diskio.f_open_w() error handling back to what it was before

Otherwise it eats the status message. Added comment that you have to check this manually to be sure if the call succeeded or not!

.idea/kotlinc.xml

@@ -4,6 +4,6 @@
     <option name="jvmTarget" value="11" />
   </component>
   <component name="KotlinJpsPluginSettings">
-    <option name="version" value="1.9.0-release-358" />
+    <option name="version" value="1.9.20" />
   </component>
 </project>

.idea/libraries/antlr_antlr4.xml

@@ -1,13 +1,13 @@
 <component name="libraryTable">
   <library name="antlr.antlr4" type="repository">
-    <properties maven-id="org.antlr:antlr4:4.12.0">
+    <properties maven-id="org.antlr:antlr4:4.13.1">
       <exclude>
         <dependency maven-id="com.ibm.icu:icu4j" />
       </exclude>
     </properties>
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4/4.12.0/antlr4-4.12.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4-runtime/4.12.0/antlr4-runtime-4.12.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4/4.13.1/antlr4-4.13.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4-runtime/4.13.1/antlr4-runtime-4.13.1.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr-runtime/3.5.3/antlr-runtime-3.5.3.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/antlr/ST4/4.3.4/ST4-4.3.4.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/abego/treelayout/org.abego.treelayout.core/1.0.3/org.abego.treelayout.core-1.0.3.jar!/" />

.idea/libraries/io_kotest_assertions_core_jvm.xml

@@ -1,19 +1,19 @@
 <component name="libraryTable">
   <library name="io.kotest.assertions.core.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-assertions-core-jvm:5.6.2" />
+    <properties maven-id="io.kotest:kotest-assertions-core-jvm:5.7.2" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.6.2/kotest-assertions-core-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.7.2/kotest-assertions-core-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.8.10/kotlin-stdlib-jdk8-1.8.10.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.8.10/kotlin-stdlib-1.8.10.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.8.10/kotlin-stdlib-jdk7-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.6.2/kotest-assertions-shared-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.7.2/kotest-assertions-shared-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.12/java-diff-utils-4.12.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.3.0/opentest4j-1.3.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.8.10/kotlin-stdlib-common-1.8.10.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.7.0/kotlinx-coroutines-jdk8-1.7.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.8.10/kotlin-reflect-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.6.2/kotest-common-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.6.2/kotest-assertions-api-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.7.2/kotest-common-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.7.2/kotest-assertions-api-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.7.0/kotlinx-coroutines-core-jvm-1.7.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/23.0.0/annotations-23.0.0.jar!/" />
     </CLASSES>

.idea/libraries/io_kotest_property_jvm.xml

@@ -1,24 +0,0 @@
-<component name="libraryTable">
-  <library name="io.kotest.property.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-property-jvm:5.6.2" />
-    <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-property-jvm/5.6.2/kotest-property-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/com/github/curious-odd-man/rgxgen/1.4/rgxgen-1.4.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.8.10/kotlin-stdlib-jdk8-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.8.10/kotlin-stdlib-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.8.10/kotlin-stdlib-jdk7-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.6.2/kotest-common-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.6.2/kotest-assertions-shared-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.6.2/kotest-assertions-api-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.7.0/kotlinx-coroutines-jdk8-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-common/1.8.10/kotlin-stdlib-common-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.12/java-diff-utils-4.12.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.8.10/kotlin-reflect-1.8.10.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.7.0/kotlinx-coroutines-core-jvm-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/23.0.0/annotations-23.0.0.jar!/" />
-    </CLASSES>
-    <JAVADOC />
-    <SOURCES />
-  </library>
-</component>

.idea/libraries/io_kotest_runner_junit5_jvm.xml

@@ -1,29 +1,29 @@
 <component name="libraryTable">
   <library name="io.kotest.runner.junit5.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-runner-junit5-jvm:5.6.2" />
+    <properties maven-id="io.kotest:kotest-runner-junit5-jvm:5.7.2" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-runner-junit5-jvm/5.6.2/kotest-runner-junit5-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-api-jvm/5.6.2/kotest-framework-api-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.6.2/kotest-assertions-shared-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-runner-junit5-jvm/5.7.2/kotest-runner-junit5-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-api-jvm/5.7.2/kotest-framework-api-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.7.2/kotest-assertions-shared-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/io/github/java-diff-utils/java-diff-utils/4.12/java-diff-utils-4.12.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-test-jvm/1.7.0/kotlinx-coroutines-test-jvm-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.6.2/kotest-common-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-engine-jvm/5.6.2/kotest-framework-engine-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/github/classgraph/classgraph/4.8.157/classgraph-4.8.157.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.7.2/kotest-common-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-engine-jvm/5.7.2/kotest-framework-engine-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/github/classgraph/classgraph/4.8.162/classgraph-4.8.162.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/opentest4j/opentest4j/1.2.0/opentest4j-1.2.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/opentest4j/opentest4j/1.3.0/opentest4j-1.3.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-debug/1.7.0/kotlinx-coroutines-debug-1.7.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/java/dev/jna/jna/5.9.0/jna-5.9.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/java/dev/jna/jna-platform/5.9.0/jna-platform-5.9.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy/1.10.9/byte-buddy-1.10.9.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy-agent/1.10.9/byte-buddy-agent-1.10.9.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-discovery-jvm/5.6.2/kotest-framework-discovery-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.6.2/kotest-assertions-core-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-discovery-jvm/5.7.2/kotest-framework-discovery-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.7.2/kotest-assertions-core-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.7.0/kotlinx-coroutines-jdk8-1.7.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.6.2/kotest-assertions-api-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-extensions-jvm/5.6.2/kotest-extensions-jvm-5.6.2.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-concurrency-jvm/5.6.2/kotest-framework-concurrency-jvm-5.6.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.7.2/kotest-assertions-api-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-extensions-jvm/5.7.2/kotest-extensions-jvm-5.7.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-concurrency-jvm/5.7.2/kotest-framework-concurrency-jvm-5.7.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.7.0/kotlinx-coroutines-core-jvm-1.7.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/23.0.0/annotations-23.0.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-engine/1.8.2/junit-platform-engine-1.8.2.jar!/" />

.idea/libraries/jetbrains_kotlinx_cli_jvm.xml

@@ -1,8 +1,8 @@
 <component name="libraryTable">
   <library name="jetbrains.kotlinx.cli.jvm" type="repository">
-    <properties include-transitive-deps="false" maven-id="org.jetbrains.kotlinx:kotlinx-cli-jvm:0.3.5" />
+    <properties include-transitive-deps="false" maven-id="org.jetbrains.kotlinx:kotlinx-cli-jvm:0.3.6" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-cli-jvm/0.3.5/kotlinx-cli-jvm-0.3.5.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-cli-jvm/0.3.6/kotlinx-cli-jvm-0.3.6.jar!/" />
     </CLASSES>
     <JAVADOC />
     <SOURCES />

.idea/libraries/slf4j_simple.xml

@@ -1,9 +1,9 @@
 <component name="libraryTable">
   <library name="slf4j.simple" type="repository">
-    <properties maven-id="org.slf4j:slf4j-simple:2.0.7" />
+    <properties maven-id="org.slf4j:slf4j-simple:2.0.9" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-simple/2.0.7/slf4j-simple-2.0.7.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-api/2.0.7/slf4j-api-2.0.7.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-simple/2.0.9/slf4j-simple-2.0.9.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/slf4j/slf4j-api/2.0.9/slf4j-api-2.0.9.jar!/" />
     </CLASSES>
     <JAVADOC />
     <SOURCES />

.idea/misc.xml

@@ -16,6 +16,9 @@
       </list>
     </option>
   </component>
+  <component name="Black">
+    <option name="sdkName" value="Python 3.11" />
+  </component>
   <component name="FrameworkDetectionExcludesConfiguration">
     <type id="Python" />
   </component>

README.md

@@ -74,6 +74,8 @@ What does Prog8 provide?
 - "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)
+- "pet32": Commodore PET (experimental)
+- "atari": Atari 8 bit such as 800XL (experimental)
 - 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)
 
 

codeCore/build.gradle

@@ -32,10 +32,10 @@ dependencies {
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
 }

codeCore/src/prog8/code/SymbolTable.kt

@@ -196,13 +196,10 @@ class StStaticVariable(name: String,
         }
         if(onetimeInitializationNumericValue!=null) {
             require(dt in NumericDatatypes)
-            require(onetimeInitializationNumericValue!=0.0) { "zero as init value should just remain uninitialized"}
         }
         if(onetimeInitializationArrayValue!=null) {
             require(dt in ArrayDatatypes)
-            if(onetimeInitializationArrayValue.all { it.number!=null} ) {
-                require(onetimeInitializationArrayValue.any { it.number != 0.0 }) { "array of all zerors as init value should just remain uninitialized" }
-            }
+            require(length==onetimeInitializationArrayValue.size)
         }
         if(onetimeInitializationStringValue!=null) {
             require(dt == DataType.STR)

codeCore/src/prog8/code/SymbolTableMaker.kt

@@ -74,10 +74,9 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                             numElements = value.value.length + 1   // include the terminating 0-byte
                         }
                         is PtArray -> {
-                            val array = makeInitialArray(value)
-                            initialArray = if(array.all { it.number==0.0 }) null else array            // all 0 as init value -> just uninitialized
+                            initialArray = makeInitialArray(value)
                             initialString = null
-                            numElements = array.size
+                            numElements = initialArray.size
                             require(node.arraySize?.toInt()==numElements)
                         }
                         else -> {
@@ -126,7 +125,11 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
     private fun makeInitialArray(value: PtArray): List<StArrayElement> {
         return value.children.map {
             when(it) {
-                is PtAddressOf -> StArrayElement(null, it.identifier.name)
+                is PtAddressOf -> {
+                    if(it.isFromArrayElement)
+                        TODO("address-of array element $it in initial array value")
+                    StArrayElement(null, it.identifier.name)
+                }
                 is PtIdentifier -> StArrayElement(null, it.name)
                 is PtNumber -> StArrayElement(it.number, null)
                 else -> throw AssemblyError("invalid array element $it")

codeCore/src/prog8/code/ast/AstBase.kt

@@ -75,6 +75,7 @@ class PtBlock(name: String,
               val library: Boolean,
               val forceOutput: Boolean,
               val noSymbolPrefixing: Boolean,
+              val veraFxMuls: Boolean,
               val alignment: BlockAlignment,
               val source: SourceCode,       // taken from the module the block is defined in.
               position: Position

codeCore/src/prog8/code/ast/AstExpressions.kt

@@ -24,7 +24,17 @@ sealed class PtExpression(val type: DataType, position: Position) : PtNode(posit
 
     infix fun isSameAs(other: PtExpression): Boolean {
         return when(this) {
-            is PtAddressOf -> other is PtAddressOf && other.type==type && other.identifier isSameAs identifier
+            is PtAddressOf -> {
+                if(other !is PtAddressOf)
+                    return false
+                if (other.type!==type || !(other.identifier isSameAs identifier))
+                    return false
+                if(other.children.size!=children.size)
+                    return false
+                if(children.size==1)
+                    return true
+                return arrayIndexExpr!! isSameAs other.arrayIndexExpr!!
+            }
             is PtArrayIndexer -> other is PtArrayIndexer && other.type==type && other.variable isSameAs variable && other.index isSameAs index && other.splitWords==splitWords
             is PtBinaryExpression -> other is PtBinaryExpression && other.left isSameAs left && other.right isSameAs right
             is PtContainmentCheck -> other is PtContainmentCheck && other.type==type && other.element isSameAs element && other.iterable isSameAs iterable
@@ -105,7 +115,12 @@ sealed class PtExpression(val type: DataType, position: Position) : PtNode(posit
 
 class PtAddressOf(position: Position) : PtExpression(DataType.UWORD, position) {
     val identifier: PtIdentifier
-        get() = children.single() as PtIdentifier
+        get() = children[0] as PtIdentifier
+    val arrayIndexExpr: PtExpression?
+        get() = if(children.size==2) children[1] as PtExpression else null
+
+    val isFromArrayElement: Boolean
+        get() = children.size==2
 }
 
 

codeCore/src/prog8/code/core/BuiltinFunctions.kt

@@ -70,6 +70,8 @@ class FSignature(val pure: Boolean,      // does it have side effects?
 
 val BuiltinFunctions: Map<String, FSignature> = mapOf(
     // this set of function have no return value and operate in-place:
+    "setlsb"    to FSignature(false, listOf(FParam("variable", arrayOf(DataType.WORD, DataType.UWORD)), FParam("value", arrayOf(DataType.BYTE, DataType.UBYTE))), null),
+    "setmsb"    to FSignature(false, listOf(FParam("variable", arrayOf(DataType.WORD, DataType.UWORD)), FParam("value", arrayOf(DataType.BYTE, DataType.UBYTE))), null),
     "rol"       to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
     "ror"       to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
     "rol2"      to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
@@ -78,7 +80,9 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "reverse"   to FSignature(false, listOf(FParam("array", ArrayDatatypes)), null),
     // cmp returns a status in the carry flag, but not a proper return value
     "cmp"       to FSignature(false, listOf(FParam("value1", IntegerDatatypesNoBool), FParam("value2", NumericDatatypesNoBool)), null),
-    "prog8_lib_stringcompare"       to FSignature(true, listOf(FParam("str1", arrayOf(DataType.STR)), FParam("str2", arrayOf(DataType.STR))), DataType.BYTE),
+    "prog8_lib_stringcompare"     to FSignature(true, listOf(FParam("str1", arrayOf(DataType.STR)), FParam("str2", arrayOf(DataType.STR))), DataType.BYTE),
+    "prog8_lib_square_byte"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.BYTE, DataType.UBYTE))), DataType.UBYTE),
+    "prog8_lib_square_word"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.WORD, DataType.UWORD))), DataType.UWORD),
     "abs"       to FSignature(true, listOf(FParam("value", NumericDatatypesNoBool)), null),
     "abs__byte"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.BYTE))), DataType.BYTE),
     "abs__word"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.WORD))), DataType.WORD),
@@ -117,8 +121,8 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "peek"      to FSignature(true, listOf(FParam("address", arrayOf(DataType.UWORD))), DataType.UBYTE),
     "peekw"     to FSignature(true, listOf(FParam("address", arrayOf(DataType.UWORD))), DataType.UWORD),
     "poke"      to FSignature(false, listOf(FParam("address", arrayOf(DataType.UWORD)), FParam("value", arrayOf(DataType.UBYTE))), null),
-    "pokemon"   to FSignature(false, listOf(FParam("address", arrayOf(DataType.UWORD)), FParam("value", arrayOf(DataType.UBYTE))), null),
     "pokew"     to FSignature(false, listOf(FParam("address", arrayOf(DataType.UWORD)), FParam("value", arrayOf(DataType.UWORD))), null),
+    "pokemon"   to FSignature(false, listOf(FParam("address", arrayOf(DataType.UWORD)), FParam("value", arrayOf(DataType.UBYTE))), DataType.UBYTE),
     "pop"       to FSignature(false, listOf(FParam("target", ByteDatatypes)), null),
     "popw"      to FSignature(false, listOf(FParam("target", WordDatatypes)), null),
     "push"      to FSignature(false, listOf(FParam("value", ByteDatatypes)), null),
@@ -129,4 +133,4 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "callfar"   to FSignature(false, listOf(FParam("bank", arrayOf(DataType.UBYTE)), FParam("address", arrayOf(DataType.UWORD)), FParam("arg", arrayOf(DataType.UWORD))), DataType.UWORD),
 )
 
