+                    if (assign1.target.isSameAs(assign2.target, program) && !assign1.target.isIOAddress(compTarget.machine))  {
+                        if(assign2.target.identifier==null || !assign2.value.referencesIdentifier(assign2.target.identifier!!.nameInSource))
+                            // only remove the second assignment if its value is a simple expression!
+                            when(assign2.value) {
+                                is PrefixExpression,
+                                is BinaryExpression,
+                                is TypecastExpression,
+                                is FunctionCallExpr -> { /* don't remove */ }
+                                else -> linesToRemove.add(assign1)
+                            }
+                    }
+                }
+            }
+        }
+
+        return modifications + linesToRemove.map { IAstModification.Remove(it, scope) }
+    }
+}

compiler/build.gradle

@@ -1,49 +1,55 @@
 plugins {
     id 'java'
     id 'application'
-    id "org.jetbrains.kotlin.jvm" version "1.4.21"
-    id 'org.jetbrains.dokka' version "0.9.18"
-    id 'com.github.johnrengelman.shadow' version '6.1.0'
+    id "org.jetbrains.kotlin.jvm"
+    id 'com.github.johnrengelman.shadow' version '7.1.0'
+    id "io.kotest" version "0.3.8"
 }
 
-targetCompatibility = 11
-sourceCompatibility = 11
-
-repositories {
-    mavenLocal()
-    mavenCentral()
-    maven { url "https://kotlin.bintray.com/kotlinx" }
-}
-
-def prog8version = rootProject.file('compiler/res/version.txt').text.trim()
-
-dependencies {
-    implementation project(':parser')
-    implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
-    // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation 'org.antlr:antlr4-runtime:4.8'
-    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.1'
-    // implementation 'net.razorvine:ksim65:1.8'
-    // implementation "com.github.hypfvieh:dbus-java:3.2.4"
-    implementation project(':parser')
-
-    testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
-    testImplementation 'org.junit.jupiter:junit-jupiter-api:5.3.2'
-    testImplementation 'org.hamcrest:hamcrest-junit:2.0.0.0'
-    testRuntimeOnly 'org.junit.jupiter:junit-jupiter-engine:5.3.2'
+java {
+    toolchain {
+        languageVersion = JavaLanguageVersion.of(javaVersion)
+    }
 }
 
 compileKotlin {
     kotlinOptions {
-        jvmTarget = "11"
-        // verbose = true
-        // freeCompilerArgs += "-XXLanguage:+NewInference"
+        jvmTarget = javaVersion
     }
 }
 
 compileTestKotlin {
     kotlinOptions {
-        jvmTarget = "11"
+        jvmTarget = 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.antlr:antlr4-runtime:4.9.2'
+    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 'io.kotest:kotest-runner-junit5-jvm:4.6.3'
+}
+
+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 }
     }
 }
 
@@ -58,7 +64,8 @@ sourceSets {
     }
     test {
         java {
-            srcDirs = ["${project.projectDir}/test"]
+            srcDir "${project.projectDir}/test"
+            srcDir "${project(':compilerAst').projectDir}/test/helpers"
         }
     }
 }
@@ -67,15 +74,9 @@ startScripts.enabled = true
 
 application {
     mainClass = 'prog8.CompilerMainKt'
-    mainClassName = 'prog8.CompilerMainKt'  // deprecated
     applicationName = 'p8compile'
 }
 
-artifacts {
-    archives shadowJar
-}
-
-
 shadowJar {
     archiveBaseName = 'prog8compiler'
     archiveVersion = prog8version
@@ -96,12 +97,4 @@ test {
     }
 }
 
-
-dokka {
-    outputFormat = 'html'
-    outputDirectory = "$buildDir/kdoc"
-}
-
-task wrapper(type: Wrapper) {
-    gradleVersion = '6.7'
-}
+build.finalizedBy installDist, installShadowDist

compiler/compiler.iml

@@ -8,12 +8,17 @@
       <sourceFolder url="file://$MODULE_DIR$/test" isTestSource="true" />
       <excludeFolder url="file://$MODULE_DIR$/build" />
     </content>
-    <orderEntry type="jdk" jdkName="11" jdkType="JavaSDK" />
+    <orderEntry type="inheritedJdk" />
     <orderEntry type="sourceFolder" forTests="false" />
     <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
-    <orderEntry type="module" module-name="parser" />
-    <orderEntry type="library" name="unittest-libs" level="project" />
-    <orderEntry type="library" name="kotlinx-cli-jvm" level="project" />
-    <orderEntry type="library" name="antlr-runtime-4.9" level="project" />
+    <orderEntry type="module" module-name="compilerAst" />
+    <orderEntry type="library" name="jetbrains.kotlinx.cli.jvm" 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" />
+    <orderEntry type="library" name="io.kotest.assertions.core.jvm" level="project" />
+    <orderEntry type="library" name="io.kotest.runner.junit5.jvm" level="project" />
+    <orderEntry type="library" name="antlr.antlr4" level="project" />
   </component>
 </module>

compiler/res/prog8lib/c128/floats.asm

@@ -0,0 +1,683 @@
+; --- low level floating point assembly routines for the C128
+;     these are almost all identical to the C64 except for a few details
+;     so we have to have a separate library file for the C128 unfortunately.
+
+
+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  $66
+		lda  $67
+		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  $66
+		sta  P8ESTACK_HI,x
+		lda  $67
+		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  ROMUPK		; 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/c128/floats.p8

@@ -0,0 +1,159 @@
+; Prog8 definitions for floating point handling on the Commodore 128
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+%option enable_floats
+
+floats {
+	; ---- this block contains C-128 compatible floating point related functions ----
+
+        const float  PI     = 3.141592653589793
+        const float  TWOPI  = 6.283185307179586
+
+        float  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 $af00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 102-103 ($66-$67) 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 $af03 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
+romsub $af09 = VAL(ubyte length @ A) clobbers(A,X,Y)       ; str -> fac1, $24/25 must point to string in bank1, A=string length   Don't call this from bank0
+
+; fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $16/17)
+; (tip: use GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
+romsub $af0c = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
+
+romsub $af12 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
+romsub $af15 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
+romsub $af18 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y in bank 1
+romsub $af1b = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
+romsub $af2a = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
+romsub $af1e = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
+romsub $af21 = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
+romsub $af5a = CONUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory bank 1  in A/Y into fac2
+romsub $af5d = ROMUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in current bank in A/Y into fac2
+romsub $af24 = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
+romsub $af27 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
+
+romsub $af63 = MOVFM(uword mflpt @ AY) clobbers(A,X,Y)      ; load mflpt value from memory  in A/Y into fac1
+romsub $af60 = MOVFRM() clobbers(A,X,Y)                     ; load mflpt value from memory  in $24/$25 into fac2
+romsub $af66 = MOVMF(uword mflpt @ XY) clobbers(A,X,Y)      ; store fac1 to memory  X/Y as 5-byte mflpt
+romsub $af69 = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+romsub $af6c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2
+romsub $af6c = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+romsub $af51 = SIGN() clobbers(X,Y) -> ubyte @ A            ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+romsub $af4e = ABS() clobbers(A,X,Y)                        ; fac1 = ABS(fac1)
+romsub $af54 = 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 $af2d = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
+romsub $af06 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY ($0100)
+romsub $af30 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+romsub $af36 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** mflpt from A/Y
+romsub $af39 = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+romsub $af33 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
+romsub $af3c = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+romsub $af57 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
+romsub $af3f = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+romsub $af42 = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+romsub $af45 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+romsub $af48 = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+
+
+asmsub  FREADSA  (byte value @A) clobbers(A,X,Y) {
+    ; ---- 8 bit signed A -> float in fac1
+    %asm {{
+        tay
+        bpl  +
+        lda  #$ff
+        jmp  GIVAYF
++       lda  #0
+        jmp  GIVAYF
+    }}
+}
+
+asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
+	%asm {{
+            stx  P8ZP_SCRATCH_REG
+            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  ROMUPK
+            jsr  FADDT
++           ldx  P8ZP_SCRATCH_REG
+            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  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 {{
+		stx  P8ZP_SCRATCH_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  P8ZP_SCRATCH_REG
+		rts
+	}}
+}
+
+%asminclude "library:c128/floats.asm"
+%asminclude "library:c64/floats_funcs.asm"
+
+}