-val InplaceModifyingBuiltinFunctions = setOf("rol", "ror", "rol2", "ror2", "sort", "reverse")
+val InplaceModifyingBuiltinFunctions = setOf("setlsb", "setmsb", "rol", "ror", "rol2", "ror2", "sort", "reverse")

codeCore/src/prog8/code/core/CompilationOptions.kt

@@ -8,11 +8,13 @@ class CompilationOptions(val output: OutputType,
                          val launcher: CbmPrgLauncherType,
                          val zeropage: ZeropageType,
                          val zpReserved: List<UIntRange>,
+                         val zpAllowed: List<UIntRange>,
                          val floats: Boolean,
                          val noSysInit: Boolean,
                          val compTarget: ICompilationTarget,
                          // these are set later, based on command line arguments or options in the source code:
                          var loadAddress: UInt,
+                         var warnSymbolShadowing: Boolean = false,
                          var optimize: Boolean = false,
                          var asmQuiet: Boolean = false,
                          var asmListfile: Boolean = false,
@@ -20,10 +22,15 @@ class CompilationOptions(val output: OutputType,
                          var experimentalCodegen: Boolean = false,
                          var varsHighBank: Int? = null,
                          var splitWordArrays: Boolean = false,
+                         var breakpointCpuInstruction: Boolean = false,
                          var outputDir: Path = Path(""),
                          var symbolDefs: Map<String, String> = emptyMap()
 ) {
     init {
         compTarget.machine.initializeMemoryAreas(this)
     }
+
+    companion object {
+        val AllZeropageAllowed: List<UIntRange> = listOf(0u..255u)
+    }
 }

codeCore/src/prog8/code/core/Enumerations.kt

@@ -76,6 +76,13 @@ enum class RegisterOrPair {
 
     companion object {
         val names by lazy { values().map { it.toString()} }
+        fun fromCpuRegister(cpu: CpuRegister): RegisterOrPair {
+            return when(cpu) {
+                CpuRegister.A -> A
+                CpuRegister.X -> X
+                CpuRegister.Y -> Y
+            }
+        }
     }
 
     fun asCpuRegister(): CpuRegister = when(this) {
@@ -152,6 +159,7 @@ val ElementToArrayTypes = mapOf(
     DataType.FLOAT to DataType.ARRAY_F,
     DataType.BOOL to DataType.ARRAY_BOOL
 )
+
 val Cx16VirtualRegisters = arrayOf(
     RegisterOrPair.R0, RegisterOrPair.R1, RegisterOrPair.R2, RegisterOrPair.R3,
     RegisterOrPair.R4, RegisterOrPair.R5, RegisterOrPair.R6, RegisterOrPair.R7,
@@ -159,6 +167,10 @@ val Cx16VirtualRegisters = arrayOf(
     RegisterOrPair.R12, RegisterOrPair.R13, RegisterOrPair.R14, RegisterOrPair.R15
 )
 
+val CpuRegisters = setOf(
+    RegisterOrPair.A, RegisterOrPair.X, RegisterOrPair.Y,
+    RegisterOrPair.AX, RegisterOrPair.AY, RegisterOrPair.XY
+)
 
 
 enum class OutputType {

codeCore/src/prog8/code/core/MemoryRegions.kt

@@ -42,6 +42,13 @@ abstract class Zeropage(options: CompilationOptions): MemoryAllocator(options) {
         }
     }
 
+    fun retainAllowed() {
+        synchronized(this) {
+            for(allowed in options.zpAllowed)
+                free.retainAll { it in allowed }
+        }
+    }
+
     fun availableBytes() = if(options.zeropage== ZeropageType.DONTUSE) 0 else free.size
     fun hasByteAvailable() = if(options.zeropage== ZeropageType.DONTUSE) false else free.isNotEmpty()
     fun hasWordAvailable(): Boolean {

codeCore/src/prog8/code/target/PETTarget.kt

@@ -0,0 +1,20 @@
+package prog8.code.target
+
+import prog8.code.core.Encoding
+import prog8.code.core.ICompilationTarget
+import prog8.code.core.IMemSizer
+import prog8.code.core.IStringEncoding
+import prog8.code.target.cbm.CbmMemorySizer
+import prog8.code.target.pet.PETMachineDefinition
+
+
+class PETTarget: ICompilationTarget, IStringEncoding by Encoder, IMemSizer by CbmMemorySizer {
+    override val name = NAME
+    override val machine = PETMachineDefinition()
+    override val supportedEncodings = setOf(Encoding.PETSCII, Encoding.SCREENCODES)
+    override val defaultEncoding = Encoding.PETSCII
+
+    companion object {
+        const val NAME = "pet32"
+    }
+}

codeCore/src/prog8/code/target/atari/AtariZeropage.kt

@@ -48,6 +48,7 @@ class AtariZeropage(options: CompilationOptions) : Zeropage(options) {
         free.addAll(distinctFree)
 
         removeReservedFromFreePool()
+        retainAllowed()
     }
 
     override fun allocateCx16VirtualRegisters() {

codeCore/src/prog8/code/target/c128/C128Zeropage.kt

@@ -69,6 +69,7 @@ class C128Zeropage(options: CompilationOptions) : Zeropage(options) {
         free.addAll(distinctFree)
 
         removeReservedFromFreePool()
+        retainAllowed()
     }
 
     override fun allocateCx16VirtualRegisters() {

codeCore/src/prog8/code/target/c64/C64Zeropage.kt

@@ -53,11 +53,11 @@ class C64Zeropage(options: CompilationOptions) : Zeropage(options) {
                 ).map{it.toUInt()})
             }
 
-            if(options.zeropage!= ZeropageType.DONTUSE) {
+            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(0x02, 0x03, 0x04, 0x05, 0x06, 0x0a, 0x0e,
-                        0x92, 0x96, 0x9b, 0x9c, 0x9e, 0x9f, 0xa5, 0xa6,
+                        0x92, 0x96, 0x9b, 0x9c, 0x9e, 0x9f, 0xa6,
                         0xb0, 0xb1, 0xbe, 0xbf, 0xf9).map{it.toUInt()})
             } else {
                 // don't use the zeropage at all
@@ -75,6 +75,8 @@ class C64Zeropage(options: CompilationOptions) : Zeropage(options) {
             // in these cases there is enough space on the zero page to stick the cx16 virtual registers in there as well.
             allocateCx16VirtualRegisters()
         }
+
+        retainAllowed()
     }
 
     override fun allocateCx16VirtualRegisters() {

codeCore/src/prog8/code/target/cbm/PetsciiEncoding.kt

@@ -24,7 +24,7 @@ object PetsciiEncoding {
         '\ufffe',    //       0x0A -> UNDEFINED
         '\ufffe',    //       0x0B -> UNDEFINED
         '\ufffe',    //       0x0C -> UNDEFINED
-        '\r'    ,    //       0x0D -> CARRIAGE RETURN
+        '\n'    ,    //       0x0D -> LINE FEED (RETURN)
         '\u000e',    //       0x0E -> SHIFT OUT
         '\ufffe',    //       0x0F -> UNDEFINED
         '\ufffe',    //       0x10 -> UNDEFINED
@@ -152,7 +152,7 @@ object PetsciiEncoding {
         '\uf113',    //      0x8A -> FUNCTION KEY 4 (CUS)
         '\uf115',    //      0x8B -> FUNCTION KEY 6 (CUS)
         '\uf117',    //      0x8C -> FUNCTION KEY 8 (CUS)
-        '\n'    ,    //       0x8D -> LINE FEED
+        '\r'    ,    //       0x8D -> CARRIAGE RETURN (SHIFT-RETURN)
         '\u000f',    //      0x8E -> SHIFT IN
         '\ufffe',    //       0x8F -> UNDEFINED
         '\uf105',    //       0x90 -> BLACK COLOR SWITCH (CUS)
@@ -283,7 +283,7 @@ object PetsciiEncoding {
         '\ufffe',    //       0x0A -> UNDEFINED
         '\ufffe',    //       0x0B -> UNDEFINED
         '\ufffe',    //       0x0C -> UNDEFINED
-        '\r'    ,    //       0x0D -> CARRIAGE RETURN
+        '\n'    ,    //       0x0D -> LINE FEED (RETURN)
         '\u000e',    //       0x0E -> SHIFT OUT
         '\ufffe',    //       0x0F -> UNDEFINED
         '\ufffe',    //       0x10 -> UNDEFINED
@@ -411,7 +411,7 @@ object PetsciiEncoding {
         '\uf113',    //       0x8A -> FUNCTION KEY 4 (CUS)
         '\uf115',    //       0x8B -> FUNCTION KEY 6 (CUS)
         '\uf117',    //       0x8C -> FUNCTION KEY 8 (CUS)
-        '\n'    ,    //       0x8D -> LINE FEED
+        '\r'    ,    //       0x8D -> CARRIAGE RETURN (SHIFT-RETURN)
         '\u000f',    //       0x8E -> SHIFT IN
         '\ufffe',    //       0x8F -> UNDEFINED
         '\uf105',    //       0x90 -> BLACK COLOR SWITCH (CUS)
@@ -1061,6 +1061,7 @@ object PetsciiEncoding {
             '}' -> '├'
             '|' -> '│'
             '\\' -> '╲'
+            '\r' -> '\n'        // to make \r (carriage returrn) equivalent to \n (line feed): RETURN ($0d)
             else -> chr
         }
 
@@ -1076,7 +1077,10 @@ object PetsciiEncoding {
                 }
                 else -> {
                     val case = if (lowercase) "lower" else "upper"
-                    throw CharConversionException("no ${case}Petscii character for '${chr}' (${chr.code})")
+                    if(chr.isISOControl())
+                        throw CharConversionException("no ${case}Petscii character for char #${chr.code}")
+                    else
+                        throw CharConversionException("no ${case}Petscii character for char #${chr.code} '${chr}'")
                 }
             }
         }
@@ -1119,7 +1123,10 @@ object PetsciiEncoding {
                 }
                 else -> {
                     val case = if (lowercase) "lower" else "upper"
-                    throw CharConversionException("no ${case}Screencode character for '${chr}' (${chr.code})")
+                    if(chr.isISOControl())
+                        throw CharConversionException("no ${case}Screencode character for char #${chr.code}")
+                    else
+                        throw CharConversionException("no ${case}Screencode character for char #${chr.code} '${chr}'")
                 }
             }
         }

codeCore/src/prog8/code/target/cx16/CX16Zeropage.kt

@@ -47,8 +47,8 @@ class CX16Zeropage(options: CompilationOptions) : Zeropage(options) {
             free.addAll(distinctFree)
 
             removeReservedFromFreePool()
-
             allocateCx16VirtualRegisters()
+            retainAllowed()
         }
     }
 

codeCore/src/prog8/code/target/pet/PETMachineDefinition.kt

@@ -0,0 +1,55 @@
+package prog8.code.target.pet
+
+import prog8.code.core.*
+import prog8.code.target.C64Target
+import prog8.code.target.cbm.Mflpt5
+import java.nio.file.Path
+
+
+class PETMachineDefinition: 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 PROGRAM_LOAD_ADDRESS = 0x0401u
+
+    override val BSSHIGHRAM_START = 0xffffu
+    override val BSSHIGHRAM_END = 0xffffu
+
+    override lateinit var zeropage: Zeropage
+    override lateinit var golden: GoldenRam
+
+    override fun getFloatAsmBytes(num: Number) = Mflpt5.fromNumber(num).makeFloatFillAsm()
+
+    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
+    }
+
+    override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
+        if(selectedEmulator!=1) {
+            System.err.println("The pet target only supports the main emulator (Vice).")
+            return
+        }
+
+        println("\nStarting PET emulator...")
+        val viceMonlist = C64Target.viceMonListName(programNameWithPath.toString())
+        val cmdline = listOf("xpet", "-model", "4032", "-ramsize", "32", "-videosize", "40", "-silent", "-moncommands", viceMonlist,
+                "-autostartprgmode", "1", "-autostart-warp", "-autostart", "${programNameWithPath}.prg")
+        val processb = ProcessBuilder(cmdline).inheritIO()
+        val process=processb.start()
+        process.waitFor()
+    }
+
+    override fun isIOAddress(address: UInt): Boolean = address in 0xe800u..0xe8ffu
+
+    override fun initializeMemoryAreas(compilerOptions: CompilationOptions) {
+        zeropage = PETZeropage(compilerOptions)
+        // there's no golden ram.
+    }
+
+}

codeCore/src/prog8/code/target/pet/PETZeropage.kt

@@ -0,0 +1,59 @@
+package prog8.code.target.pet
+
+import prog8.code.core.CompilationOptions
+import prog8.code.core.InternalCompilerException
+import prog8.code.core.Zeropage
+import prog8.code.core.ZeropageType
+
+
+// reference: http://www.zimmers.net/cbmpics/cbm/PETx/petmem.txt
+
+class PETZeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0xb3u      // temp storage for a single byte
+    override val SCRATCH_REG = 0xb4u     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0xb6u      // temp storage 1 for a word
+    override val SCRATCH_W2 = 0xb8u      // temp storage 2 for a word
+
+    init {
+        if (options.floats) {
+            throw InternalCompilerException("PET target doesn't yet support floating point routines")
+        }
+
+        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 -> {
+                free.addAll(0x00u..0xffu)
+                free.removeAll(listOf(0x8du, 0x8eu, 0x8fu, 0x97u, 0x98u, 0x99u, 0x9au, 0x9bu, 0x9eu, 0xa7u, 0xa8u, 0xa9u, 0xaau))        // these are updated/used by IRQ
+            }
+            ZeropageType.KERNALSAFE -> {
+                free.addAll(0x00u..0xffu)
+                free.removeAll(listOf(0x8du, 0x8eu, 0x8fu, 0x97u, 0x98u, 0x99u, 0x9au, 0x9bu, 0x9eu, 0xa7u, 0xa8u, 0xa9u, 0xaau))        // these are updated/used by IRQ
+            }
+            ZeropageType.FLOATSAFE,
+            ZeropageType.BASICSAFE -> {
+                free.addAll(0xb3u..0xbau)       // TODO more?
+            }
+            ZeropageType.DONTUSE -> {
+                free.clear()  // don't use zeropage at all
+            }
+        }
+
+        val distinctFree = free.distinct()
+        free.clear()
+        free.addAll(distinctFree)
+
+        removeReservedFromFreePool()
+        retainAllowed()
+    }
+
+    override fun allocateCx16VirtualRegisters() {
+        TODO("Not known if PET can put the virtual regs in ZP")
+    }
+}

codeGenCpu6502/build.gradle

@@ -30,16 +30,16 @@ dependencies {
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
+    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.7.2'
 }
 
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
     test {

codeGenCpu6502/src/prog8/codegen/cpu6502/AsmGen.kt

@@ -503,24 +503,18 @@ class AsmGen6502Internal (
             return
         }
 
-        val indexVar = expr.index as? PtIdentifier
-            ?: throw AssemblyError("array indexer should have been replaced with a temp var @ ${expr.index.position}")
-
-        val indexName = asmVariableName(indexVar)
-
         if(expr.splitWords) {
-            when (register) {
-                CpuRegister.A -> out("  lda  $indexName")
-                CpuRegister.X -> out("  ldx  $indexName")
-                CpuRegister.Y -> out("  ldy  $indexName")
-            }
+            assignExpressionToRegister(expr.index, RegisterOrPair.fromCpuRegister(register), false)
             return
         }
 
         when (elementDt) {
-            in ByteDatatypes -> out("  ld$reg  $indexName")
+            in ByteDatatypes -> {
+                assignExpressionToRegister(expr.index, RegisterOrPair.fromCpuRegister(register), false)
+            }
             in WordDatatypes -> {
-                out("  lda  $indexName |  asl  a")
+                assignExpressionToRegister(expr.index, RegisterOrPair.A, false)
+                out("  asl  a")
                 when (register) {
                     CpuRegister.A -> {}
                     CpuRegister.X -> out(" tax")
@@ -529,12 +523,13 @@ class AsmGen6502Internal (
             }
             DataType.FLOAT -> {
                 require(options.compTarget.memorySize(DataType.FLOAT) == 5) {"invalid float size ${expr.position}"}
+                assignExpressionToRegister(expr.index, RegisterOrPair.A, false)
                 out("""
-                    lda  $indexName
+                    sta  P8ZP_SCRATCH_REG
                     asl  a
                     asl  a
                     clc
-                    adc  $indexName"""
+                    adc  P8ZP_SCRATCH_REG"""
                 )
                 when (register) {
                     CpuRegister.A -> {}
@@ -568,7 +563,7 @@ class AsmGen6502Internal (
         when(reg) {
             RegisterOrPair.A,
             RegisterOrPair.X,
-            RegisterOrPair.Y -> assignmentAsmGen.assignRegisterByte(target, reg.asCpuRegister(), target.datatype in SignedDatatypes)
+            RegisterOrPair.Y -> assignmentAsmGen.assignRegisterByte(target, reg.asCpuRegister(), target.datatype in SignedDatatypes, true)
             RegisterOrPair.AX,
             RegisterOrPair.AY,
             RegisterOrPair.XY,
@@ -582,6 +577,34 @@ class AsmGen6502Internal (
     internal fun assignExpressionTo(value: PtExpression, target: AsmAssignTarget) {
         when (target.datatype) {
             in ByteDatatypes -> {
+                if (value.asConstInteger()==0) {
+                    when(target.kind) {
+                        TargetStorageKind.VARIABLE -> {
+                            if (isTargetCpu(CpuType.CPU6502))
+                                out("lda  #0 |  sta  ${target.asmVarname}")
+                            else
+                                out("stz  ${target.asmVarname}")
+                        }
+                        TargetStorageKind.MEMORY -> {
+                            val address = target.memory!!.address.asConstInteger()
+                            if(address!=null) {
+                                if (isTargetCpu(CpuType.CPU6502))
+                                    out("lda  #0 |  sta  ${address.toHex()}")
+                                else
+                                    out("  stz  ${address.toHex()}")
+                                return
+                            }
+                        }
+                        TargetStorageKind.REGISTER -> {
+                            val zero = PtNumber(DataType.UBYTE, 0.0, value.position)
+                            zero.parent = value
+                            assignExpressionToRegister(zero, target.register!!, false)
+                            return
+                        }
+                        else -> { }
+                    }
+                }
+
                 assignExpressionToRegister(value, RegisterOrPair.A)
                 assignRegister(RegisterOrPair.A, target)
             }
@@ -1026,6 +1049,15 @@ $repeatLabel""")
         val label = "_prog8_breakpoint_${breakpointLabels.size+1}"
         breakpointLabels.add(label)
         out(label)
+        if(options.breakpointCpuInstruction) {
+            val instruction =
+                when(options.compTarget.machine.cpu) {
+                    CpuType.CPU6502 -> "brk"
+                    CpuType.CPU65c02 -> "stp"
+                    else -> throw AssemblyError("invalid cpu type")
+                }
+            out("  $instruction")
+        }
     }
 
     internal fun signExtendAYlsb(valueDt: DataType) {
@@ -2417,8 +2449,7 @@ $repeatLabel""")
                     cmp  #<$number
                     bne  $jumpIfFalseLabel
                     cpy  #>$number
-                    bne  $jumpIfFalseLabel
-                    """)
+                    bne  $jumpIfFalseLabel""")
             }
             is PtIdentifier -> {
                 assignExpressionToRegister(left, RegisterOrPair.AY)
@@ -2426,18 +2457,21 @@ $repeatLabel""")
                     cmp  ${asmVariableName(right)}
                     bne  $jumpIfFalseLabel
                     cpy  ${asmVariableName(right)}+1
-                    bne  $jumpIfFalseLabel
-                    """)
+                    bne  $jumpIfFalseLabel""")
             }
             is PtAddressOf -> {
                 assignExpressionToRegister(left, RegisterOrPair.AY)
                 val name = asmSymbolName(right.identifier)
-                out("""
-                    cmp  #<$name
-                    bne  $jumpIfFalseLabel
-                    cpy  #>$name
-                    bne  $jumpIfFalseLabel
-                    """)
+                if(right.isFromArrayElement) {
+                    TODO("address-of array element $name at ${right.position}")
+                    // assignmentAsmGen.assignAddressOf(target, name, right.arrayIndexExpr)
+                } else {
+                    out("""
+                        cmp  #<$name
+                        bne  $jumpIfFalseLabel
+                        cpy  #>$name
+                        bne  $jumpIfFalseLabel""")
+                }
             }
             else -> {
                 if(left.isSimple()) {
@@ -2518,12 +2552,16 @@ $repeatLabel""")
             is PtAddressOf -> {
                 assignExpressionToRegister(left, RegisterOrPair.AY)
                 val name = asmSymbolName(right.identifier)
-                out("""
-                    cmp  #<$name
-                    bne  +
-                    cpy  #>$name
-                    beq  $jumpIfFalseLabel
-+""")
+                if(right.isFromArrayElement) {
+                    TODO("address-of array element $name at ${right.position}")
+                } else {
+                    out("""
+                        cmp  #<$name
+                        bne  +
+                        cpy  #>$name
+                        beq  $jumpIfFalseLabel
+    +""")
+                }
             }
             else -> {
                 if(left.isSimple()) {
@@ -2849,8 +2887,12 @@ $repeatLabel""")
             is PtAddressOf -> {
                 assignExpressionToRegister(right, RegisterOrPair.AY)
                 val name = asmSymbolName(left.identifier)
-                code("#>$name", "#<$name")
-                return true
+                if(left.isFromArrayElement) {
+                    TODO("address-of array element $name at ${left.position}")
+                } else {
+                    code("#>$name", "#<$name")
+                    return true
+                }
             }
             is PtIdentifier -> {
                 assignExpressionToRegister(right, RegisterOrPair.AY)
@@ -2898,8 +2940,12 @@ $repeatLabel""")
             is PtAddressOf -> {
                 assignExpressionToRegister(left, RegisterOrPair.AY)
                 val name = asmSymbolName(right.identifier)
-                code("#>$name", "#<$name")
-                return true
+                if(right.isFromArrayElement) {
+                    TODO("address-of array element $name at ${right.position}")
+                } else {
+                    code("#>$name", "#<$name")
+                    return true
+                }
             }
             is PtIdentifier -> {
                 assignExpressionToRegister(left, RegisterOrPair.AY)
@@ -3031,7 +3077,11 @@ $repeatLabel""")
     }
 
     internal fun popFAC1() {
-        out("  jsr  floats.popFAC1")
+        out("  clc |  jsr  floats.popFAC")
+    }
+
+    internal fun popFAC2() {
+        out("  sec |  jsr  floats.popFAC")
     }
 
     internal fun needAsaveForExpr(arg: PtExpression): Boolean =

codeGenCpu6502/src/prog8/codegen/cpu6502/AsmOptimizer.kt

@@ -43,6 +43,13 @@ internal fun optimizeAssembly(lines: MutableList<String>, machine: IMachineDefin
         numberOfOptimizations++
     }
 
+    mods = optimizeUnneededTempvarInAdd(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
     var linesByFourteen = getLinesBy(lines, 14)
     mods = optimizeSameAssignments(linesByFourteen, machine, symbolTable)
     if(mods.isNotEmpty()) {
@@ -508,4 +515,27 @@ private fun optimizeUselessPushPopStack(linesByFour: List<List<IndexedValue<Stri
         optimize('y', lines)
     }
     return mods
-}
+}
+
+private fun optimizeUnneededTempvarInAdd(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
+    // sequence:  sta  P8ZP_SCRATCH_XX  / lda  something / clc / adc  P8ZP_SCRATCH_XX
+    // this can be performed without the scratch variable:  clc  /  adc  something
+    val mods = mutableListOf<Modification>()
+
+    for(lines in linesByFour) {
+        val first = lines[0].value.trimStart()
+        val second = lines[1].value.trimStart()
+        val third = lines[2].value.trimStart()
+        val fourth = lines[3].value.trimStart()
+        if(first.startsWith("sta  P8ZP_SCRATCH_") && second.startsWith("lda") && third.startsWith("clc") && fourth.startsWith("adc  P8ZP_SCRATCH_") ) {
+            if(fourth.substring(4)==first.substring(4)) {
+                mods.add(Modification(lines[0].index, false, "    clc"))
+                mods.add(Modification(lines[1].index, false, "    adc  ${second.substring(3).trimStart()}"))
+                mods.add(Modification(lines[2].index, true, null))
+                mods.add(Modification(lines[3].index, true, null))
+            }
+        }
+    }
+
+    return mods
+}

codeGenCpu6502/src/prog8/codegen/cpu6502/AsmsubHelpers.kt

@@ -1,9 +1,8 @@
 package prog8.codegen.cpu6502
 
-import prog8.code.ast.*
+import prog8.code.ast.PtAsmSub
 import prog8.code.core.Cx16VirtualRegisters
 import prog8.code.core.RegisterOrPair
-import prog8.code.core.RegisterOrStatusflag
 
 
 fun asmsub6502ArgsEvalOrder(sub: PtAsmSub): List<Int> {
@@ -11,50 +10,19 @@ fun asmsub6502ArgsEvalOrder(sub: PtAsmSub): List<Int> {
     // order is:
     //  1) cx16 virtual word registers,
     //  2) paired CPU registers,
-    //  3) single CPU registers (X last), except A,
+    //  3) single CPU registers (order Y,X,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.withIndex()
     val (cx16regs, args2) = args.partition { it.value.first.registerOrPair in Cx16VirtualRegisters }
     val pairedRegisters = arrayOf(RegisterOrPair.AX, RegisterOrPair.AY, RegisterOrPair.XY)
     val (pairedRegs , args3) = args2.partition { it.value.first.registerOrPair in pairedRegisters }
-    val (regsWithoutA, args4) = args3.partition { it.value.first.registerOrPair != RegisterOrPair.A }
-    val (regA, rest) = args4.partition { it.value.first.registerOrPair != null }
+    val (singleRegs, rest) = args3.partition { it.value.first.registerOrPair != null }
 
     cx16regs.forEach { order += it.index }
     pairedRegs.forEach { order += it.index }
-    regsWithoutA.forEach {
-        if(it.value.first.registerOrPair != RegisterOrPair.X)
-            order += it.index
-    }
-    regsWithoutA.firstOrNull { it.value.first.registerOrPair==RegisterOrPair.X } ?.let { order += it.index}
+    singleRegs.sortedBy { it.value.first.registerOrPair!!.asCpuRegister() }.asReversed().forEach { order += it.index }
+    require(rest.all { it.value.first.registerOrPair==null && it.value.first.statusflag!=null})
     rest.forEach { order += it.index }
-    regA.forEach { order += it.index }
     require(order.size==sub.parameters.size)
     return order
 }
-
-fun asmsub6502ArgsHaveRegisterClobberRisk(
-    args: List<PtExpression>,
-    params: List<Pair<RegisterOrStatusflag, PtSubroutineParameter>>
-): Boolean {
-    fun isClobberRisk(expr: PtExpression): Boolean {
-        when (expr) {
-            is PtArrayIndexer -> {
-                return params.any {
-                    it.first.registerOrPair in listOf(RegisterOrPair.Y, RegisterOrPair.AY, RegisterOrPair.XY)
-                }
-            }
-            is PtBuiltinFunctionCall -> {
-                if (expr.name == "lsb" || expr.name == "msb")
-                    return isClobberRisk(expr.args[0])
-                if (expr.name == "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) }
-}

codeGenCpu6502/src/prog8/codegen/cpu6502/AssemblyProgram.kt

@@ -22,15 +22,20 @@ internal class AssemblyProgram(
         val assemblerCommand: List<String>
 
         when (compTarget.name) {
-            in setOf("c64", "c128", "cx16") -> {
+            in setOf("c64", "c128", "cx16", "pet32") -> {
                 // CBM machines .prg generation.
 
                 // 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",
+                    "-Wall", // "-Wno-strict-bool", "-Werror",
                     "--dump-labels", "--vice-labels", "--labels=$viceMonListFile", "--no-monitor"
                 )
 
+                if(options.warnSymbolShadowing)
+                    command.add("-Wshadow")
+                else
+                    command.add("-Wno-shadow")
+
                 if(options.asmQuiet)
                     command.add("--quiet")
 
@@ -59,10 +64,15 @@ internal class AssemblyProgram(
 
                 // TODO are these options okay for atari?
                 val command = mutableListOf("64tass", "--ascii", "--case-sensitive", "--long-branch",
-                    "-Wall", "-Wno-strict-bool", "-Wno-shadow", // "-Werror",
+                    "-Wall", // "-Werror", "-Wno-strict-bool"
                     "--no-monitor"
                 )
 