compiler/res/prog8lib/c128/graphics.p8

@@ -0,0 +1,60 @@
+%import syslib
+%import textio
+
+; Bitmap pixel graphics module for the Commodore 128
+
+; TODO c128 actually implement the graphics routines. Ideally a way to 'borrow' the code form the C64 version without just copy-pasting that here?
+
+graphics {
+    const uword WIDTH = 320
+    const ubyte HEIGHT = 200
+
+    sub enable_bitmap_mode() {
+        ; enable bitmap screen, erase it and set colors to black/white.
+        ; TODO
+    }
+
+    sub disable_bitmap_mode() {
+        ; enables text mode, erase the text screen, color white
+        ; TODO
+    }
+
+
+    sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
+        ; TODO
+    }
+
+    sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
+        ; TODO
+    }
+
+    sub fillrect(uword x, uword y, uword width, uword height) {
+        ; TODO
+    }
+
+    sub rect(uword x, uword y, uword width, uword height) {
+        ; TODO
+    }
+
+    sub horizontal_line(uword x, uword y, uword length) {
+        ; TODO
+    }
+
+    sub vertical_line(uword x, uword y, uword height) {
+        ; TODO
+    }
+
+    sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; TODO
+    }
+
+    sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; TODO
+    }
+
+    inline asmsub  plot(uword plotx @R0, uword ploty @R1) clobbers(A, X, Y) {
+        %asm {{
+            nop     ; TODO
+        }}
+    }
+}

compiler/res/prog8lib/c128/syslib.p8

@@ -0,0 +1,796 @@
+; Prog8 definitions for the Commodore-128
+; Including memory registers, I/O registers, Basic and Kernal subroutines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+;
+
+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)     // TODO c128 ??
+
+        &ubyte  COLOR           = $00f1     ; cursor color
+        ;;&ubyte  HIBASE          = $0288     ; screen base address / 256 (hi-byte of screen memory address)        // TODO c128 ??
+        &uword  CINV            = $0314     ; IRQ vector (in ram)
+        &uword  CBINV           = $0316     ; BRK vector (in ram)
+        &uword  NMINV           = $0318     ; NMI vector (in ram)
+        &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 ----
+
+; ---- 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.home or txt.plot
+
+romsub $FA65 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
+romsub $FF33 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
+
+; TODO c128 a bunch of kernal routines are missing here that are specific to the c128
+
+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 compatible 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
+    }}
+}
+
+
+; ---- system utility routines that are essentially the same as on the C64: -----
+asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #$80
+        sta  247    ; disable charset switching
+        lda  #112
+        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
+	}}
+}
+
+}
+
+c128 {
+; ---- C128 specific registers ----
+
+    &ubyte  VM1     = $0A2C         ; shadow for VUC $d018 in text mode
+    &ubyte  VM2     = $0A2D         ; shadow for VIC $d018 in bitmap screen mode
+    &ubyte  VM3     = $0A2E         ; starting page for VDC screen mem
+    &ubyte  VM4     = $0A2F         ; starting page for VDC attribute mem
+
+
+; ---- C128 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                     ; TODO c128 ram and rom bank selection how?
+        ;;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  c64.disable_runstop_and_charsetswitch
+        clc
+        clv
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        rts     ; no phase 2 steps on the C128
+    }}
+}
+
+asmsub  disable_basic() clobbers(A) {
+    %asm {{
+        lda $0a04   ; disable BASIC shadow registers
+        and #$fe
+        sta $0a04
+
+        lda #$01    ; disable BASIC IRQ service routine
+        sta $12fd
+
+        lda #$ff    ; disable screen editor IRQ setup
+        sta $d8
+
+        lda #$b7    ; skip programmable function key check
+        sta $033c
+
+        lda #$0e    ; bank out BASIC ROM
+        sta $ff00
+        rts
+    }}
+}
+
+; ---- end of C128 specific system utility routines ----
+
+}
+
+sys {
+    ; ------- lowlevel system routines --------
+
+    const ubyte target = 128         ;  compilation target specifier.  64 = C64, 128 = C128,  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       TODO c128 how to do this?
+            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 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 C128 as well but their location in memory is different
+    ; (because there's no room for them in the zeropage)
+    ; $1300-$1bff is unused RAM on C128. We'll use $1a00-$1bff as the lo/hi evalstack.
+    ; the virtual registers are allocated at the bottom of the eval-stack (should be ample space unless
+    ; you're doing insane nesting of expressions...)
+    &uword r0  = $1b00
+    &uword r1  = $1b02
+    &uword r2  = $1b04
+    &uword r3  = $1b06
+    &uword r4  = $1b08
+    &uword r5  = $1b0a
+    &uword r6  = $1b0c
+    &uword r7  = $1b0e
+    &uword r8  = $1b10
+    &uword r9  = $1b12
+    &uword r10 = $1b14
+    &uword r11 = $1b16
+    &uword r12 = $1b18
+    &uword r13 = $1b1a
+    &uword r14 = $1b1c
+    &uword r15 = $1b1e
+
+    &word r0s  = $1b00
+    &word r1s  = $1b02
+    &word r2s  = $1b04
+    &word r3s  = $1b06
+    &word r4s  = $1b08
+    &word r5s  = $1b0a
+    &word r6s  = $1b0c
+    &word r7s  = $1b0e
+    &word r8s  = $1b10
+    &word r9s  = $1b12
+    &word r10s = $1b14
+    &word r11s = $1b16
+    &word r12s = $1b18
+    &word r13s = $1b1a
+    &word r14s = $1b1c
+    &word r15s = $1b1e
+
+    &ubyte r0L  = $1b00
+    &ubyte r1L  = $1b02
+    &ubyte r2L  = $1b04
+    &ubyte r3L  = $1b06
+    &ubyte r4L  = $1b08
+    &ubyte r5L  = $1b0a
+    &ubyte r6L  = $1b0c
+    &ubyte r7L  = $1b0e
+    &ubyte r8L  = $1b10
+    &ubyte r9L  = $1b12
+    &ubyte r10L = $1b14
+    &ubyte r11L = $1b16
+    &ubyte r12L = $1b18
+    &ubyte r13L = $1b1a
+    &ubyte r14L = $1b1c
+    &ubyte r15L = $1b1e
+
+    &ubyte r0H  = $1b01
+    &ubyte r1H  = $1b03
+    &ubyte r2H  = $1b05
+    &ubyte r3H  = $1b07
+    &ubyte r4H  = $1b09
+    &ubyte r5H  = $1b0b
+    &ubyte r6H  = $1b0d
+    &ubyte r7H  = $1b0f
+    &ubyte r8H  = $1b11
+    &ubyte r9H  = $1b13
+    &ubyte r10H = $1b15
+    &ubyte r11H = $1b17
+    &ubyte r12H = $1b19
+    &ubyte r13H = $1b1b
+    &ubyte r14H = $1b1d
+    &ubyte r15H = $1b1f
+
+    &byte r0sL  = $1b00
+    &byte r1sL  = $1b02
+    &byte r2sL  = $1b04
+    &byte r3sL  = $1b06
+    &byte r4sL  = $1b08
+    &byte r5sL  = $1b0a
+    &byte r6sL  = $1b0c
+    &byte r7sL  = $1b0e
+    &byte r8sL  = $1b10
+    &byte r9sL  = $1b12
+    &byte r10sL = $1b14
+    &byte r11sL = $1b16
+    &byte r12sL = $1b18
+    &byte r13sL = $1b1a
+    &byte r14sL = $1b1c
+    &byte r15sL = $1b1e
+
+    &byte r0sH  = $1b01
+    &byte r1sH  = $1b03
+    &byte r2sH  = $1b05
+    &byte r3sH  = $1b07
+    &byte r4sH  = $1b09
+    &byte r5sH  = $1b0b
+    &byte r6sH  = $1b0d
+    &byte r7sH  = $1b0f
+    &byte r8sH  = $1b11
+    &byte r9sH  = $1b13
+    &byte r10sH = $1b15
+    &byte r11sH = $1b17
+    &byte r12sH = $1b19
+    &byte r13sH = $1b1b
+    &byte r14sH = $1b1d
+    &byte r15sH = $1b1f
+}

compiler/res/prog8lib/c128/textio.p8

@@ -0,0 +1,619 @@
+; 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
+    c128.VM1 |= 2
+}
+
+sub uppercase() {
+    c64.VMCSB &= ~2
+    c128.VM1 &= ~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/c64/floats.asm