+                if(options.warnSymbolShadowing)
+                    command.add("-Wshadow")
+                else
+                    command.add("-Wno-shadow")
+
                 if(options.asmQuiet)
                     command.add("--quiet")
 

codeGenCpu6502/src/prog8/codegen/cpu6502/BuiltinFunctionsAsmGen.kt

@@ -41,13 +41,24 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "rol2" -> funcRol2(fcall)
             "ror" -> funcRor(fcall)
             "ror2" -> funcRor2(fcall)
+            "setlsb" -> funcSetLsbMsb(fcall, false)
+            "setmsb" -> funcSetLsbMsb(fcall, true)
             "sort" -> funcSort(fcall)
             "reverse" -> funcReverse(fcall)
             "memory" -> funcMemory(fcall, discardResult, resultRegister)
             "peekw" -> funcPeekW(fcall, resultRegister)
             "peek" -> throw AssemblyError("peek() should have been replaced by @()")
             "pokew" -> funcPokeW(fcall)
-            "pokemon" -> { /* meme function */ }
+            "pokemon" -> {
+                val memread = PtMemoryByte(fcall.position)
+                memread.add(fcall.args[0])
+                memread.parent = fcall
+                asmgen.assignExpressionToRegister(memread, RegisterOrPair.A, false)
+                asmgen.out("  pha")
+                val memtarget = AsmAssignTarget(TargetStorageKind.MEMORY, asmgen, DataType.UBYTE, fcall.definingISub(), fcall.position, memory=memread)
+                asmgen.assignExpressionTo(fcall.args[1], memtarget)
+                asmgen.out("  pla")
+            }
             "poke" -> throw AssemblyError("poke() should have been replaced by @()")
             "push" -> asmgen.pushCpuStack(DataType.UBYTE, fcall.args[0])
             "pushw" -> asmgen.pushCpuStack(DataType.UWORD, fcall.args[0])
@@ -72,12 +83,31 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "cmp" -> funcCmp(fcall)
             "callfar" -> funcCallFar(fcall)
             "prog8_lib_stringcompare" -> funcStringCompare(fcall)
+            "prog8_lib_square_byte" -> funcSquare(fcall, DataType.UBYTE)
+            "prog8_lib_square_word" -> funcSquare(fcall, DataType.UWORD)
             else -> throw AssemblyError("missing asmgen for builtin func ${fcall.name}")
         }
 
         return BuiltinFunctions.getValue(fcall.name).returnType
     }
 
+    private fun funcSquare(fcall: PtBuiltinFunctionCall, resultType: DataType) {
+        // square of word value is faster with dedicated routine, square of byte just use the regular multiplication routine.
+        when (resultType) {
+            DataType.UBYTE -> {
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)
+                asmgen.out("  tay  |  jsr  math.multiply_bytes")
+            }
+            DataType.UWORD -> {
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+                asmgen.out("  jsr  math.square")
+            }
+            else -> {
+                throw AssemblyError("optimized square only for integer types")
+            }
+        }
+    }
+
     private fun funcDivmod(fcall: PtBuiltinFunctionCall) {
         assignAsmGen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A, false)
         asmgen.saveRegisterStack(CpuRegister.A, false)
@@ -89,8 +119,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         val var3name = asmgen.asmVariableName(fcall.args[3] as PtIdentifier)
         val divisionTarget = AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UBYTE, fcall.definingISub(), fcall.args[2].position, var2name)
         val remainderTarget = AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UBYTE, fcall.definingISub(), fcall.args[3].position, var3name)
-        assignAsmGen.assignRegisterByte(remainderTarget, CpuRegister.A, false)
-        assignAsmGen.assignRegisterByte(divisionTarget, CpuRegister.Y, false)
+        assignAsmGen.assignRegisterByte(remainderTarget, CpuRegister.A, false, false)
+        assignAsmGen.assignRegisterByte(divisionTarget, CpuRegister.Y, false, false)
     }
 
     private fun funcDivmodW(fcall: PtBuiltinFunctionCall) {
@@ -162,7 +192,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         asmgen.out("  sta  (+)+0 |  sty  (+)+1")
         asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
         asmgen.out("""
-            jsr  cx16.jsrfar
+            jsr  cx16.JSRFAR
 +           .word  0
 +           .byte  0""")
         // note that by convention the values in A+Y registers are now the return value of the call.
@@ -252,11 +282,11 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         when(fcall.args[0].type) {
             DataType.UBYTE -> {
                 asmgen.out("  ldy  #0 |  jsr  prog8_lib.func_sqrt16_into_A")
-                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false, false)
             }
             DataType.UWORD -> {
                 asmgen.out("  jsr  prog8_lib.func_sqrt16_into_A")
-                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false)
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false, false)
             }
             DataType.FLOAT -> {
                 asmgen.out("  jsr  floats.func_sqrt_into_FAC1")
@@ -590,6 +620,85 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         asmgen.assignExpressionToVariable(indexer.index, "prog8_lib.${operation}_array_u${dt}._arg_index", DataType.UBYTE)
     }
 
+    private fun funcSetLsbMsb(fcall: PtBuiltinFunctionCall, msb: Boolean) {
+        val target: AsmAssignTarget
+        when(fcall.args[0]) {
+            is PtIdentifier -> {
+                val varname = asmgen.asmVariableName(fcall.args[0] as PtIdentifier) + if(msb) "+1" else ""
+                target = AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UBYTE, fcall.definingSub(), fcall.position, variableAsmName = varname)
+            }
+            is PtNumber -> {
+                val num = (fcall.args[0] as PtNumber).number + if(msb) 1 else 0
+                val mem = PtMemoryByte(fcall.position)
+                mem.add(PtNumber(DataType.UBYTE, num, fcall.position))
+                target = AsmAssignTarget(TargetStorageKind.MEMORY, asmgen, DataType.UBYTE, fcall.definingSub(), fcall.position, memory = mem)
+            }
+            is PtAddressOf -> {
+                val mem = PtMemoryByte(fcall.position)
+                if(msb) {
+                    val address = PtBinaryExpression("+", DataType.UWORD, fcall.args[0].position)
+                    address.add(fcall.args[0])
+                    address.add(PtNumber(address.type, 1.0, fcall.args[0].position))
+                    mem.add(address)
+                } else {
+                    mem.add(fcall.args[0])
+                }
+                target = AsmAssignTarget(TargetStorageKind.MEMORY, asmgen, DataType.UBYTE, fcall.definingSub(), fcall.position, memory = mem)
+            }
+            is PtArrayIndexer -> {
+                val indexer = fcall.args[0] as PtArrayIndexer
+                require(!indexer.usesPointerVariable)
+                val elementSize: Int
+                val msbAdd: Int
+                if(indexer.splitWords) {
+                    val arrayVariable = indexer.variable
+                    indexer.children[0] = PtIdentifier(arrayVariable.name + if(msb) "_msb" else "_lsb", DataType.ARRAY_UB, arrayVariable.position)
+                    indexer.children[0].parent = indexer
+                    elementSize = 1
+                    msbAdd = 0
+                } else {
+                    elementSize = 2
+                    msbAdd = if(msb) 1 else 0
+                }
+
+                // double the index because of word array (if not split), add one if msb (if not split)
+                val constIndexNum = (indexer.index as? PtNumber)?.number
+                if(constIndexNum!=null) {
+                    indexer.children[1] = PtNumber(indexer.index.type, constIndexNum*elementSize + msbAdd, indexer.position)
+                    indexer.children[1].parent = indexer
+                } else {
+                    val multipliedIndex: PtExpression
+                    if(elementSize==1) {
+                        multipliedIndex = indexer.index
+                    } else {
+                        multipliedIndex = PtBinaryExpression("<<", indexer.index.type, indexer.position)
+                        multipliedIndex.add(indexer.index)
+                        multipliedIndex.add(PtNumber(DataType.UBYTE, 1.0, indexer.position))
+                    }
+                    if(msbAdd>0) {
+                        val msbIndex = PtBinaryExpression("+", indexer.index.type, indexer.position)
+                        msbIndex.add(multipliedIndex)
+                        msbIndex.add(PtNumber(DataType.UBYTE, msbAdd.toDouble(), indexer.position))
+                        indexer.children[1] = msbIndex
+                        msbIndex.parent = indexer
+                    } else {
+                        indexer.children[1] = multipliedIndex
+                        multipliedIndex.parent=indexer
+                    }
+                }
+                target = AsmAssignTarget(TargetStorageKind.ARRAY, asmgen, DataType.UBYTE, fcall.definingSub(), fcall.position, array = indexer)
+            }
+            else -> throw AssemblyError("setlsb/setmsb on weird target ${fcall.args[0]}")
+        }
+
+        if(fcall.args[1].asConstInteger() == 0) {
+            assignAsmGen.assignConstantByte(target, 0)
+        } else {
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A, false)
+            assignAsmGen.assignRegisterByte(target, CpuRegister.A, false, false)
+        }
+    }
+
     private fun funcSgn(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         translateArguments(fcall, scope)
         val dt = fcall.args.single().type
@@ -601,7 +710,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             DataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_into_A")
             else -> throw AssemblyError("weird type $dt")
         }
-        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, true)
+        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, true, true)
     }
 
     private fun funcAnyAll(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
@@ -614,7 +723,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             in SplitWordArrayTypes -> TODO("split word any/all")
             else -> throw AssemblyError("weird type $dt")
         }
-        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, dt in SignedDatatypes)
+        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, dt in SignedDatatypes, true)
     }
 
     private fun funcAbs(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
@@ -623,7 +732,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         when (dt) {
             DataType.BYTE -> {
                 asmgen.out("  jsr  prog8_lib.abs_b_into_A")
-                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A,false)
+                assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A,false, true)
             }
             DataType.WORD -> {
                 asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
@@ -723,11 +832,11 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         asmgen.out("""
                             ldy  #0
                             lda  ($varname),y
-                            pha
+                            tax
                             iny
                             lda  ($varname),y
                             tay
-                            pla""")
+                            txa""")
                     }
                 } else fallback()
             }
@@ -740,11 +849,11 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(result.second, RegisterOrPair.Y)
                     asmgen.out("""
                         lda  ($varname),y
-                        pha
+                        tax
                         iny
                         lda  ($varname),y
                         tay
-                        pla""")
+                        txa""")
                 } else fallback()
             }
             else -> fallback()
@@ -771,7 +880,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 assignAsmGen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A, signed)    // value
                 asmgen.out("  jsr  prog8_lib.func_clamp_${fcall.type.toString().lowercase()}")
                 val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-                assignAsmGen.assignRegisterByte(targetReg, CpuRegister.A, signed)
+                assignAsmGen.assignRegisterByte(targetReg, CpuRegister.A, signed, true)
             }
             in WordDatatypes -> {
                 assignAsmGen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W1", fcall.args[1].type)  // minimum
@@ -787,55 +896,59 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
 
     private fun funcMin(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type in SignedDatatypes
-        if(fcall.type in ByteDatatypes) {
-            asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_B1", fcall.type)     // right
-            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)          // left
-            asmgen.out("  cmp  P8ZP_SCRATCH_B1")
-            if(signed) asmgen.out("  bmi  +") else asmgen.out("  bcc  +")
-            asmgen.out("""
-                lda  P8ZP_SCRATCH_B1
-+""")
-            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-            asmgen.assignRegister(RegisterOrPair.A, targetReg)
-        } else if(fcall.type in WordDatatypes) {
-            asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
-            asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
-            if(signed) {
+        when (fcall.type) {
+            in ByteDatatypes -> {
+                asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_B1", fcall.type)     // right
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)          // left
+                asmgen.out("  cmp  P8ZP_SCRATCH_B1")
+                if(signed) asmgen.out("  bmi  +") else asmgen.out("  bcc  +")
                 asmgen.out("""
-                    lda  P8ZP_SCRATCH_W1
-                    ldy  P8ZP_SCRATCH_W1+1
-                    cmp  P8ZP_SCRATCH_W2
-                    tya
-                    sbc  P8ZP_SCRATCH_W2+1
-                    bvc  +
-                    eor  #$80
-+                   bpl  +
-                    lda  P8ZP_SCRATCH_W1                   
-                    ldy  P8ZP_SCRATCH_W1+1
-                    jmp  ++
-+                   lda  P8ZP_SCRATCH_W2
-                    ldy  P8ZP_SCRATCH_W2+1
-+""")
-            } else {
-                asmgen.out("""
-                    lda  P8ZP_SCRATCH_W1+1
-                    cmp  P8ZP_SCRATCH_W2+1
-                    bcc  ++
-                    bne  +
-                    lda  P8ZP_SCRATCH_W1
-                    cmp  P8ZP_SCRATCH_W2
-                    bcc  ++
-+                   lda  P8ZP_SCRATCH_W2
-                    ldy  P8ZP_SCRATCH_W2+1
-                    jmp  ++
-+                   lda  P8ZP_SCRATCH_W1
-                    ldy  P8ZP_SCRATCH_W1+1
-+""")
+                    lda  P8ZP_SCRATCH_B1
+    +""")
+                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
+                asmgen.assignRegister(RegisterOrPair.A, targetReg)
+            }
+            in WordDatatypes -> {
+                asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
+                asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
+                if(signed) {
+                    asmgen.out("""
+                        lda  P8ZP_SCRATCH_W1
+                        ldy  P8ZP_SCRATCH_W1+1
+                        cmp  P8ZP_SCRATCH_W2
+                        tya
+                        sbc  P8ZP_SCRATCH_W2+1
+                        bvc  +
+                        eor  #$80
+    +                   bpl  +
+                        lda  P8ZP_SCRATCH_W1                   
+                        ldy  P8ZP_SCRATCH_W1+1
+                        jmp  ++
+    +                   lda  P8ZP_SCRATCH_W2
+                        ldy  P8ZP_SCRATCH_W2+1
+    +""")
+                } else {
+                    asmgen.out("""
+                        lda  P8ZP_SCRATCH_W1+1
+                        cmp  P8ZP_SCRATCH_W2+1
+                        bcc  ++
+                        bne  +
+                        lda  P8ZP_SCRATCH_W1
+                        cmp  P8ZP_SCRATCH_W2
+                        bcc  ++
+    +                   lda  P8ZP_SCRATCH_W2
+                        ldy  P8ZP_SCRATCH_W2+1
+                        jmp  ++
+    +                   lda  P8ZP_SCRATCH_W1
+                        ldy  P8ZP_SCRATCH_W1+1
+    +""")
+                }
+                val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
+                asmgen.assignRegister(RegisterOrPair.AY, targetReg)
+            }
+            else -> {
+                throw AssemblyError("min float not supported")
             }
-            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
-            asmgen.assignRegister(RegisterOrPair.AY, targetReg)
-        } else {
-            throw AssemblyError("min float not supported")
         }
     }
 
@@ -967,17 +1080,17 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         } else {
             when(resultRegister) {
                 null, RegisterOrPair.A -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  tya")
                 }
                 RegisterOrPair.X -> {
                     asmgen.out("  pha")
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AX)
                     asmgen.out("  pla")
                 }
                 RegisterOrPair.Y -> {
                     asmgen.out("  pha")
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  pla")
                 }
                 else -> throw AssemblyError("invalid reg")
@@ -1011,37 +1124,55 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 else -> throw AssemblyError("invalid reg")
             }
         } else {
-            when(resultRegister) {
+            if(arg is PtArrayIndexer && resultRegister in setOf(null, RegisterOrPair.A, RegisterOrPair.Y, RegisterOrPair.X)) {
+                // just read the lsb byte out of the word array
+                val arrayVar = if(arg.splitWords) asmgen.asmVariableName(arg.variable)+"_lsb" else asmgen.asmVariableName(arg.variable)
+                when(resultRegister) {
+                    null, RegisterOrPair.A -> {
+                        asmgen.loadScaledArrayIndexIntoRegister(arg, arg.type, CpuRegister.Y)
+                        asmgen.out("  lda  $arrayVar,y")
+                    }
+                    RegisterOrPair.Y -> {
+                        asmgen.loadScaledArrayIndexIntoRegister(arg, arg.type, CpuRegister.X)
+                        asmgen.out("  lda  $arrayVar,x")
+                    }
+                    RegisterOrPair.X -> {
+                        asmgen.loadScaledArrayIndexIntoRegister(arg, arg.type, CpuRegister.Y)
+                        asmgen.out("  ldx  $arrayVar,y")
+                    }
+                    else -> throw AssemblyError("invalid reg")
+                }
+            } else  when(resultRegister) {
                 null, RegisterOrPair.A -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
                     //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
                     //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
                 }
                 RegisterOrPair.X -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.XY)
                     // NOTE: we rely on the fact that the above assignment to XY, assigns the Lsb to X as the last instruction.
                     //       this is required because the compiler assumes the status bits are set according to what X is (lsb)
                     //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
                 }
                 RegisterOrPair.Y -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  tay |  cpy  #0")
                 }
                 RegisterOrPair.AY -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  ldy  #0 |  cmp  #0")
                 }
                 RegisterOrPair.AX -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AX)
                     asmgen.out("  ldx  #0 |  cmp  #0")
                 }
                 RegisterOrPair.XY -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.XY)
                     asmgen.out("  ldy  #0 |  cpx  #0")
                 }
                 in Cx16VirtualRegisters -> {
-                    asmgen.assignExpressionToRegister(fcall.args.single(), resultRegister)
+                    asmgen.assignExpressionToRegister(arg, resultRegister)
                     val zero = PtNumber(DataType.UBYTE, 0.0, Position.DUMMY)
                     zero.parent=fcall
                     assignAsmGen.assignExpressionToVariable(zero, "cx16.${resultRegister.toString().lowercase()}H", DataType.UBYTE)

codeGenCpu6502/src/prog8/codegen/cpu6502/ForLoopsAsmGen.kt

@@ -255,11 +255,11 @@ $loopLabel""")
                 sec
                 sbc  #<${stepsize.absoluteValue}
                 sta  $varname
-                pha
+                tax
                 lda  $varname+1
                 sbc  #>${stepsize.absoluteValue}
                 sta  $varname+1
-                pla
+                txa
 $modifiedLabel2 cmp  #0    ; modified 
                 lda  $varname+1
 $modifiedLabel  sbc  #0    ; modified

codeGenCpu6502/src/prog8/codegen/cpu6502/FunctionCallAsmGen.kt

@@ -69,37 +69,59 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
         // remember: dealing with the X register and/or dealing with return values is the responsibility of the caller
     }
 
-    private fun argumentsViaRegisters(sub: PtAsmSub, call: PtFunctionCall) {
-        if(sub.parameters.size==1) {
-            argumentViaRegister(sub, IndexedValue(0, sub.parameters.single().second), call.args[0])
-        } else {
-            if(asmsub6502ArgsHaveRegisterClobberRisk(call.args, sub.parameters)) {
-                registerArgsViaCpuStackEvaluation(call, sub)
-            } else {
-                asmsub6502ArgsEvalOrder(sub).forEach {
-                    val param = sub.parameters[it]
-                    val arg = call.args[it]
-                    argumentViaRegister(sub, IndexedValue(it, param.second), arg)
-                }
+
+    private fun usesOtherRegistersWhileEvaluating(arg: PtExpression): Boolean {
+        return when(arg) {
+            is PtBuiltinFunctionCall -> {
+                if (arg.name == "lsb" || arg.name == "msb")
+                    return usesOtherRegistersWhileEvaluating(arg.args[0])
+                if (arg.name == "mkword")
+                    return usesOtherRegistersWhileEvaluating(arg.args[0]) || usesOtherRegistersWhileEvaluating(arg.args[1])
+                return !arg.isSimple()
             }
+            is PtAddressOf -> false
+            is PtIdentifier -> false
+            is PtMachineRegister -> false
+            is PtMemoryByte -> false
+            is PtNumber -> false
+            else -> true
         }
     }
 
-    private fun registerArgsViaCpuStackEvaluation(call: PtFunctionCall, callee: PtAsmSub) {
-        // 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.
+    private fun argumentsViaRegisters(sub: PtAsmSub, call: PtFunctionCall) {
+        val registersUsed = mutableListOf<RegisterOrStatusflag>();
 
-        if(callee.parameters.isEmpty())
-            return
+        fun usedA() = registersUsed.any {it.registerOrPair==RegisterOrPair.A || it.registerOrPair==RegisterOrPair.AX || it.registerOrPair==RegisterOrPair.AY}
+        fun usedX() = registersUsed.any {it.registerOrPair==RegisterOrPair.X || it.registerOrPair==RegisterOrPair.AX || it.registerOrPair==RegisterOrPair.XY}
+        fun usedY() = registersUsed.any {it.registerOrPair==RegisterOrPair.Y || it.registerOrPair==RegisterOrPair.AY || it.registerOrPair==RegisterOrPair.XY}
 
-        // use the cpu hardware stack as intermediate storage for the arguments.
-        val argOrder = asmsub6502ArgsEvalOrder(callee)
-        argOrder.reversed().forEach {
-            asmgen.pushCpuStack(callee.parameters[it].second.type, call.args[it])
-        }
-        argOrder.forEach {
-            val param = callee.parameters[it]
-            asmgen.popCpuStack(callee, param.second, param.first)
+        if(sub.parameters.size==1) {
+            argumentViaRegister(sub, IndexedValue(0, sub.parameters.single().second), call.args[0])
+        } else {
+            val optimalEvalOrder = asmsub6502ArgsEvalOrder(sub)
+            optimalEvalOrder.forEach {
+                val param = sub.parameters[it]
+                val arg = call.args[it]
+                registersUsed += if(usesOtherRegistersWhileEvaluating(arg)) {
+                    if(!registersUsed.any{it.statusflag!=null || it.registerOrPair in CpuRegisters})
+                        argumentViaRegister(sub, IndexedValue(it, param.second), arg)
+                    else if(registersUsed.any {it.statusflag!=null}) {
+                        throw AssemblyError("call argument evaluation problem: can't save cpu statusregister parameter ${call.position}")
+                    }
+                    else {
+                        if(usedX()) asmgen.saveRegisterStack(CpuRegister.X, false)
+                        if(usedY()) asmgen.saveRegisterStack(CpuRegister.Y, false)
+                        if(usedA()) asmgen.saveRegisterStack(CpuRegister.A, false)
+                        val used = argumentViaRegister(sub, IndexedValue(it, param.second), arg)
+                        if(usedA()) asmgen.restoreRegisterStack(CpuRegister.A, false)
+                        if(usedY()) asmgen.restoreRegisterStack(CpuRegister.Y, true)
+                        if(usedX()) asmgen.restoreRegisterStack(CpuRegister.X, true)
+                        used
+                    }
+                } else {
+                    argumentViaRegister(sub, IndexedValue(it, param.second), arg)
+                }
+            }
         }
     }
 
@@ -112,7 +134,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
         asmgen.assignExpressionToVariable(value, varName, parameter.type)
     }
 
-    private fun argumentViaRegister(sub: IPtSubroutine, parameter: IndexedValue<PtSubroutineParameter>, value: PtExpression, registerOverride: RegisterOrPair? = null) {
+    private fun argumentViaRegister(sub: IPtSubroutine, parameter: IndexedValue<PtSubroutineParameter>, value: PtExpression, registerOverride: RegisterOrPair? = null): RegisterOrStatusflag {
         // pass argument via a register parameter
         if(!isArgumentTypeCompatible(value.type, parameter.value.type))
             throw AssemblyError("argument type incompatible")
@@ -141,6 +163,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                     }
                     is PtIdentifier -> {
                         val sourceName = asmgen.asmVariableName(value)
+                        // note: cannot use X register here to store A because it might be used for other arguments
                         asmgen.out("""
                             pha
                             clc
@@ -155,6 +178,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                     }
                 }
             } else throw AssemblyError("can only use Carry as status flag parameter")
+            return RegisterOrStatusflag(null, statusflag)
         }
         else {
             // via register or register pair
@@ -187,6 +211,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                 }
                 asmgen.translateNormalAssignment(AsmAssignment(src, target, program.memsizer, Position.DUMMY), scope)
             }
+            return RegisterOrStatusflag(register, null)
         }
     }
 
@@ -208,3 +233,5 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
         return false
     }
 }
+
+

codeGenCpu6502/src/prog8/codegen/cpu6502/ProgramAndVarsGen.kt

@@ -428,12 +428,12 @@ internal class ProgramAndVarsGen(
                 asmgen.out("""
                     lda  #<${name}_init_value
                     ldy  #>${name}_init_value
-                    sta  cx16.r0L
-                    sty  cx16.r0H
+                    sta  cx16.r0
+                    sty  cx16.r0+1
                     lda  #<${name}
                     ldy  #>${name}
-                    sta  cx16.r1L
-                    sty  cx16.r1H
+                    sta  cx16.r1
+                    sty  cx16.r1+1
                     lda  #<$size
                     ldy  #>$size
                     jsr  sys.memcopy""")
@@ -452,7 +452,6 @@ internal class ProgramAndVarsGen(
         }
 
         asmgen.out("""+        
-                    ldx  #127       ; init estack ptr (half page)
                     clv
                     clc""")
     }
@@ -498,7 +497,15 @@ internal class ProgramAndVarsGen(
         for ((scopedName, zpvar) in zpVariables) {
             if (scopedName.startsWith("cx16.r"))
                 continue        // The 16 virtual registers of the cx16 are not actual variables in zp, they're memory mapped
-            asmgen.out("${scopedName.substringAfterLast('.')} \t= ${zpvar.address} \t; zp ${zpvar.dt}")
+            val variable = symboltable.flat.getValue(scopedName) as StStaticVariable
+            if(variable.dt in SplitWordArrayTypes) {
+                val lsbAddr = zpvar.address
+                val msbAddr = zpvar.address + (zpvar.size/2).toUInt()
+                asmgen.out("${scopedName.substringAfterLast('.')}_lsb \t= $lsbAddr \t; zp ${zpvar.dt} (lsbs)")
+                asmgen.out("${scopedName.substringAfterLast('.')}_msb \t= $msbAddr \t; zp ${zpvar.dt} (msbs)")
+            } else {
+                asmgen.out("${scopedName.substringAfterLast('.')} \t= ${zpvar.address} \t; zp ${zpvar.dt}")
+            }
         }
     }
 

codeGenCpu6502/src/prog8/codegen/cpu6502/assignment/AnyExprAsmGen.kt

@@ -1,7 +1,10 @@
 package prog8.codegen.cpu6502.assignment
 
 import prog8.code.ast.PtBinaryExpression
+import prog8.code.ast.PtExpression
 import prog8.code.core.*
+import prog8.code.target.C64Target
+import prog8.code.target.Cx16Target
 import prog8.codegen.cpu6502.AsmGen6502Internal
 
 //
@@ -20,13 +23,13 @@ internal class AnyExprAsmGen(
                     return assignByteBinExpr(expr, assign)
                 if (expr.left.type in WordDatatypes && expr.right.type in WordDatatypes) {
                     require(expr.operator in ComparisonOperators)
-                    TODO("words operands comparison -> byte")
+                    throw AssemblyError("words operands comparison -> byte, should have been handled by assignOptimizedComparisonWords()")
                 }
                 if (expr.left.type==DataType.FLOAT && expr.right.type==DataType.FLOAT) {
                     require(expr.operator in ComparisonOperators)
                     return assignFloatBinExpr(expr, assign)
                 }
-                TODO("weird expr operand types")
+                throw AssemblyError("weird expr operand types: ${expr.left.type} and {${expr.right.type}")
             }
             in WordDatatypes -> {
                 require(expr.left.type in WordDatatypes && expr.right.type in WordDatatypes) {
@@ -180,37 +183,25 @@ internal class AnyExprAsmGen(
     private fun assignFloatBinExpr(expr: PtBinaryExpression, assign: AsmAssignment): Boolean {
         when(expr.operator) {
             "+" -> {
-                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC1, true)
-                if(!expr.right.isSimple()) asmgen.pushFAC1()
-                asmgen.assignExpressionToRegister(expr.right, RegisterOrPair.FAC2, true)
-                if(!expr.right.isSimple()) asmgen.popFAC1()
+                assignFloatOperandsToFACandARG(expr.left, expr.right)
                 asmgen.out("  jsr  floats.FADDT")
                 asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
                 return true
             }
             "-" -> {
-                asmgen.assignExpressionToRegister(expr.right, RegisterOrPair.FAC1, true)
-                if(!expr.left.isSimple()) asmgen.pushFAC1()
-                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC2, true)
-                if(!expr.left.isSimple()) asmgen.popFAC1()
+                assignFloatOperandsToFACandARG(expr.right, expr.left)
                 asmgen.out("  jsr  floats.FSUBT")
                 asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
                 return true
             }
             "*" -> {
-                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC1, true)
-                if(!expr.right.isSimple()) asmgen.pushFAC1()
-                asmgen.assignExpressionToRegister(expr.right, RegisterOrPair.FAC2, true)
-                if(!expr.right.isSimple()) asmgen.popFAC1()
+                assignFloatOperandsToFACandARG(expr.left, expr.right)
                 asmgen.out("  jsr  floats.FMULTT")
                 asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
                 return true
             }
             "/" -> {
-                asmgen.assignExpressionToRegister(expr.right, RegisterOrPair.FAC1, true)
-                if(!expr.left.isSimple()) asmgen.pushFAC1()
-                asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC2, true)
-                if(!expr.left.isSimple()) asmgen.popFAC1()
+                assignFloatOperandsToFACandARG(expr.right, expr.left)
                 asmgen.out("  jsr  floats.FDIVT")
                 asmgen.assignRegister(RegisterOrPair.FAC1, assign.target)
                 return true
@@ -255,6 +246,31 @@ internal class AnyExprAsmGen(
         }
     }
 
+    private fun assignFloatOperandsToFACandARG(left: PtExpression, right: PtExpression) {
+        when(asmgen.options.compTarget.name) {
+            C64Target.NAME -> {
+                // c64 has a quirk: always make sure FAC2 is loaded last (done using CONUPK)  otherwise the result will be corrupt on C64
+                // this requires some more forced copying around of float values in certain cases
+                if (right.isSimple()) {
+                    asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                    asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC2, true)
+                } else {
+                    asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC1, true)
+                    asmgen.pushFAC1()
+                    asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                    asmgen.popFAC2()
+                }
+            }
+            Cx16Target.NAME -> {
+                asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                if (!right.isSimple()) asmgen.pushFAC1()
+                asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC2, true)
+                if (!right.isSimple()) asmgen.popFAC1()
+            }
+            else -> TODO("don't know how to evaluate float expression for selected compilation target")
+        }
+    }
+
     private fun setupFloatComparisonFAC1vsVarAY(expr: PtBinaryExpression) {
         asmgen.assignExpressionToRegister(expr.left, RegisterOrPair.FAC1, true)
         if(!expr.right.isSimple()) asmgen.pushFAC1()

codeGenCpu6502/test/TestCodegen.kt

@@ -22,6 +22,7 @@ class TestCodegen: FunSpec({
             CbmPrgLauncherType.NONE,
             ZeropageType.DONTUSE,
             zpReserved = emptyList(),
+            zpAllowed = CompilationOptions.AllZeropageAllowed,
             floats = true,
             noSysInit = false,
             compTarget = target,
@@ -44,7 +45,7 @@ class TestCodegen: FunSpec({
 //}
         val codegen = AsmGen6502(prefixSymbols = false)
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("pi", DataType.UBYTE, ZeropageWish.DONTCARE, PtNumber(DataType.UBYTE, 0.0, Position.DUMMY), null, Position.DUMMY))
         sub.add(PtVariable("particleX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
@@ -94,7 +95,7 @@ class TestCodegen: FunSpec({
         program.add(block)
 
         // define the "cx16.r0" virtual register
-        val cx16block = PtBlock("cx16", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val cx16block = PtBlock("cx16", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         cx16block.add(PtMemMapped("r0", DataType.UWORD, 100u, null, Position.DUMMY))
         program.add(cx16block)
 

codeGenExperimental/build.gradle

@@ -36,10 +36,10 @@ dependencies {
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
 }

codeGenIntermediate/build.gradle

@@ -30,16 +30,16 @@ dependencies {
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
+    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.7.2'
 }
 