@@ -428,7 +428,9 @@ var_fac1_greater_f	.proc
 		cmp  #1
 		beq  +
 		lda  #0
-+		rts
+		rts
++		lda  #1
+		rts
 		.pend
 
 var_fac1_greatereq_f	.proc

compiler/res/prog8lib/c64/floats.p8

@@ -4,7 +4,6 @@
 ;
 ; indent format: TABS, size=8
 
-%target c64
 %option enable_floats
 
 floats {
@@ -13,6 +12,7 @@ floats {
         const float  PI     = 3.141592653589793
         const float  TWOPI  = 6.283185307179586
 
+        float  tempvar_swap_float     ; used for some swap() operations
 
 ; ---- C64 basic and kernal ROM float constants and functions ----
 
@@ -83,7 +83,7 @@ romsub $bc58 = ABS()                                        ; fac1 = ABS(fac1)
 romsub $bf71 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
 romsub $bf74 = SQRA() clobbers(A,X,Y)                       ; fac1 = SQRT(fac2)
 romsub $bfed = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
-romsub $bfb4 = NEGOP() clobbers(A)                          ; switch the sign of fac1
+romsub $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)
@@ -195,7 +195,7 @@ sub  print_f  (float value) {
 	}}
 }
 
-%asminclude "library:c64/floats.asm", ""
-%asminclude "library:c64/floats_funcs.asm", ""
+%asminclude "library:c64/floats.asm"
+%asminclude "library:c64/floats_funcs.asm"
 
 }

compiler/res/prog8lib/c64/graphics.p8

@@ -1,8 +1,7 @@
-%target c64
 %import textio
 
 ; bitmap pixel graphics module for the C64
-; only black/white monchrome 320x200 for now
+; only black/white monochrome 320x200 for now
 ; assumes bitmap screen memory is $2000-$3fff
 
 graphics {
@@ -34,35 +33,36 @@ graphics {
     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.
+        ; TODO implement this as optimized assembly, for instance https://github.com/EgonOlsen71/bresenham/blob/main/src/asm/graphics.asm  ??
+        ;      or from here https://retro64.altervista.org/blog/an-introduction-to-vector-based-graphics-the-commodore-64-rotating-simple-3d-objects/
+
         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-x1 as word
-        word @zp dy = y2-y1
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
 
         if dx==0 {
-            vertical_line(x1, y1, abs(dy)+1 as ubyte)
+            vertical_line(x1, y1, abs(dy) as ubyte +1)
             return
         }
         if dy==0 {
             if x1>x2
                 x1=x2
-            horizontal_line(x1, y1, abs(dx)+1 as uword)
+            horizontal_line(x1, y1, abs(dx) as uword +1)
             return
         }
 
-        ; TODO rewrite the rest in optimized assembly
-
         word @zp d = 0
         ubyte positive_ix = true
         if dx < 0 {
             dx = -dx
             positive_ix = false
         }
-        dx *= 2
-        dy *= 2
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
         internal_plotx = x1
 
         if dx >= dy {
@@ -72,10 +72,10 @@ graphics {
                     if internal_plotx==x2
                         return
                     internal_plotx++
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             } else {
@@ -84,10 +84,10 @@ graphics {
                     if internal_plotx==x2
                         return
                     internal_plotx--
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             }
@@ -99,10 +99,10 @@ graphics {
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         internal_plotx++
-                        d -= dy
+                        d -= dy2
                     }
                 }
             } else {
@@ -111,10 +111,10 @@ graphics {
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         internal_plotx--
-                        d -= dy
+                        d -= dy2
                     }
                 }
             }

compiler/res/prog8lib/c64/syslib.p8

@@ -5,19 +5,19 @@
 ;
 ; indent format: TABS, size=8
 
-%target c64
-
 c64 {
         &ubyte  TIME_HI         = $a0       ; software jiffy clock, hi byte
         &ubyte  TIME_MID        = $a1       ;  .. mid byte
         &ubyte  TIME_LO         = $a2       ;    .. lo byte. Updated by IRQ every 1/60 sec
-        &ubyte  STATUS          = $90       ; kernel status variable for I/O
+        &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  CINV            = $0314     ; IRQ vector (in ram)
+        &uword  CBINV           = $0316     ; BRK vector (in ram)
+        &uword  NMINV           = $0318     ; NMI vector (in ram)
         &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
@@ -178,7 +178,7 @@ c64 {
 
 ; ---- C64 ROM kernal routines ----
 
-romsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use c64scr.print instead)
+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
@@ -202,7 +202,7 @@ romsub $FFAE = UNLSN() clobbers(A)                              ; command serial
 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 address @ Y)   ; set logical file parameters
+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
@@ -211,10 +211,10 @@ romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320
 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, ubyte @ X, ubyte @ Y     ; (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 $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
-romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
+romsub $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
@@ -246,7 +246,7 @@ asmsub STOP2() -> ubyte @A  {
 }
 
 asmsub RDTIM16() -> uword @AY {
-    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
     %asm {{
         stx  P8ZP_SCRATCH_REG
         jsr  c64.RDTIM
@@ -294,6 +294,12 @@ asmsub  init_system()  {
     }}
 }
 
+asmsub  init_system_phase2()  {
+    %asm {{
+        rts     ; no phase 2 steps on the C64
+    }}
+}
+
 asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
     %asm {{
         lda  #$80
@@ -304,27 +310,13 @@ asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
     }}
 }
 
-asmsub  set_irqvec_excl() clobbers(A)  {
-	%asm {{
-		sei
-		lda  #<_irq_handler
-		sta  c64.CINV
-		lda  #>_irq_handler
-		sta  c64.CINV+1
-		cli
-		rts
-_irq_handler	jsr  set_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		lda  c64.CIA1ICR		; acknowledge CIA1 interrupt
-		jmp  c64.IRQDFEND		; end irq processing - don't call kernel
-	}}
-}
-
-asmsub  set_irqvec() clobbers(A)  {
+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
@@ -333,9 +325,23 @@ asmsub  set_irqvec() clobbers(A)  {
 		cli
 		rts
 _irq_handler    jsr  _irq_handler_init
-		jsr  irq.irq
+_modified	jsr  $ffff                      ; modified
 		jsr  _irq_handler_end
-		jmp  c64.IRQDFRT		; continue with normal kernel irq routine
+		lda  _use_kernal
+		bne  +
+		lda  #$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
@@ -388,7 +394,7 @@ IRQ_SCRATCH_ZPWORD2	.word  0
 		}}
 }
 
-asmsub  restore_irqvec() clobbers(A) {
+asmsub  restore_irq() clobbers(A) {
 	%asm {{
 		sei
 		lda  #<c64.IRQDFRT
@@ -404,8 +410,15 @@ asmsub  restore_irqvec() clobbers(A) {
 	}}
 }
 
-asmsub  set_rasterirq(uword rasterpos @ AY) clobbers(A) {
+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
@@ -416,12 +429,21 @@ asmsub  set_rasterirq(uword rasterpos @ AY) clobbers(A) {
 		rts
 
 _raster_irq_handler
-		jsr  set_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		jmp  c64.IRQDFRT
+		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
@@ -445,28 +467,6 @@ _setup_raster_irq
 	}}
 }
 
-asmsub  set_rasterirq_excl(uword rasterpos @ AY) clobbers(A) {
-	%asm {{
-		sei
-		jsr  set_rasterirq._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_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		jmp  c64.IRQDFEND		; end irq processing - don't call kernel
-
-	}}
-}
-
 ; ---- end of C64 specific system utility routines ----
 
 }
@@ -474,11 +474,11 @@ _raster_irq_handler
 sys {
     ; ------- lowlevel system routines --------
 
-    const ubyte target = 64         ;  compilation target specifier.  64 = C64,  16 = CommanderX16.
+    const ubyte target = 64         ;  compilation target specifier.  64 = C64, 128 = C128, 16 = CommanderX16.
 
 
     asmsub  reset_system()  {
-        ; Soft-reset the system back to Basic prompt.
+        ; Soft-reset the system back to initial power-on Basic prompt.
         %asm {{
             sei
             lda  #14
@@ -489,6 +489,7 @@ sys {
 
     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()) {
@@ -497,6 +498,27 @@ sys {
         }
     }
 
+    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
@@ -577,62 +599,34 @@ _longcopy
         }}
     }
 
-
-    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
+        php
+        pla
         }}
     }
 
     inline asmsub clear_carry() {
         %asm {{
-            clc
+        clc
         }}
     }
 
     inline asmsub set_carry() {
         %asm {{
-            sec
+        sec
         }}
     }
 
     inline asmsub clear_irqd() {
         %asm {{
-            cli
+        cli
         }}
     }
 
     inline asmsub set_irqd() {
         %asm {{
-            sei
+        sei
         }}
     }
 
@@ -679,4 +673,88 @@ cx16 {
     &uword r14 = $cf1c
     &uword r15 = $cf1e
 
+    &word r0s  = $cf00
+    &word r1s  = $cf02
+    &word r2s  = $cf04
+    &word r3s  = $cf06
+    &word r4s  = $cf08
+    &word r5s  = $cf0a
+    &word r6s  = $cf0c
+    &word r7s  = $cf0e
+    &word r8s  = $cf10
+    &word r9s  = $cf12
+    &word r10s = $cf14
+    &word r11s = $cf16
+    &word r12s = $cf18
+    &word r13s = $cf1a
+    &word r14s = $cf1c
+    &word r15s = $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
+
+    &byte r0sL  = $cf00
+    &byte r1sL  = $cf02
+    &byte r2sL  = $cf04
+    &byte r3sL  = $cf06
+    &byte r4sL  = $cf08
+    &byte r5sL  = $cf0a
+    &byte r6sL  = $cf0c
+    &byte r7sL  = $cf0e
+    &byte r8sL  = $cf10
+    &byte r9sL  = $cf12
+    &byte r10sL = $cf14
+    &byte r11sL = $cf16
+    &byte r12sL = $cf18
+    &byte r13sL = $cf1a
+    &byte r14sL = $cf1c
+    &byte r15sL = $cf1e
+
+    &byte r0sH  = $cf01
+    &byte r1sH  = $cf03
+    &byte r2sH  = $cf05
+    &byte r3sH  = $cf07
+    &byte r4sH  = $cf09
+    &byte r5sH  = $cf0b
+    &byte r6sH  = $cf0d
+    &byte r7sH  = $cf0f
+    &byte r8sH  = $cf11
+    &byte r9sH  = $cf13
+    &byte r10sH = $cf15
+    &byte r11sH = $cf17
+    &byte r12sH = $cf19
+    &byte r13sH = $cf1b
+    &byte r14sH = $cf1d
+    &byte r15sH = $cf1f
 }

compiler/res/prog8lib/c64/textio.p8

@@ -4,7 +4,6 @@
 ;
 ; indent format: TABS, size=8
 
-%target c64
 %import syslib
 %import conv
 
@@ -31,6 +30,16 @@ 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.
@@ -576,7 +585,7 @@ _colormod	sta  $ffff		; modified
 }
 
 asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernel routine, to save the X register.
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
 	%asm  {{
 		stx  P8ZP_SCRATCH_REG
 		tax

compiler/res/prog8lib/conv.p8

@@ -7,239 +7,213 @@ conv {
 
 ; ----- 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)
+    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 {{
-		ldy  #uword2decimal.ASCII_0_OFFSET
-		bne  uword2decimal.hex_try200
-		rts
+            stx  P8ZP_SCRATCH_REG
+            jsr  conv.ubyte2decimal
+            sty  string_out
+            sta  string_out+1
+            stx  string_out+2
+            lda  #0
+            sta  string_out+3
+            ldx  P8ZP_SCRATCH_REG
+            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)
+asmsub  str_ub  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in decimal string form, without left padding 0s
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
+		ldy  #0
+		sty  P8ZP_SCRATCH_B1
+		jsr  conv.ubyte2decimal
+_output_byte_digits
+                ; hundreds?
+		cpy  #'0'
+		beq  +
 		pha
-		and  #$0f
-		tax
-		ldy  _hex_digits,x
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		sta  string_out,y
 		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
+		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
+                ldx  P8ZP_SCRATCH_REG
+                rts
 	}}
 }
 
-asmsub  uword2hex  (uword value @AY) clobbers(A,Y)  {
-	; ---- convert 16 bit uword in A/Y into 4-character hexadecimal string 'uword2hex.output' (0-terminated)
+asmsub  str_b  (byte value @ A) clobbers(A,Y)  {
+	; ---- convert the byte in A in decimal string form, without left padding 0s
 	%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)
+            stx  P8ZP_SCRATCH_REG
+            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 {{
+	    stx  P8ZP_SCRATCH_REG
+	    jsr  conv.uword2decimal
+	    ldy  #0
+-           lda  conv.uword2decimal.decTenThousands,y
+            sta  string_out,y
+            beq  +
+            iny
+            bne  -
++           ldx  P8ZP_SCRATCH_REG
+	    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 {{
+	    stx  P8ZP_SCRATCH_REG
+	    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
+            ldx  P8ZP_SCRATCH_REG
+            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
+	    stx  P8ZP_SCRATCH_REG
+	    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
 	}}
 }
 
@@ -257,7 +231,7 @@ asmsub  any2uword(str string @AY) clobbers(Y) -> ubyte @A {
 	sta  P8ZP_SCRATCH_W1
 	sty  P8ZP_SCRATCH_W1+1
 	ldy  #0
-	lda  (P8ZP_SCRATCH_W1)
+	lda  (P8ZP_SCRATCH_W1),y
 	ldy  P8ZP_SCRATCH_W1+1
 	cmp  #'$'
 	beq  _hex
@@ -520,4 +494,243 @@ _stop
 	}}
 }
 
+
+; ----- 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 https://forums.nesdev.org/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/cx16diskio.p8

@@ -0,0 +1,37 @@
+; Cx16 specific disk drive I/O routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+%import diskio
+
+cx16diskio {
+
+    ; Use kernal LOAD routine to load the given program file in memory.
+    ; This is similar to Basic's  LOAD "filename",drive  /  LOAD "filename",drive,1
+    ; If you don't give an address_override, the location in memory is taken from the 2-byte file header.
+    ; If you specify a custom address_override, the first 2 bytes in the file are ignored
+    ; and the rest is loaded at the given location in memory.
+    ; Returns the end load address+1 if successful or 0 if a load error occurred.
+    ; You can use the load_size() function to calcuate the size of the file that was loaded.
+    sub load(ubyte drivenumber, uword filenameptr, ubyte bank, uword address_override) -> uword {
+        cx16.rambank(bank)
+        return diskio.load(drivenumber, filenameptr, address_override)
+    }
+
+    ; Use kernal LOAD routine to load the given file in memory.
+    ; INCLUDING the first 2 bytes in the file: no program header is assumed in the file.
+    ; This is different from Basic's LOAD instruction which always skips the first two bytes.
+    ; The load address is mandatory. Returns the number of bytes loaded.
+    ; Returns the end load address+1 if successful or 0 if a load error occurred.
+    ; You can use the load_size() function to calcuate the size of the file that was loaded.
+    sub load_raw(ubyte drivenumber, uword filenameptr, ubyte bank, uword address) -> uword {
+        cx16.rambank(bank)
+        return diskio.load_raw(drivenumber, filenameptr, address)
+    }
+
+    ; For use directly after a load or load_raw call (don't mess with the ram bank yet):
+    ; Calculates the number of bytes loaded (files > 64Kb ar truncated to 16 bits)
+    sub load_size(ubyte startbank, uword startaddress, uword endaddress) -> uword {
+        return $2000 * (cx16.getrambank() - startbank) + endaddress - startaddress
+    }
+}

compiler/res/prog8lib/cx16/floats.p8

@@ -4,15 +4,17 @@
 ;
 ; indent format: TABS, size=8
 
-%target cx16
 %option enable_floats
 
 floats {
-	; ---- this block contains C-64 floating point related functions ----
+	; ---- 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
 