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
     test {

codeGenIntermediate/src/prog8/codegen/intermediate/BuiltinFuncGen.kt

@@ -33,24 +33,45 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "peekw" -> funcPeekW(call)
             "poke" -> funcPoke(call)
             "pokew" -> funcPokeW(call)
-            "pokemon" -> ExpressionCodeResult.EMPTY     // easter egg function
+            "pokemon" -> funcPokemon(call)
             "mkword" -> funcMkword(call)
             "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(call)
             "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(call)
             "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(call)
             "sort" -> funcSort(call)
             "reverse" -> funcReverse(call)
+            "setlsb" -> funcSetLsbMsb(call, false)
+            "setmsb" -> funcSetLsbMsb(call, true)
             "rol" -> funcRolRor(Opcode.ROXL, call)
             "ror" -> funcRolRor(Opcode.ROXR, call)
             "rol2" -> funcRolRor(Opcode.ROL, call)
             "ror2" -> funcRolRor(Opcode.ROR, call)
             "prog8_lib_stringcompare" -> funcStringCompare(call)
+            "prog8_lib_square_byte" -> funcSquare(call, IRDataType.BYTE)
+            "prog8_lib_square_word" -> funcSquare(call, IRDataType.WORD)
             else -> throw AssemblyError("missing builtinfunc for ${call.name}")
         }
     }
 
+    private fun funcSquare(call: PtBuiltinFunctionCall, resultType: IRDataType): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val valueTr = exprGen.translateExpression(call.args[0])
+        addToResult(result, valueTr, valueTr.resultReg, valueTr.resultFpReg)
+        return if(resultType==IRDataType.FLOAT) {
+            val resultFpReg = codeGen.registers.nextFreeFloat()
+            addInstr(result, IRInstruction(Opcode.SQUARE, resultType, fpReg1 = resultFpReg, fpReg2 = valueTr.resultFpReg), null)
+            ExpressionCodeResult(result, resultType, -1, resultFpReg)
+        }
+        else {
+            val resultReg = codeGen.registers.nextFree()
+            addInstr(result, IRInstruction(Opcode.SQUARE, resultType, reg1 = resultReg, reg2 = valueTr.resultReg), null)
+            ExpressionCodeResult(result, resultType, resultReg, -1)
+        }
+    }
+
     private fun funcCallfar(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 3), null)
         val bankTr = exprGen.translateExpression(call.args[0])
         val addressTr = exprGen.translateExpression(call.args[1])
         val argumentwordTr = exprGen.translateExpression(call.args[2])
@@ -92,6 +113,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     private fun funcStringCompare(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
         val left  = exprGen.translateExpression(call.args[0])
         val right = exprGen.translateExpression(call.args[1])
         addToResult(result, left, left.resultReg, -1)
@@ -126,6 +148,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 else -> throw IllegalArgumentException("weird type")
             }
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
@@ -147,6 +170,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 else -> throw IllegalArgumentException("weird type")
             }
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
@@ -290,6 +314,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 else -> throw IllegalArgumentException("weird type to reverse")
             }
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
@@ -313,6 +338,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 else -> throw IllegalArgumentException("weird type to sort")
             }
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
         val tr = exprGen.translateExpression(call.args[0])
         addToResult(result, tr, tr.resultReg, -1)
         val lengthReg = codeGen.registers.nextFree()
@@ -323,19 +349,25 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     private fun funcMkword(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
-        val msbTr = exprGen.translateExpression(call.args[0])
-        addToResult(result, msbTr, msbTr.resultReg, -1)
-        val lsbTr = exprGen.translateExpression(call.args[1])
-        addToResult(result, lsbTr, lsbTr.resultReg, -1)
         val resultReg = codeGen.registers.nextFree()
-        result += IRCodeChunk(null, null).also {
-            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=resultReg, reg2 = lsbTr.resultReg, reg3 = msbTr.resultReg)
+        if((call.args[0] as? PtNumber)?.number == 0.0) {
+            // msb is 0, use EXT
+            val lsbTr = exprGen.translateExpression(call.args[1])
+            addToResult(result, lsbTr, lsbTr.resultReg, -1)
+            addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=resultReg, reg2 = lsbTr.resultReg), null)
+        } else {
+            val msbTr = exprGen.translateExpression(call.args[0])
+            addToResult(result, msbTr, msbTr.resultReg, -1)
+            val lsbTr = exprGen.translateExpression(call.args[1])
+            addToResult(result, lsbTr, lsbTr.resultReg, -1)
+            addInstr(result, IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=resultReg, reg2 = msbTr.resultReg, reg3 = lsbTr.resultReg), null)
         }
         return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
     }
 
     private fun funcClamp(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
+        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 3), null)
         val type = irType(call.type)
         val valueTr = exprGen.translateExpression(call.args[0])
         val minimumTr = exprGen.translateExpression(call.args[1])
@@ -520,6 +552,59 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         }
     }
 
+    private fun funcPokemon(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val address = call.args[0]
+
+        fun pokeM(result: MutableList<IRCodeChunkBase>, address: Int, value: PtExpression) {
+            if(codeGen.isZero(value)) {
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.STOREZM, IRDataType.BYTE, address = address)
+                }
+            } else {
+                val tr = exprGen.translateExpression(value)
+                addToResult(result, tr, tr.resultReg, -1)
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = tr.resultReg, address = address)
+                }
+            }
+        }
+
+        fun pokeI(result: MutableList<IRCodeChunkBase>, register: Int, value: PtExpression) {
+            if(codeGen.isZero(value)) {
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.STOREZI, IRDataType.BYTE, reg1 = register)
+                }
+            } else {
+                val valueTr = exprGen.translateExpression(value)
+                addToResult(result, valueTr, valueTr.resultReg, -1)
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.STOREI, IRDataType.BYTE, reg1 = valueTr.resultReg, reg2 = register)
+                }
+            }
+        }
+
+        return if(address is PtNumber) {
+            val resultRegister = codeGen.registers.nextFree()
+            val addressNum = address.number.toInt()
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1 = resultRegister, address = addressNum)
+            }
+            pokeM(result, addressNum, call.args[1])
+            ExpressionCodeResult(result, IRDataType.BYTE, resultRegister, -1)
+        } else {
+            val addressTr = exprGen.translateExpression(address)
+            addToResult(result, addressTr, addressTr.resultReg, -1)
+            val resultReg = codeGen.registers.nextFree()
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.LOADI, IRDataType.BYTE, reg1 = resultReg, reg2 = addressTr.resultReg)
+            }
+            pokeI(result, addressTr.resultReg, call.args[1])
+            ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
+        }
+    }
+
+
     private fun funcMemory(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val name = (call.args[0] as PtString).value
         val code = IRCodeChunk(null, null)
@@ -531,6 +616,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
     private fun funcLsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         return exprGen.translateExpression(call.args.single())
         // note: if a word result is needed, the upper byte is cleared by the typecast that follows. No need to do it here.
+        // TODO to be more strict, maybe we *should* introduce a new result register that is of type .b?
     }
 
     private fun funcMsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {
@@ -557,6 +643,122 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         return ExpressionCodeResult(result, vmDt, -1, -1)
     }
 
+    private fun funcSetLsbMsb(call: PtBuiltinFunctionCall, msb: Boolean): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val target = call.args[0]
+        val isConstZeroValue = call.args[1].asConstInteger()==0
+        when(target) {
+            is PtIdentifier -> {
+                if(isConstZeroValue) {
+                    result += IRCodeChunk(null, null).also {
+                        val pointerReg = codeGen.registers.nextFree()
+                        it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1 = pointerReg, labelSymbol = target.name)
+                        if (msb)
+                            it += IRInstruction(Opcode.INC, IRDataType.WORD, reg1 = pointerReg)
+                        it += IRInstruction(Opcode.STOREZI, IRDataType.BYTE, reg1 = pointerReg)
+                    }
+                } else {
+                    val valueTr = exprGen.translateExpression(call.args[1])
+                    addToResult(result, valueTr, valueTr.resultReg, -1)
+                    result += IRCodeChunk(null, null).also {
+                        val pointerReg = codeGen.registers.nextFree()
+                        it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1 = pointerReg, labelSymbol = target.name)
+                        if (msb)
+                            it += IRInstruction(Opcode.INC, IRDataType.WORD, reg1 = pointerReg)
+                        it += IRInstruction(Opcode.STOREI, IRDataType.BYTE, reg1 = valueTr.resultReg, reg2 = pointerReg)
+                    }
+                }
+            }
+            is PtArrayIndexer -> {
+                require(!target.usesPointerVariable)
+                if(target.splitWords) {
+                    // lsb/msb in split arrays, element index 'size' is always 1
+                    val constIndex = target.index.asConstInteger()
+                    val varName = target.variable.name + if(msb) "_msb" else "_lsb"
+                    if(isConstZeroValue) {
+                        if(constIndex!=null) {
+                            val offsetReg = codeGen.registers.nextFree()
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=offsetReg, immediate = constIndex)
+                                it += IRInstruction(Opcode.STOREZX, IRDataType.BYTE, reg1=offsetReg, labelSymbol = varName)
+                            }
+                        } else {
+                            val indexTr = exprGen.translateExpression(target.index)
+                            addToResult(result, indexTr, indexTr.resultReg, -1)
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.STOREZX, IRDataType.BYTE, reg1=indexTr.resultReg, labelSymbol = varName)
+                            }
+                        }
+                    } else {
+                        val valueTr = exprGen.translateExpression(call.args[1])
+                        addToResult(result, valueTr, valueTr.resultReg, -1)
+                        if(constIndex!=null) {
+                            val offsetReg = codeGen.registers.nextFree()
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=offsetReg, immediate = constIndex)
+                                it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1=valueTr.resultReg, reg2=offsetReg, labelSymbol = varName)
+                            }
+                        } else {
+                            val indexTr = exprGen.translateExpression(target.index)
+                            addToResult(result, indexTr, indexTr.resultReg, -1)
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1=valueTr.resultReg, reg2=indexTr.resultReg, labelSymbol = varName)
+                            }
+                        }
+                    }
+                }
+                else {
+                    val eltSize = codeGen.program.memsizer.memorySize(target.type)
+                    val constIndex = target.index.asConstInteger()
+                    if(isConstZeroValue) {
+                        if(constIndex!=null) {
+                            val offsetReg = codeGen.registers.nextFree()
+                            val offset = eltSize*constIndex + if(msb) 1 else 0
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=offsetReg, immediate = offset)
+                                it += IRInstruction(Opcode.STOREZX, IRDataType.BYTE, reg1=offsetReg, labelSymbol = target.variable.name)
+                            }
+                        } else {
+                            val indexTr = exprGen.translateExpression(target.index)
+                            addToResult(result, indexTr, indexTr.resultReg, -1)
+                            result += IRCodeChunk(null, null).also {
+                                if(eltSize>1)
+                                    it += codeGen.multiplyByConst(IRDataType.BYTE, indexTr.resultReg, eltSize)
+                                if(msb)
+                                    it += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1=indexTr.resultReg)
+                                it += IRInstruction(Opcode.STOREZX, IRDataType.BYTE, reg1=indexTr.resultReg, labelSymbol = target.variable.name)
+                            }
+                        }
+                    } else {
+                        val valueTr = exprGen.translateExpression(call.args[1])
+                        addToResult(result, valueTr, valueTr.resultReg, -1)
+                        if(constIndex!=null) {
+                            val offsetReg = codeGen.registers.nextFree()
+                            val offset = eltSize*constIndex + if(msb) 1 else 0
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=offsetReg, immediate = offset)
+                                it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1=valueTr.resultReg, reg2=offsetReg, labelSymbol = target.variable.name)
+                            }
+                        } else {
+                            val indexTr = exprGen.translateExpression(target.index)
+                            addToResult(result, indexTr, indexTr.resultReg, -1)
+                            result += IRCodeChunk(null, null).also {
+                                if(eltSize>1)
+                                    it += codeGen.multiplyByConst(IRDataType.BYTE, indexTr.resultReg, eltSize)
+                                if(msb)
+                                    it += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1=indexTr.resultReg)
+                                it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1=valueTr.resultReg, reg2=indexTr.resultReg, labelSymbol = target.variable.name)
+                            }
+                        }
+                    }
+                }
+            }
+            else -> throw AssemblyError("weird target for setlsb/setmsb: $target")
+        }
+        return ExpressionCodeResult(result, IRDataType.WORD, -1, -1)
+    }
+
+
     private fun assignRegisterTo(target: PtExpression, register: Int): IRCodeChunks {
         val assignment = PtAssignment(target.position)
         val assignTarget = PtAssignTarget(target.position)

codeGenIntermediate/src/prog8/codegen/intermediate/ExpressionGen.kt

@@ -3,10 +3,7 @@ package prog8.codegen.intermediate
 import prog8.code.StRomSub
 import prog8.code.StSub
 import prog8.code.ast.*
-import prog8.code.core.AssemblyError
-import prog8.code.core.DataType
-import prog8.code.core.PassByValueDatatypes
-import prog8.code.core.SignedDatatypes
+import prog8.code.core.*
 import prog8.intermediate.*
 