+        float  tempvar_swap_float     ; used for some swap() operations
+
 
 ; ---- ROM float functions ----
 
@@ -43,46 +45,44 @@ romsub $fe1e = NORMAL() clobbers(A,X,Y)                     ; normalize fac1 (?)
 romsub $fe24 = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
 romsub $fe27 = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
 romsub $fe2a = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
-romsub $fe33 = CONUPK(uword mflpt @ AY) clobbers(A,Y)       ; load mflpt value from memory  in A/Y into fac2
-romsub $fe36 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
-romsub $fe3c = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive!
-romsub $fe3f = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1  (remainder in fac2)
-romsub $fe42 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  (remainder in fac2)  mind the order of the operands
+romsub $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 $fe48 = MOVFM(uword mflpt @ AY) clobbers(A,Y)        ; load mflpt value from memory  in A/Y into fac1
-romsub $fe4b = MOVMF(uword mflpt @ XY) clobbers(A,Y)        ; store fac1 to memory  X/Y as 5-byte mflpt
-romsub $fe4e = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
-romsub $fe51 = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded)
-romsub $fe54 = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
-romsub $fe5a = SIGN() -> ubyte @ A                          ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
-romsub $fe5d = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
-romsub $fe60 = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
-romsub $fe6c = ABS()                                        ; fac1 = ABS(fac1)
-romsub $fe6f = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
-romsub $fe78 = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to round instead of trunc
-romsub $fe7e = FINLOG(byte value @A) clobbers (A, X, Y)           ; fac1 += signed byte in A
-romsub $fe81 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
-romsub $fe8a = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
-romsub $fe8d = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
-; note: there is no FPWR() on the Cx16
-romsub $fe93 = NEGOP() clobbers(A)                          ; switch the sign of fac1
-romsub $fe96 = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
-romsub $fe9f = RND2(byte value @A) clobbers(A,X,Y)                ; fac1 = RND(A) float random number generator
-romsub $fea2 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
-romsub $fea5 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
-romsub $fea8 = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
-romsub $feab = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
-romsub $feae = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+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  P8ZP_SCRATCH_W2
+        sta  _tmp
         sty  P8ZP_SCRATCH_B1
         tya
-        ldy  P8ZP_SCRATCH_W2
+        ldy  _tmp
         jsr  GIVAYF                 ; load it as signed... correct afterwards
         lda  P8ZP_SCRATCH_B1
 	    bpl  +
@@ -91,6 +91,7 @@ asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	    jsr  FADD
 +	    plx
         rts
+_tmp        .byte 0
 _flt65536    .byte 145,0,0,0,0       ; 65536.0
 	}}
 }
@@ -128,6 +129,14 @@ asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
 	}}
 }
 
+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 {{
@@ -149,7 +158,7 @@ sub  print_f  (float value) {
 	}}
 }
 
-%asminclude "library:c64/floats.asm", ""
-%asminclude "library:c64/floats_funcs.asm", ""
+%asminclude "library:c64/floats.asm"
+%asminclude "library:c64/floats_funcs.asm"
 
 }

compiler/res/prog8lib/cx16/gfx2.p8

@@ -1,33 +1,41 @@
-%target cx16
-
-; Bitmap pixel graphics module for the CommanderX16
+; Bitmap pixel graphics routines for the CommanderX16
 ; Custom routines to use the full-screen 640x480 and 320x240 screen modes.
-; This only works on the Cx16. No text layer is currently shown, text can be drawn as part of the bitmap itself.
-; Note: for compatible graphics code that words on C64 too, use the "graphics" module instead.
-; Note: there is no color palette manipulation here, you have to do that yourself or use the "palette" module.
+; (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 = 255
+    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) {
-        ; mode 0 = bitmap 320 x 240 x 1c monochrome
-        ; mode 1 = bitmap 320 x 240 x 256c
-        ; mode 128 = bitmap 640 x 480 x 1c monochrome
-        ; ...other modes?
-
-        ; copy the lower-case charset to the upper part of the vram, so we can use it later to plot text
-
         when mode {
-            0 -> {
-                ; 320 x 240 x 1c
+            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
@@ -38,8 +46,9 @@ gfx2 {
                 height = 240
                 bpp = 1
             }
-            1 -> {
-                ; 320 x 240 x 256c
+            ; 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
@@ -50,8 +59,8 @@ gfx2 {
                 height = 240
                 bpp = 8
             }
-            128 -> {
-                ; 640 x 480 x 1c
+            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
@@ -62,13 +71,26 @@ gfx2 {
                 height = 480
                 bpp = 1
             }
-            255 -> {
+            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
             }
         }
 
@@ -81,21 +103,28 @@ gfx2 {
         monochrome_stipple(false)
         position(0, 0)
         when active_mode {
-            0 -> {
-                ; 320 x 240 x 1c
+            1 -> {
+                ; lores monochrome
                 repeat 240/2/8
                     cs_innerloop640()
             }
-            1 -> {
-                ; 320 x 240 x 256c
+            ; TODO mode 2, 3
+            4 -> {
+                ; lores 256c
                 repeat 240/2
                     cs_innerloop640()
             }
-            128 -> {
-                ; 640 x 480 x 1c
+            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)
     }
@@ -130,35 +159,13 @@ gfx2 {
         if length==0
             return
         when active_mode {
-            1 -> {
-                ; 8bpp mode
-                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
-                }}
-            }
-            0, 128 -> {
-                ; 1 bpp 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 {
-                    ; this could be  optimized by setting this byte in 1 go but probably not worth it due to code size
+                    ; TODO optimize this by writing a masked byte in 1 go
                     plot(x, y, color)
                     x++
                 }
@@ -200,116 +207,188 @@ _loop                   lda  length
 _done
                     }}
                     repeat separate_pixels {
-                        ; this could be  optimized by setting this byte in 1 go but probably not worth it due to code size
+                        ; TODO optimize this by writing a masked byte in 1 go
                         plot(x, y, color)
                         x++
                     }
                 }
-                cx16.VERA_ADDR_H = (cx16.VERA_ADDR_H & %00000111)   ; vera auto-increment off again
+                cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off again
+            }
+            4 -> {
+                ; lores 256c
+                position(x, y)
+                %asm {{
+                    lda  color
+                    phx
+                    ldx  length+1
+                    beq  +
+                    ldy  #0
+-                   sta  cx16.VERA_DATA0
+                    iny
+                    bne  -
+                    dex
+                    bne  -
++                   ldy  length     ; remaining
+                    beq  +
+-                   sta  cx16.VERA_DATA0
+                    dey
+                    bne  -
++                   plx
+                }}
+            }
+            6 -> {
+                ; highres 4c
+                ; TODO also mostly usable for lores 4c?
+                color &= 3
+                ubyte[4] colorbits
+                ubyte ii
+                for ii in 3 downto 0 {
+                    colorbits[ii] = color
+                    color <<= 2
+                }
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                %asm {{
+                    lda  cx16.VERA_ADDR_H
+                    and  #%00000111         ; no auto advance
+                    sta  cx16.VERA_ADDR_H
+                    stz  cx16.VERA_CTRL     ; setup vera addr 0
+                    lda  cx16.r1
+                    and  #1
+                    sta  cx16.VERA_ADDR_H
+                    lda  cx16.r0
+                    sta  cx16.VERA_ADDR_L
+                    lda  cx16.r0+1
+                    sta  cx16.VERA_ADDR_M
+                    phx
+                    ldx  x
+                }}
+
+                repeat length {
+                    %asm {{
+                        txa
+                        and  #3
+                        tay
+                        lda  cx16.VERA_DATA0
+                        and  gfx2.plot.mask4c,y
+                        ora  colorbits,y
+                        sta  cx16.VERA_DATA0
+                        cpy  #%00000011         ; next vera byte?
+                        bne  ++
+                        inc  cx16.VERA_ADDR_L
+                        bne  ++
+                        inc  cx16.VERA_ADDR_M
++                       bne  +
+                        inc  cx16.VERA_ADDR_H
++                       inx                     ; next pixel
+                    }}
+                }
+
+                %asm {{
+                    plx
+                }}
             }
         }
     }
 
     sub vertical_line(uword x, uword y, uword height, ubyte color) {
-        position(x,y)
-        if active_mode==1 {
-            ; set vera auto-increment to 320 pixel increment (=next line)
-            cx16.VERA_ADDR_H = (cx16.VERA_ADDR_H & %00000111) | (14<<4)
-            %asm {{
-                ldy  height
-                beq  +
-                lda  color
--               sta  cx16.VERA_DATA0
-                dey
-                bne  -
-+
-            }}
-            return
-        }
-
-        ; note for the 1 bpp modes we can't use vera's auto increment mode because we have to 'or' the pixel data in place.
-        cx16.VERA_ADDR_H = (cx16.VERA_ADDR_H & %00000111)   ; no auto advance
-        cx16.r15 = gfx2.plot.bits[x as ubyte & 7]           ; bitmask
-        if active_mode>=128
-            cx16.r14 = 640/8
-        else
-            cx16.r14 = 320/8
-        if color {
-            if monochrome_dont_stipple_flag {
-                repeat height {
-                    %asm {{
-                        lda  cx16.VERA_DATA0
-                        ora  cx16.r15
-                        sta  cx16.VERA_DATA0
-                        lda  cx16.VERA_ADDR_L
-                        clc
-                        adc  cx16.r14                 ; advance vera ptr to go to the next line
-                        sta  cx16.VERA_ADDR_L
-                        lda  cx16.VERA_ADDR_M
-                        adc  #0
-                        sta  cx16.VERA_ADDR_M
-                        ; lda  cx16.VERA_ADDR_H     ; the bitmap size is small enough to not have to deal with the _H part.
-                        ; adc  #0
-                        ; sta  cx16.VERA_ADDR_H
-                    }}
+        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
+                        }}
+                    }
                 }
-            } else {
-                ; stippling.
-                height = (height+1)/2
+            }
+            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 {{
-                    lda x
-                    eor y
-                    and #1
-                    bne +
-                    lda  cx16.VERA_ADDR_L
-                    clc
-                    adc  cx16.r14                ; advance vera ptr to go to the next line for correct stipple pattern
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.VERA_ADDR_M
-                    adc  #0
-                    sta  cx16.VERA_ADDR_M
-+
-                    asl  cx16.r14
                     ldy  height
                     beq  +
--                   lda  cx16.VERA_DATA0
-                    ora  cx16.r15
-                    sta  cx16.VERA_DATA0
-                    lda  cx16.VERA_ADDR_L
-                    clc
-                    adc  cx16.r14               ; advance vera data ptr to go to the next-next line
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.VERA_ADDR_M
-                    adc  #0
-                    sta  cx16.VERA_ADDR_M
-                    ; lda  cx16.VERA_ADDR_H      ; the bitmap size is small enough to not have to deal with the _H part.
-                    ; adc  #0
-                    ; sta  cx16.VERA_ADDR_H
+                    lda  color
+-                   sta  cx16.VERA_DATA0
                     dey
                     bne  -
 +
                 }}
             }
-        } else {
-            cx16.r15 = ~cx16.r15
-            repeat height {
-                %asm {{
-                    lda  cx16.VERA_DATA0
-                    and  cx16.r15
-                    sta  cx16.VERA_DATA0
-                    lda  cx16.VERA_ADDR_L
-                    clc
-                    adc  cx16.r14             ; advance vera data ptr to go to the next line
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.VERA_ADDR_M
-                    adc  #0
-                    sta  cx16.VERA_ADDR_M
-                    ; lda  cx16.VERA_ADDR_H      ; the bitmap size is small enough to not have to deal with the _H part.
-                    ; adc  #0
-                    ; sta  cx16.VERA_ADDR_H
-                }}
+            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) {
@@ -320,42 +399,41 @@ _done
             swap(x1, x2)
             swap(y1, y2)
         }
-        word @zp dx = x2-x1 as word
-        word @zp dy = y2-y1 as word
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
 
         if dx==0 {
-            vertical_line(x1, y1, abs(dy)+1 as uword, color)
+            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)+1 as uword, color)
+            horizontal_line(x1, y1, abs(dx) as uword +1, color)
             return
         }
 
-        ; TODO rewrite the rest in optimized assembly (or reuse GRAPH_draw_line if we can get the FB replacement vector layer working)
         word @zp d = 0
-        ubyte positive_ix = true
+        cx16.r13 = true      ; 'positive_ix'
         if dx < 0 {
             dx = -dx
-            positive_ix = false
+            cx16.r13 = false
         }
-        dx *= 2
-        dy *= 2
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
         cx16.r14 = x1       ; internal plot X
 
         if dx >= dy {
-            if positive_ix {
+            if cx16.r13 {
                 repeat {
                     plot(cx16.r14, y1, color)
                     if cx16.r14==x2
                         return
                     cx16.r14++
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             } else {
@@ -364,25 +442,25 @@ _done
                     if cx16.r14==x2
                         return
                     cx16.r14--
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             }
         }
         else {
-            if positive_ix {
+            if cx16.r13 {
                 repeat {
                     plot(cx16.r14, y1, color)
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         cx16.r14++
-                        d -= dy
+                        d -= dy2
                     }
                 }
             } else {
@@ -391,10 +469,10 @@ _done
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         cx16.r14--
-                        d -= dy
+                        d -= dy2
                     }
                 }
             }
@@ -468,79 +546,160 @@ _done
 
     sub plot(uword @zp x, uword y, ubyte color) {
         ubyte[8] bits = [128, 64, 32, 16, 8, 4, 2, 1]
-        uword addr
-        ubyte value
+        ubyte[4] mask4c = [%00111111, %11001111, %11110011, %11111100]
+        ubyte[4] shift4c = [6,4,2,0]
 
         when active_mode {
-            0 -> {
-                %asm {{
-                    lda  x
-                    eor  y
-                    ora  monochrome_dont_stipple_flag
-                    and  #1
-                }}
-                if_nz {
-                    addr = x/8 + y*(320/8)
-                    value = bits[lsb(x)&7]
-                    if color
-                        cx16.vpoke_or(0, addr, value)
-                    else {
-                        value = ~value
-                        cx16.vpoke_and(0, addr, value)
-                    }
-                }
-            }
-            128 -> {
-                %asm {{
-                    lda  x
-                    eor  y
-                    ora  monochrome_dont_stipple_flag
-                    and  #1
-                }}
-                if_nz {
-                    addr = x/8 + y*(640/8)
-                    value = bits[lsb(x)&7]
-                    if color
-                        cx16.vpoke_or(0, addr, value)
-                    else {
-                        value = ~value
-                        cx16.vpoke_and(0, addr, value)
-                    }
-                }
-            }
             1 -> {
-                void addr_mul_320_add_24(y, x)      ; 24 bits result is in r0 and r1L
-                value = lsb(cx16.r1)
-                cx16.vpoke(value, cx16.r0, color)
-                ; activate vera auto-increment mode so next_pixel() can be used after this
-                cx16.VERA_ADDR_H = (cx16.VERA_ADDR_H & %00000111) | %00010000
-                color = cx16.VERA_DATA0
+                ; 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 {
-            0 -> {
+            1 -> {
+                ; lores monochrome
                 cx16.r0 = y*(320/8) + x/8
                 cx16.vaddr(0, cx16.r0, 0, 1)
             }
-            128 -> {
+            ; 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)
             }
-            1 -> {
-                void addr_mul_320_add_24(y, x)      ; 24 bits result is in r0 and r1L
-                ubyte bank = lsb(cx16.r1)
+            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
         }}
@@ -550,6 +709,7 @@ _done
         ; -- 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
@@ -614,13 +774,14 @@ _done
     sub text(uword @zp x, uword y, ubyte color, uword sctextptr) {
         ; -- Write some text at the given pixel position. The text string must be in screencode encoding (not petscii!).
         ;    You must also have called text_charset() first to select and prepare the character set to use.
-        ;    NOTE: in monochrome (1bpp) screen modes, x position is currently constrained to mulitples of 8 !
+        ;    NOTE: in monochrome (1bpp) screen modes, x position is currently constrained to multiples of 8 !  TODO allow per-pixel horizontal positioning
+        ; TODO draw whole horizontal spans using vera auto increment if possible, instead of per-character columns
         uword chardataptr
         when active_mode {
-            0, 128 -> {
-                ; 1-bitplane modes
+            1, 5 -> {
+                ; monochrome mode, either resolution
                 cx16.r2 = 40
-                if active_mode>=128
+                if active_mode==5
                     cx16.r2 = 80
                 while @(sctextptr) {
                     chardataptr = charset_addr + (@(sctextptr) as uword)*8
@@ -648,11 +809,8 @@ _done
                         sta  cx16.VERA_ADDR_L
                         bcc  +
                         inc  cx16.VERA_ADDR_M
-+                       lda  x
-                        clc
-                        adc  #1
-                        sta  x
-                        bcc  +
++                       inc  x
+                        bne  +
                         inc  x+1
 +                       dey
                         bne  -
@@ -660,8 +818,8 @@ _done
                     sctextptr++
                 }
             }
-            1 -> {
-                ; 320 x 240 x 256c
+            4 -> {
+                ; lores 256c
                 while @(sctextptr) {
                     chardataptr = charset_addr + (@(sctextptr) as uword)*8
                     cx16.vaddr(charset_bank, chardataptr, 1, 1)
@@ -689,6 +847,30 @@ _done
                     sctextptr++
                 }
             }
+            6 -> {
+                ; hires 4c
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    repeat 8 {
+                        ; TODO rewrite this inner loop partly in assembly
+                        ;      requires expanding the charbits to 2-bits per pixel (based on color)
+                        ;      also it's way more efficient to draw whole horizontal spans instead of per-character
+                        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++
+                }
+            }
         }
     }
 