 internal class ExpressionCodeResult(val chunks: IRCodeChunks, val dt: IRDataType, val resultReg: Int, val resultFpReg: Int) {
@@ -69,10 +66,28 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 val vmDt = irType(expr.type)
                 val symbol = expr.identifier.name
                 // note: LOAD <symbol>  gets you the address of the symbol, whereas LOADM <symbol> would get you the value stored at that location
-                val code = IRCodeChunk(null, null)
+                val result = mutableListOf<IRCodeChunkBase>()
                 val resultRegister = codeGen.registers.nextFree()
-                code += IRInstruction(Opcode.LOAD, vmDt, reg1=resultRegister, labelSymbol = symbol)
-                ExpressionCodeResult(code, vmDt, resultRegister, -1)
+                addInstr(result, IRInstruction(Opcode.LOAD, vmDt, reg1 = resultRegister, labelSymbol = symbol), null)
+                if(expr.isFromArrayElement) {
+                    val indexTr = translateExpression(expr.arrayIndexExpr!!)
+                    addToResult(result, indexTr, indexTr.resultReg, -1)
+                    if(expr.identifier.type in SplitWordArrayTypes) {
+                        result += IRCodeChunk(null, null).also {
+                            // multiply indexTr resultreg by the eltSize and add this to the resultRegister.
+                            it += IRInstruction(Opcode.ADDR, IRDataType.BYTE, reg1=resultRegister, reg2=indexTr.resultReg)
+                        }
+                    } else {
+                        val eltSize = codeGen.program.memsizer.memorySize(expr.identifier.type, 1)
+                        result += IRCodeChunk(null, null).also {
+                            // multiply indexTr resultreg by the eltSize and add this to the resultRegister.
+                            if(eltSize>1)
+                                it += IRInstruction(Opcode.MUL, IRDataType.BYTE, reg1=indexTr.resultReg, immediate = eltSize)
+                            it += IRInstruction(Opcode.ADDR, IRDataType.BYTE, reg1=resultRegister, reg2=indexTr.resultReg)
+                        }
+                    }
+                }
+                ExpressionCodeResult(result, vmDt, resultRegister, -1)
             }
             is PtMemoryByte -> {
                 val result = mutableListOf<IRCodeChunkBase>()
@@ -107,6 +122,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         val result = mutableListOf<IRCodeChunkBase>()
         when(check.iterable.type) {
             DataType.STR -> {
+                addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
                 val elementTr = translateExpression(check.element)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
                 val iterableTr = translateExpression(check.iterable)
@@ -115,6 +131,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 return ExpressionCodeResult(result, IRDataType.BYTE, elementTr.resultReg, -1)
             }
             DataType.ARRAY_UB, DataType.ARRAY_B -> {
+                addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 3), null)
                 val elementTr = translateExpression(check.element)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
                 val iterableTr = translateExpression(check.iterable)
@@ -126,6 +143,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 return ExpressionCodeResult(result, IRDataType.BYTE, elementTr.resultReg, -1)
             }
             DataType.ARRAY_UW, DataType.ARRAY_W -> {
+                addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 3), null)
                 val elementTr = translateExpression(check.element)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
                 val iterableTr = translateExpression(check.iterable)
@@ -170,18 +188,18 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 val memOffset = (arrayIx.index as PtNumber).number.toInt()
                 result += IRCodeChunk(null, null).also {
                     val tmpRegMsb = codeGen.registers.nextFree()
-                    it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1=resultRegister, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb+$memOffset")
                     it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1=tmpRegMsb, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_msb+$memOffset")
-                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=resultRegister, reg3=tmpRegMsb)
+                    it += IRInstruction(Opcode.LOADM, IRDataType.BYTE, reg1=resultRegister, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb+$memOffset")
+                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=tmpRegMsb, reg3=resultRegister)
                 }
             } else {
                 val tr = translateExpression(arrayIx.index)
                 addToResult(result, tr, tr.resultReg, -1)
                 result += IRCodeChunk(null, null).also {
                     val tmpRegMsb = codeGen.registers.nextFree()
-                    it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=resultRegister, reg2 = tr.resultReg, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb")
                     it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegMsb, reg2 = tr.resultReg, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_msb")
-                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=resultRegister, reg3=tmpRegMsb)
+                    it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=resultRegister, reg2 = tr.resultReg, immediate = arrayLength, labelSymbol= "${arrayVarSymbol}_lsb")
+                    it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=finalResultReg, reg2=tmpRegMsb, reg3=resultRegister)
                 }
             }
             return ExpressionCodeResult(result, vmDt, finalResultReg, -1)
@@ -366,6 +384,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         when (val callTarget = codeGen.symbolTable.flat.getValue(fcall.name)) {
             is StSub -> {
                 val result = mutableListOf<IRCodeChunkBase>()
+                addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = callTarget.parameters.size), null)
                 // assign the arguments
                 val argRegisters = mutableListOf<FunctionCallArgs.ArgumentSpec>()
                 for ((arg, parameter) in fcall.args.zip(callTarget.parameters)) {
@@ -386,8 +405,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                         FunctionCallArgs.RegSpec(returnIrType, codeGen.registers.nextFree(), null)
                 }
                 // create the call
-                val call = IRInstruction(Opcode.CALL, labelSymbol = fcall.name, fcallArgs = FunctionCallArgs(argRegisters, returnRegSpec))
-                addInstr(result, call, null)
+                addInstr(result, IRInstruction(Opcode.CALL, labelSymbol = fcall.name, fcallArgs = FunctionCallArgs(argRegisters, returnRegSpec)), null)
                 return if(fcall.void)
                     ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
                 else if(fcall.type==DataType.FLOAT)
@@ -397,6 +415,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             }
             is StRomSub -> {
                 val result = mutableListOf<IRCodeChunkBase>()
+                addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = callTarget.parameters.size), null)
                 // assign the arguments
                 val argRegisters = mutableListOf<FunctionCallArgs.ArgumentSpec>()
                 for ((arg, parameter) in fcall.args.zip(callTarget.parameters)) {
@@ -543,14 +562,16 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             val resultRegister = codeGen.registers.nextFree()
             val valueReg = codeGen.registers.nextFree()
             val label = codeGen.createLabelName()
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=resultRegister, immediate = 1), null)
-            addInstr(result, IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=valueReg, fpReg1 = leftTr.resultFpReg, fpReg2 = rightTr.resultFpReg), null)
-            if (notEquals) {
-                addInstr(result, IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=valueReg, immediate = 0, labelSymbol = label), null)
-            } else {
-                addInstr(result, IRInstruction(Opcode.BEQ, IRDataType.BYTE, reg1=valueReg, immediate = 0, labelSymbol = label), null)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=resultRegister, immediate = 1)
+                it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=valueReg, fpReg1 = leftTr.resultFpReg, fpReg2 = rightTr.resultFpReg)
+                it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=valueReg, immediate = 0)
+                it += if (notEquals)
+                    IRInstruction(Opcode.BSTNE, labelSymbol = label)
+                else
+                    IRInstruction(Opcode.BSTEQ, labelSymbol = label)
+                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=resultRegister, immediate = 0)
             }
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=resultRegister, immediate = 0), null)
             result += IRCodeChunk(label, null)
             return ExpressionCodeResult(result, IRDataType.BYTE, resultRegister, -1)
         } else {

codeGenIntermediate/src/prog8/codegen/intermediate/IRCodeGen.kt

@@ -415,13 +415,19 @@ class IRCodeGen(
                 if(choice.statements.children.isEmpty()) {
                     // no statements for this choice value, jump to the end immediately
                     choice.values.children.map { it as PtNumber }.sortedBy { it.number }.forEach { value ->
-                        addInstr(result, IRInstruction(Opcode.BEQ, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt(), labelSymbol = endLabel), null)
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.CMPI, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt())
+                            it += IRInstruction(Opcode.BSTEQ, labelSymbol = endLabel)
+                        }
                     }
                 } else {
                     val choiceLabel = createLabelName()
                     choices.add(choiceLabel to choice)
                     choice.values.children.map { it as PtNumber }.sortedBy { it.number }.forEach { value ->
-                        addInstr(result, IRInstruction(Opcode.BEQ, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt(), labelSymbol = choiceLabel), null)
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.CMPI, valueDt, reg1=valueTr.resultReg, immediate = value.number.toInt())
+                            it += IRInstruction(Opcode.BSTEQ, labelSymbol = choiceLabel)
+                        }
                     }
                 }
             }
@@ -464,7 +470,8 @@ class IRCodeGen(
                         addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = indexReg, immediate = 0), null)
                         result += IRCodeChunk(loopLabel, null).also {
                             it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1 = tmpReg, reg2 = indexReg, labelSymbol = iterable.name)
-                            it += IRInstruction(Opcode.BEQ, IRDataType.BYTE, reg1 = tmpReg, immediate = 0, labelSymbol = endLabel)
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = tmpReg, immediate = 0)
+                            it += IRInstruction(Opcode.BSTEQ, labelSymbol = endLabel)
                             it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = tmpReg, labelSymbol = loopvarSymbol)
                         }
                         result += translateNode(forLoop.statements)
@@ -484,15 +491,16 @@ class IRCodeGen(
                             val tmpRegLsb = registers.nextFree()
                             val tmpRegMsb = registers.nextFree()
                             val concatReg = registers.nextFree()
-                            it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegLsb, reg2=indexReg, immediate = iterableLength, labelSymbol=iterable.name+"_lsb")
                             it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegMsb, reg2=indexReg, immediate = iterableLength, labelSymbol=iterable.name+"_msb")
-                            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=concatReg, reg2=tmpRegLsb, reg3=tmpRegMsb)
+                            it += IRInstruction(Opcode.LOADX, IRDataType.BYTE, reg1=tmpRegLsb, reg2=indexReg, immediate = iterableLength, labelSymbol=iterable.name+"_lsb")
+                            it += IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=concatReg, reg2=tmpRegMsb, reg3=tmpRegLsb)
                             it += IRInstruction(Opcode.STOREM, irType(elementDt), reg1=concatReg, labelSymbol = loopvarSymbol)
                         }
                         result += translateNode(forLoop.statements)
                         result += IRCodeChunk(null, null).also {
                             it += IRInstruction(Opcode.INC, IRDataType.BYTE, reg1=indexReg)
-                            it += IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(iterableLength==256) 0 else iterableLength, labelSymbol = loopLabel)
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=indexReg, immediate = if(iterableLength==256) 0 else iterableLength)
+                            it += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
                         }
                     }
                     else -> {
@@ -507,7 +515,10 @@ class IRCodeGen(
                         }
                         result += translateNode(forLoop.statements)
                         result += addConstReg(IRDataType.BYTE, indexReg, elementSize)
-                        addInstr(result, IRInstruction(Opcode.BNE, IRDataType.BYTE, reg1=indexReg, immediate = if(lengthBytes==256) 0 else lengthBytes, labelSymbol = loopLabel), null)
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=indexReg, immediate = if(lengthBytes==256) 0 else lengthBytes)
+                            it += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+                        }
                     }
                 }
             }
@@ -555,8 +566,11 @@ class IRCodeGen(
         result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
         if(step==1 || step==-1) {
             // if endvalue == loopvar, stop loop, else iterate
-            addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = fromTr.resultReg, labelSymbol = loopvarSymbol), null)
-            addInstr(result, IRInstruction(Opcode.BEQR, loopvarDtIr, reg1=toTr.resultReg, reg2=fromTr.resultReg, labelSymbol = labelAfterFor), null)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = fromTr.resultReg, labelSymbol = loopvarSymbol)
+                it += IRInstruction(Opcode.CMP, loopvarDtIr, reg1=toTr.resultReg, reg2=fromTr.resultReg)
+                it += IRInstruction(Opcode.BSTEQ, labelSymbol = labelAfterFor)
+            }
             result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
             addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = loopLabel), null)
         } else {
@@ -601,8 +615,8 @@ class IRCodeGen(
         result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
         val chunk2 = addConstMem(loopvarDtIr, null, loopvarSymbol, iterable.step)
         chunk2 += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = indexReg, labelSymbol = loopvarSymbol)
-        chunk2 += IRInstruction(Opcode.XOR, loopvarDtIr, reg1 = 999, immediate = 111)
-        chunk2 += IRInstruction(Opcode.BNE, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped, labelSymbol = loopLabel)
+        chunk2 += IRInstruction(Opcode.CMPI, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped)
+        chunk2 += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
         result += chunk2
         return result
     }
@@ -961,21 +975,33 @@ class IRCodeGen(
                         fpReg1 = leftTr.resultFpReg,
                         fpReg2 = rightTr.resultFpReg
                     )
-                    val gotoOpcode = when (condition.operator) {
-                        "==" -> Opcode.BEQ
-                        "!=" -> Opcode.BNE
-                        "<" -> Opcode.BLTS
-                        ">" -> Opcode.BGTS
-                        "<=" -> Opcode.BLES
-                        ">=" -> Opcode.BGES
-                        else -> throw AssemblyError("weird operator")
+                    when(condition.operator) {
+                        // TODO: the converted list of operators
+                        "==" -> {
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = compResultReg, immediate = 0)
+                            it += branchInstr(goto, Opcode.BSTEQ)
+                        }
+                        "!=" -> {
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = compResultReg, immediate = 0)
+                            it += branchInstr(goto, Opcode.BSTNE)
+                        }
+                        else -> {
+                            // TODO: the old list of operators, still to be converted
+                            val gotoOpcode = when (condition.operator) {
+                                "<" -> Opcode.BLTS
+                                ">" -> Opcode.BGTS
+                                "<=" -> Opcode.BLES
+                                ">=" -> Opcode.BGES
+                                else -> throw AssemblyError("weird operator")
+                            }
+                            it += if (goto.address != null)
+                                IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, address = goto.address?.toInt())
+                            else if (goto.generatedLabel != null)
+                                IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = goto.generatedLabel)
+                            else
+                                IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = goto.identifier!!.name)
+                        }
                     }
-                    it += if (goto.address != null)
-                        IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, address = goto.address?.toInt())
-                    else if (goto.generatedLabel != null)
-                        IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = goto.generatedLabel)
-                    else
-                        IRInstruction(gotoOpcode, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = goto.identifier!!.name)
                 }
                 return result
             } else {
@@ -990,6 +1016,14 @@ class IRCodeGen(
         }
     }
 
+    // TODO use this everywhere
+    private fun branchInstr(goto: PtJump, branchOpcode: Opcode) = if (goto.address != null)
+        IRInstruction(branchOpcode, address = goto.address?.toInt())
+    else if (goto.generatedLabel != null)
+        IRInstruction(branchOpcode, labelSymbol = goto.generatedLabel)
+    else
+        IRInstruction(branchOpcode, labelSymbol = goto.identifier!!.name)
+
     private fun ifZeroIntThenJump(
         result: MutableList<IRCodeChunkBase>,
         ifElse: PtIfElse,
@@ -1000,21 +1034,33 @@ class IRCodeGen(
         val condition = ifElse.condition as PtBinaryExpression
         val leftTr = expressionEval.translateExpression(condition.left)
         addToResult(result, leftTr, leftTr.resultReg, -1)
-        val opcode = when (condition.operator) {
-            "==" -> Opcode.BEQ
-            "!=" -> Opcode.BNE
-            "<" -> if (signed) Opcode.BLTS else Opcode.BLT
-            ">" -> if (signed) Opcode.BGTS else Opcode.BGT
-            "<=" -> if (signed) Opcode.BLES else Opcode.BLE
-            ">=" -> if (signed) Opcode.BGES else Opcode.BGE
-            else -> throw AssemblyError("invalid comparison operator")
+        when(condition.operator) {
+            // TODO: converted list of operators
+            "==" -> {
+                addInstr(result, IRInstruction(Opcode.CMPI, irDtLeft, reg1 = leftTr.resultReg, immediate = 0), null)
+                addInstr(result, branchInstr(goto, Opcode.BSTEQ), null)
+            }
+            "!=" -> {
+                addInstr(result, IRInstruction(Opcode.CMPI, irDtLeft, reg1 = leftTr.resultReg, immediate = 0), null)
+                addInstr(result, branchInstr(goto, Opcode.BSTNE), null)
+            }
+            else -> {
+                // TODO: to-be converted operators
+                val opcode = when (condition.operator) {
+                    "<" -> if (signed) Opcode.BLTS else Opcode.BLT
+                    ">" -> if (signed) Opcode.BGTS else Opcode.BGT
+                    "<=" -> if (signed) Opcode.BLES else Opcode.BLE
+                    ">=" -> if (signed) Opcode.BGES else Opcode.BGE
+                    else -> throw AssemblyError("invalid comparison operator")
+                }
+                if (goto.address != null)
+                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, address = goto.address?.toInt()), null)
+                else if (goto.generatedLabel != null)
+                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, labelSymbol = goto.generatedLabel), null)
+                else
+                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, labelSymbol = goto.identifier!!.name), null)
+            }
         }
-        if (goto.address != null)
-            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, address = goto.address?.toInt()), null)
-        else if (goto.generatedLabel != null)
-            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, labelSymbol = goto.generatedLabel), null)
-        else
-            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = leftTr.resultReg, immediate = 0, labelSymbol = goto.identifier!!.name), null)
     }
 