@@ -709,48 +891,98 @@ _done
         }}
     }
 
-    asmsub addr_mul_320_add_24(uword address @R0, uword value @AY) clobbers(A) -> uword @R0, ubyte @R1  {
-            %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
-            }}
-        }
+    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

@@ -1,10 +1,9 @@
-%target cx16
 %import syslib
 %import textio
 
 ; Bitmap pixel graphics module for the CommanderX16
 ; wraps the graphics functions that are in ROM.
-; only black/white monchrome 320x200 for now. (i.e. truncated at the bottom)
+; 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.
 

compiler/res/prog8lib/cx16/palette.p8

@@ -1,5 +1,3 @@
-%target cx16
-
 ; Manipulate the Commander X16's display color palette.
 ; Should you want to restore the default palette, you have to reinitialize the Vera yourself.
 
@@ -9,34 +7,46 @@ palette {
     ubyte c
 
     sub set_color(ubyte index, uword color) {
-        cx16.vpoke(1, $fa00+index*2, lsb(color))
-        cx16.vpoke(1, $fa01+index*2, msb(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 palletteptr, uword num_colors) {
+    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, @(palletteptr))
-            palletteptr++
-            cx16.vpoke(1, vera_palette_ptr, @(palletteptr))
-            palletteptr++
+            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_rgb8(uword palletteptr, uword num_colors) {
+    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 = @(palletteptr) >> 4
-            palletteptr++
-            greenblue = @(palletteptr) & %11110000
-            palletteptr++
-            greenblue |= @(palletteptr) >> 4    ; add Blue
-            palletteptr++
+            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)
@@ -44,20 +54,28 @@ palette {
         }
     }
 
-    sub set_monochrome() {
+    sub set_monochrome(uword screencolorRGB, uword drawcolorRGB) {
         vera_palette_ptr = $fa00
-        cx16.vpoke(1, vera_palette_ptr, 0)
+        cx16.vpoke(1, vera_palette_ptr, lsb(screencolorRGB))   ; G,B
         vera_palette_ptr++
-        cx16.vpoke(1, vera_palette_ptr, 0)
+        cx16.vpoke(1, vera_palette_ptr, msb(screencolorRGB))   ; R
         vera_palette_ptr++
         repeat 255 {
-            cx16.vpoke(1, vera_palette_ptr, 255)
+            cx16.vpoke(1, vera_palette_ptr, lsb(drawcolorRGB)) ; G,B
             vera_palette_ptr++
-            cx16.vpoke(1, vera_palette_ptr, 255)
+            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 {
@@ -88,7 +106,7 @@ palette {
         $666,  ; 12 = medium grey
         $9D8,  ; 13 = light green
         $65B,  ; 14 = light blue
-        $999  ; 15 = light grey
+        $999   ; 15 = light grey
     ]
 
     uword[] C64_colorpalette_pepto = [  ; # this is Pepto's Commodore-64 palette  http://www.pepto.de/projects/colorvic/
@@ -107,7 +125,7 @@ palette {
         $777,  ; 12 = medium grey
         $af9,  ; 13 = light green
         $76e,  ; 14 = light blue
-        $bbb  ; 15 = light grey
+        $bbb   ; 15 = light grey
     ]
 
     uword[] C64_colorpalette_light = [  ; this is a lighter palette

compiler/res/prog8lib/cx16/syslib.p8

@@ -5,16 +5,13 @@
 ;
 ; indent format: TABS, size=8
 
-%target cx16
-
-
 c64 {
 
 ; ---- kernal routines, these are the same as on the Commodore-64 (hence the same block name) ----
 
 ; STROUT --> use txt.print
 ; CLEARSCR -> use txt.clear_screen
-; HOMECRSR -> use txt.plot
+; HOMECRSR -> use txt.home or 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)
@@ -24,7 +21,7 @@ romsub $FF8D = VECTOR(uword userptr @ XY, ubyte dir @ Pc) clobbers(A,Y)     ; re
 romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
 romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
 romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
-romsub $FF99 = MEMTOP(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer.   NOTE: as a Cx16 extension, also returns the number of RAM memory banks in register A !  See MEMTOP2
+romsub $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
@@ -35,7 +32,7 @@ romsub $FFAE = UNLSN() clobbers(A)                              ; command serial
 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 address @ Y)   ; set logical file parameters
+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
@@ -44,10 +41,10 @@ romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320
 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, ubyte @ X, ubyte @ Y     ; (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 $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
-romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
+romsub $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
@@ -74,7 +71,7 @@ asmsub STOP2() -> ubyte @A  {
 }
 
 asmsub RDTIM16() -> uword @AY {
-    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
     %asm {{
         phx
         jsr  c64.RDTIM
@@ -87,28 +84,20 @@ asmsub RDTIM16() -> uword @AY {
     }}
 }
 
-asmsub MEMTOP2() -> ubyte @A {
-    ; -- uses MEMTOP's cx16 extension to query the number of available RAM banks.
-    %asm {{
-        phx
-        sec
-        jsr  c64.MEMTOP
-        plx
-        rts
-    }}
-}
-
 }
 
 cx16 {
 
-; 65c02 hardware vectors:
-    &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
+; irq and hardware vectors:
+    &uword  CINV            = $0314     ; IRQ vector (in ram)
+    &uword  CBINV           = $0316     ; BRK vector (in ram)
+    &uword  NMINV           = $0318     ; NMI 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
+; the sixteen virtual 16-bit registers in both normal unsigned mode and signed mode (s)
     &uword r0  = $0002
     &uword r1  = $0004
     &uword r2  = $0006
@@ -126,6 +115,91 @@ cx16 {
     &uword r14 = $001e
     &uword r15 = $0020
 
+    &word r0s  = $0002
+    &word r1s  = $0004
+    &word r2s  = $0006
+    &word r3s  = $0008
+    &word r4s  = $000a
+    &word r5s  = $000c
+    &word r6s  = $000e
+    &word r7s  = $0010
+    &word r8s  = $0012
+    &word r9s  = $0014
+    &word r10s = $0016
+    &word r11s = $0018
+    &word r12s = $001a
+    &word r13s = $001c
+    &word r14s = $001e
+    &word r15s = $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
+
+    &byte r0sL  = $0002
+    &byte r1sL  = $0004
+    &byte r2sL  = $0006
+    &byte r3sL  = $0008
+    &byte r4sL  = $000a
+    &byte r5sL  = $000c
+    &byte r6sL  = $000e
+    &byte r7sL  = $0010
+    &byte r8sL  = $0012
+    &byte r9sL  = $0014
+    &byte r10sL = $0016
+    &byte r11sL = $0018
+    &byte r12sL = $001a
+    &byte r13sL = $001c
+    &byte r14sL = $001e
+    &byte r15sL = $0020
+
+    &byte r0sH  = $0003
+    &byte r1sH  = $0005
+    &byte r2sH  = $0007
+    &byte r3sH  = $0009
+    &byte r4sH  = $000b
+    &byte r5sH  = $000d
+    &byte r6sH  = $000f
+    &byte r7sH  = $0011
+    &byte r8sH  = $0013
+    &byte r9sH  = $0015
+    &byte r10sH = $0017
+    &byte r11sH = $0019
+    &byte r12sH = $001b
+    &byte r13sH = $001d
+    &byte r14sH = $001f
+    &byte r15sH = $0021
+
 ; VERA registers
 
     const uword VERA_BASE       = $9F20
@@ -171,7 +245,7 @@ cx16 {
 
 ; I/O
 
-    const uword  via1   = $9f60                  ;VIA 6522 #1
+    const uword  via1   = $9f00                  ;VIA 6522 #1
     &ubyte  d1prb	= via1+0
     &ubyte  d1pra	= via1+1
     &ubyte  d1ddrb	= via1+2
@@ -189,7 +263,7 @@ cx16 {
     &ubyte  d1ier	= via1+14
     &ubyte  d1ora	= via1+15
 
-    const uword  via2   = $9f70                  ;VIA 6522 #2
+    const uword  via2   = $9f10                  ;VIA 6522 #2
     &ubyte  d2prb	= via2+0
     &ubyte  d2pra	= via2+1
     &ubyte  d2ddrb	= via2+2
@@ -207,6 +281,11 @@ cx16 {
     &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
@@ -235,7 +314,8 @@ 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
+
+; It's not documented what registers are clobbered, so we assume the worst for all following kernal routines...:
 
 ; high level graphics & fonts
 romsub $ff20 = GRAPH_init(uword vectors @R0)  clobbers(A,X,Y)
@@ -290,19 +370,42 @@ romsub $fecc = monitor()  clobbers(A,X,Y)
 
 ; ---- utilities -----
 
-inline asmsub rombank(ubyte rombank @A) {
+inline asmsub rombank(ubyte bank @A) {
     ; -- set the rom banks
     %asm {{
-        sta  $01            ; rom bank register (new)
-        sta  cx16.d1prb     ; rom bank register (old)
+        sta  $01            ; rom bank register (v39+, used to be cx16.d1prb $9f60 in v38)
     }}
 }
 
-inline asmsub rambank(ubyte rambank @A) {
+inline asmsub rambank(ubyte bank @A) {
     ; -- set the ram bank
     %asm {{
-        sta  $00            ; ram bank register (new)
-        sta  cx16.d1pra     ; ram bank register (old)
+        sta  $00            ; ram bank register (v39+, used to be cx16.d1pra $9f61 in v38)
+    }}
+}
+
+inline asmsub getrombank() -> ubyte @A {
+    ; -- get the current rom bank
+    %asm {{
+        lda  $01            ; rom bank register (v39+, used to be cx16.d1prb $9f60 in v38)
+    }}
+}
+
+inline asmsub getrambank() -> ubyte @A {
+    ; -- get the current ram bank
+    %asm {{
+        lda  $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
     }}
 }
 
@@ -349,75 +452,126 @@ asmsub vaddr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, byte autoIncrO
 }
 
 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
-        }}
+    ; -- 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
-        }}
+    ; -- 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
-        }}
+    ; -- 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
-        }}
+    ; -- 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
@@ -453,17 +607,15 @@ _loop       ldy  #0
 }
 
 ; ---- system stuff -----
-asmsub init_system()  {
+asmsub  init_system()  {
     ; Initializes the machine to a sane starting state.
     ; Called automatically by the loader program logic.
     %asm {{
         sei
         cld
-        ;stz  $00
-        ;stz  $01
-        ;stz  d1prb      ; select rom bank 0
         lda  #$80
         sta  VERA_CTRL
+        stz  $01        ; select rom bank 0 (enable kernal)
         jsr  c64.IOINIT
         jsr  c64.RESTOR
         jsr  c64.CINT
@@ -485,32 +637,215 @@ asmsub init_system()  {
     }}
 }
 
+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.
+    const ubyte target = 16         ;  compilation target specifier.  64 = C64,  128 = C128,  16 = CommanderX16.
 
 
     asmsub reset_system() {
-        ; Soft-reset the system back to Basic prompt.
+        ; Soft-reset the system back to initial power-on Basic prompt.
         %asm {{
             sei
-            stz  $01            ; bank the kernal in (new rom bank register)
-            stz  cx16.d1prb     ; bank the kernal in (old rom bank register)
+            stz  $01            ; bank the kernal in
             jmp  (cx16.RESET_VEC)
         }}
     }
 
-    sub wait(uword jiffies) {
-        ; --- wait approximately the given number of jiffies (1/60th seconds)
-        repeat jiffies {
-            ubyte jiff = lsb(c64.RDTIM16())
-            while jiff==lsb(c64.RDTIM16()) {
-                ; wait until 1 jiffy has passed
-            }
-        }
+    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) {
@@ -541,27 +876,6 @@ sys {
         }}
     }
 
-    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
@@ -571,25 +885,25 @@ sys {
 
     inline asmsub clear_carry() {
         %asm {{
-            clc
+        clc
         }}
     }
 
     inline asmsub set_carry() {
         %asm {{
-            sec
+        sec
         }}
     }
 
     inline asmsub clear_irqd() {
         %asm {{
-            cli
+        cli
         }}
     }
 
     inline asmsub set_irqd() {
         %asm {{
-            sei
+        sei
         }}
     }
 

compiler/res/prog8lib/cx16/textio.p8

@@ -4,7 +4,6 @@
 ;
 ; indent format: TABS, size=8
 
-%target cx16
 %import syslib
 %import conv
 
@@ -31,6 +30,16 @@ 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.
@@ -410,7 +419,7 @@ _print_byte_digits
 		jsr  c64.CHROUT
 		pla
 		jsr  c64.CHROUT
-		jmp  _ones
+		bra  _ones
 +       pla
         cmp  #'0'
         beq  _ones
@@ -433,7 +442,7 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 		jsr  c64.CHROUT
 +		pla
 		jsr  conv.byte2decimal
-		jmp  print_ub._print_byte_digits
+		bra  print_ub._print_byte_digits
 	}}
 }
 
@@ -484,7 +493,7 @@ asmsub  print_uwbin  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
 		jsr  print_ubbin
 		pla
 		clc
-		jmp  print_ubbin
+		bra  print_ubbin
 	}}
 }
 
@@ -497,7 +506,7 @@ asmsub  print_uwhex  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
 		jsr  print_ubhex
 		pla
 		clc
-		jmp  print_ubhex
+		bra  print_ubhex
 	}}
 }
 
@@ -556,11 +565,10 @@ asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
 		tay
 		pla
 		eor  #255
-		clc
-		adc  #1
-		bcc  +
+		ina
+		bne +
 		iny
-+		jmp  print_uw
++		bra  print_uw
 	}}
 }
 
@@ -588,95 +596,126 @@ asmsub  input_chars  (uword buffer @ AY) clobbers(A) -> ubyte @ Y  {
 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
+            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
+            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
+            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
+            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
+            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 kernel routine, to save the X register.
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
 	%asm  {{
 		phx
 		tax

compiler/res/prog8lib/cx16logo.p8

@@ -20,12 +20,12 @@ cx16logo {
     }
 
     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",
+            "\uf10d\uf11a\uf139\uf11b     \uf11a\uf13a\uf11b",
+            "\uf10b\uf11a▎\uf139\uf11b   \uf11a\uf13a\uf130\uf11b",
+            "\uf10f\uf11a▌ \uf139\uf11b \uf11a\uf13a \uf11b▌",
+            "\uf102 \uf132\uf11a▖\uf11b \uf11a▗\uf11b\uf132",
+            "\uf10e ▂\uf11a▘\uf11b \uf11a▝\uf11b▂",
+            "\uf104 \uf11a \uf11b\uf13a\uf11b \uf139\uf11a \uf11b",
             "\uf101\uf130\uf13a   \uf139▎\uf100"
         ]
 }