     private fun ifNonZeroIntThenJump(
@@ -1028,47 +1074,60 @@ class IRCodeGen(
         if(condition==null) {
             val tr = expressionEval.translateExpression(ifElse.condition)
             result += tr.chunks
-            if (goto.address != null)
-                addInstr(result, IRInstruction(Opcode.BNE, irDtLeft, reg1 = tr.resultReg, immediate = 0, address = goto.address?.toInt()), null)
-            else if (goto.generatedLabel != null)
-                addInstr(result, IRInstruction(Opcode.BNE, irDtLeft, reg1 = tr.resultReg, immediate = 0, labelSymbol = goto.generatedLabel), null)
-            else
-                addInstr(result, IRInstruction(Opcode.BNE, irDtLeft, reg1 = tr.resultReg, immediate = 0, labelSymbol = goto.identifier!!.name), null)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.CMPI, irDtLeft, reg1 = tr.resultReg, immediate = 0)
+                it += branchInstr(goto, Opcode.BSTNE)
+            }
         } else {
             val leftTr = expressionEval.translateExpression(condition.left)
             addToResult(result, leftTr, leftTr.resultReg, -1)
-            val opcode: Opcode
             val number = (condition.right as? PtNumber)?.number?.toInt()
             if(number!=null) {
                 val firstReg = leftTr.resultReg
-                opcode = when (condition.operator) {
-                    "==" -> Opcode.BEQ
-                    "!=" -> Opcode.BNE
-                    "<" -> if(signed) Opcode.BLTS else Opcode.BLT
-                    ">" -> if(signed) Opcode.BGTS else Opcode.BGT
-                    "<=" -> if(signed) Opcode.BLES else Opcode.BLE
-                    ">=" -> if(signed) Opcode.BGES else Opcode.BGE
-                    else -> throw AssemblyError("invalid comparison operator")
+                when(condition.operator) {
+                    // TODO: the converted operators
+                    "==" -> {
+                        addInstr(result, IRInstruction(Opcode.CMPI, irDtLeft, reg1 = firstReg, immediate = number), null)
+                        addInstr(result, branchInstr(goto, Opcode.BSTEQ), null)
+                    }
+                    "!=" -> {
+                        addInstr(result, IRInstruction(Opcode.CMPI, irDtLeft, reg1 = firstReg, immediate = number), null)
+                        addInstr(result, branchInstr(goto, Opcode.BSTNE), null)
+                    }
+                    else -> {
+                        // TODO: to-be converted operators
+                        val opcode = when (condition.operator) {
+                            "<" -> if(signed) Opcode.BLTS else Opcode.BLT
+                            ">" -> if(signed) Opcode.BGTS else Opcode.BGT
+                            "<=" -> if(signed) Opcode.BLES else Opcode.BLE
+                            ">=" -> if(signed) Opcode.BGES else Opcode.BGE
+                            else -> throw AssemblyError("invalid comparison operator")
+                        }
+                        if (goto.address != null)
+                            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, address = goto.address?.toInt()), null)
+                        else if (goto.generatedLabel != null)
+                            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, labelSymbol = goto.generatedLabel), null)
+                        else
+                            addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, labelSymbol = goto.identifier!!.name), null)
+                    }
                 }
-                if (goto.address != null)
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, address = goto.address?.toInt()), null)
-                else if (goto.generatedLabel != null)
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, labelSymbol = goto.generatedLabel), null)
-                else
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, immediate = number, labelSymbol = goto.identifier!!.name), null)
             } else {
                 val rightTr = expressionEval.translateExpression(condition.right)
                 addToResult(result, rightTr, rightTr.resultReg, -1)
                 val firstReg: Int
                 val secondReg: Int
+                val opcode: Opcode
+                var useCmp = false
                 when (condition.operator) {
                     "==" -> {
-                        opcode = Opcode.BEQR
+                        useCmp = true
+                        opcode = Opcode.BSTEQ
                         firstReg = leftTr.resultReg
                         secondReg = rightTr.resultReg
                     }
                     "!=" -> {
-                        opcode = Opcode.BNER
+                        useCmp = true
+                        opcode = Opcode.BSTNE
                         firstReg = leftTr.resultReg
                         secondReg = rightTr.resultReg
                     }
@@ -1096,12 +1155,20 @@ class IRCodeGen(
                     }
                     else -> throw AssemblyError("invalid comparison operator")
                 }
-                if (goto.address != null)
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, address = goto.address?.toInt()), null)
-                else if (goto.generatedLabel != null)
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, labelSymbol = goto.generatedLabel), null)
-                else
-                    addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, labelSymbol = goto.identifier!!.name), null)
+
+                if(useCmp) {
+                    result += IRCodeChunk(null, null).also {
+                        it += IRInstruction(Opcode.CMP, irDtLeft, reg1 = firstReg, reg2 = secondReg)
+                        it += branchInstr(goto, opcode)
+                    }
+                } else {
+                    if (goto.address != null)
+                        addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, address = goto.address?.toInt()), null)
+                    else if (goto.generatedLabel != null)
+                        addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, labelSymbol = goto.generatedLabel), null)
+                    else
+                        addInstr(result, IRInstruction(opcode, irDtLeft, reg1 = firstReg, reg2 = secondReg, labelSymbol = goto.identifier!!.name), null)
+                }
             }
         }
     }
@@ -1109,6 +1176,7 @@ class IRCodeGen(
     private fun translateIfElseZeroComparison(ifElse: PtIfElse, irDtLeft: IRDataType, signed: Boolean): IRCodeChunks {
         val result = mutableListOf<IRCodeChunkBase>()
         val elseBranch: Opcode
+        var useCmpi = false     // for the branch opcodes that have been converted to CMPI + BSTxx form already
         val compResultReg: Int
         val branchDt: IRDataType
         val condition = ifElse.condition as PtBinaryExpression
@@ -1122,13 +1190,19 @@ class IRCodeGen(
                 it += IRInstruction(Opcode.LOAD, IRDataType.FLOAT, fpReg1 = rightFpReg, immediateFp = 0f)
                 it += IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1=compResultReg, fpReg1 = leftTr.resultFpReg, fpReg2 = rightFpReg)
             }
-            elseBranch = when (condition.operator) {
-                "==" -> Opcode.BNE
-                "!=" -> Opcode.BEQ
-                "<" -> Opcode.BGES
-                ">" -> Opcode.BLES
-                "<=" -> Opcode.BGTS
-                ">=" -> Opcode.BLTS
+            when (condition.operator) {
+                "==" -> {
+                    elseBranch = Opcode.BSTNE
+                    useCmpi = true
+                }
+                "!=" -> {
+                    elseBranch = Opcode.BSTEQ
+                    useCmpi = true
+                }
+                "<" -> elseBranch = Opcode.BGES
+                ">" -> elseBranch = Opcode.BLES
+                "<=" -> elseBranch = Opcode.BGTS
+                ">=" -> elseBranch = Opcode.BLTS
                 else -> throw AssemblyError("weird operator")
             }
         } else {
@@ -1137,13 +1211,19 @@ class IRCodeGen(
             val tr = expressionEval.translateExpression(condition.left)
             compResultReg = tr.resultReg
             addToResult(result, tr, tr.resultReg, -1)
-            elseBranch = when (condition.operator) {
-                "==" -> Opcode.BNE
-                "!=" -> Opcode.BEQ
-                "<" -> if (signed) Opcode.BGES else Opcode.BGE
-                ">" -> if (signed) Opcode.BLES else Opcode.BLE
-                "<=" -> if (signed) Opcode.BGTS else Opcode.BGT
-                ">=" -> if (signed) Opcode.BLTS else Opcode.BLT
+            when (condition.operator) {
+                "==" -> {
+                    elseBranch = Opcode.BSTNE
+                    useCmpi = true
+                }
+                "!=" -> {
+                    elseBranch = Opcode.BSTEQ
+                    useCmpi = true
+                }
+                "<" -> elseBranch = if (signed) Opcode.BGES else Opcode.BGE
+                ">" -> elseBranch = if (signed) Opcode.BLES else Opcode.BLE
+                "<=" -> elseBranch = if (signed) Opcode.BGTS else Opcode.BGT
+                ">=" -> elseBranch = if (signed) Opcode.BLTS else Opcode.BLT
                 else -> throw AssemblyError("weird operator")
             }
         }
@@ -1151,14 +1231,26 @@ class IRCodeGen(
         if(ifElse.elseScope.children.isEmpty()) {
             // just if
             val afterIfLabel = createLabelName()
-            addInstr(result, IRInstruction(elseBranch, branchDt, reg1=compResultReg, immediate = 0, labelSymbol = afterIfLabel), null)
+            if(useCmpi) {
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.CMPI, branchDt, reg1=compResultReg, immediate = 0)
+                    it += IRInstruction(elseBranch, labelSymbol = afterIfLabel)
+                }
+            } else
+                addInstr(result, IRInstruction(elseBranch, branchDt, reg1=compResultReg, immediate = 0, labelSymbol = afterIfLabel), null)
             result += translateNode(ifElse.ifScope)
             result += IRCodeChunk(afterIfLabel, null)
         } else {
             // if and else
             val elseLabel = createLabelName()
             val afterIfLabel = createLabelName()
-            addInstr(result, IRInstruction(elseBranch, branchDt, reg1=compResultReg, immediate = 0, labelSymbol = elseLabel), null)
+            if(useCmpi) {
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.CMPI, branchDt, reg1=compResultReg, immediate = 0)
+                    it += IRInstruction(elseBranch, labelSymbol = elseLabel)
+                }
+            } else
+                addInstr(result, IRInstruction(elseBranch, branchDt, reg1=compResultReg, immediate = 0, labelSymbol = elseLabel), null)
             result += translateNode(ifElse.ifScope)
             addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
             result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
@@ -1169,31 +1261,39 @@ class IRCodeGen(
 
     private fun translateIfElseNonZeroComparison(ifElse: PtIfElse, irDtLeft: IRDataType, signed: Boolean): IRCodeChunks {
         val result = mutableListOf<IRCodeChunkBase>()
-        val elseBranchOpcode: Opcode
         val elseBranchFirstReg: Int
         val elseBranchSecondReg: Int
         val branchDt: IRDataType
         val condition = ifElse.condition as? PtBinaryExpression
         if(condition==null) {
-            if(irDtLeft==IRDataType.FLOAT)
-                throw AssemblyError("condition value should not be float")
-            val tr = expressionEval.translateExpression(ifElse.condition)
-            result += tr.chunks
-            if(ifElse.elseScope.children.isNotEmpty()) {
-                val elseLabel = createLabelName()
-                val afterIfLabel = createLabelName()
-                addInstr(result, IRInstruction(Opcode.BEQ, irDtLeft, reg1=tr.resultReg, immediate = 0, labelSymbol = elseLabel), null)
-                result += translateNode(ifElse.ifScope)
-                addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
-                result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
-                result += IRCodeChunk(afterIfLabel, null)
-            } else {
-                val afterIfLabel = createLabelName()
-                addInstr(result, IRInstruction(Opcode.BEQ, irDtLeft, reg1=tr.resultReg, immediate = 0, labelSymbol = afterIfLabel), null)
-                result += translateNode(ifElse.ifScope)
-                result += IRCodeChunk(afterIfLabel, null)
-            }
-            return result
+            throw AssemblyError("if-else condition is not a binaryexpression, should have been converted?")
+//            if(irDtLeft==IRDataType.FLOAT)
+//                throw AssemblyError("condition value should not be float")
+//            val tr = expressionEval.translateExpression(ifElse.condition)
+//            result += tr.chunks
+//            if(ifElse.elseScope.children.isNotEmpty()) {
+//                val elseLabel = createLabelName()
+//                result += IRCodeChunk(null, null).also {
+//                    it += IRInstruction(Opcode.CMPI, irDtLeft, reg1=tr.resultReg, immediate = 0)
+//                    it += IRInstruction(Opcode.BSTEQ, labelSymbol = elseLabel)
+//                    TODO("test this")
+//                }
+//                result += translateNode(ifElse.ifScope)
+//                val afterIfLabel = createLabelName()
+//                addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
+//                result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
+//                result += IRCodeChunk(afterIfLabel, null)
+//            } else {
+//                val afterIfLabel = createLabelName()
+//                result += IRCodeChunk(null, null).also {
+//                    it += IRInstruction(Opcode.CMPI, irDtLeft, reg1=tr.resultReg, immediate = 0)
+//                    it += IRInstruction(Opcode.BSTEQ, labelSymbol = afterIfLabel)
+//                    TODO("test this")
+//                }
+//                result += translateNode(ifElse.ifScope)
+//                result += IRCodeChunk(afterIfLabel, null)
+//            }
+//            return result
         } else {
             if (irDtLeft == IRDataType.FLOAT) {
                 val leftTr = expressionEval.translateExpression(condition.left)
@@ -1201,35 +1301,35 @@ class IRCodeGen(
                 val rightTr = expressionEval.translateExpression(condition.right)
                 addToResult(result, rightTr, -1, rightTr.resultFpReg)
                 val compResultReg = registers.nextFree()
-                addInstr(
-                    result,
-                    IRInstruction(
-                        Opcode.FCOMP,
-                        IRDataType.FLOAT,
-                        reg1 = compResultReg,
-                        fpReg1 = leftTr.resultFpReg,
-                        fpReg2 = rightTr.resultFpReg
-                    ),
-                    null
-                )
-                val elseBranch = when (condition.operator) {
-                    "==" -> Opcode.BNE
-                    "!=" -> Opcode.BEQ
-                    "<" -> Opcode.BGES
-                    ">" -> Opcode.BLES
-                    "<=" -> Opcode.BGTS
-                    ">=" -> Opcode.BLTS
+                addInstr(result, IRInstruction(Opcode.FCOMP, IRDataType.FLOAT, reg1 = compResultReg, fpReg1 = leftTr.resultFpReg, fpReg2 = rightTr.resultFpReg), null)
+                val elseBranch: Opcode
+                var useCmpi = false     // for the branch opcodes that have been converted to CMPI + BSTxx form already
+                when (condition.operator) {
+                    "==" -> {
+                        elseBranch = Opcode.BSTNE
+                        useCmpi = true
+                    }
+                    "!=" -> {
+                        elseBranch = Opcode.BSTEQ
+                        useCmpi = true
+                    }
+                    "<" -> elseBranch = Opcode.BGES
+                    ">" -> elseBranch = Opcode.BLES
+                    "<=" -> elseBranch = Opcode.BGTS
+                    ">=" -> elseBranch = Opcode.BLTS
                     else -> throw AssemblyError("weird operator")
                 }
                 if (ifElse.elseScope.children.isNotEmpty()) {
                     // if and else parts
                     val elseLabel = createLabelName()
                     val afterIfLabel = createLabelName()
-                    addInstr(
-                        result,
-                        IRInstruction(elseBranch, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = elseLabel),
-                        null
-                    )
+                    if(useCmpi) {
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = compResultReg, immediate = 0)
+                            it += IRInstruction(elseBranch, labelSymbol = elseLabel)
+                        }
+                    } else
+                        addInstr(result, IRInstruction(elseBranch, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = elseLabel), null)
                     result += translateNode(ifElse.ifScope)
                     addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
                     result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
@@ -1237,11 +1337,13 @@ class IRCodeGen(
                 } else {
                     // only if part
                     val afterIfLabel = createLabelName()
-                    addInstr(
-                        result,
-                        IRInstruction(elseBranch, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = afterIfLabel),
-                        null
-                    )
+                    if(useCmpi) {
+                        result += IRCodeChunk(null, null).also {
+                            it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = compResultReg, immediate = 0)
+                            it += IRInstruction(elseBranch, labelSymbol = afterIfLabel)
+                        }
+                    } else
+                        addInstr(result, IRInstruction(elseBranch, IRDataType.BYTE, reg1 = compResultReg, immediate = 0, labelSymbol = afterIfLabel), null)
                     result += translateNode(ifElse.ifScope)
                     result += IRCodeChunk(afterIfLabel, null)
                 }
@@ -1252,26 +1354,34 @@ class IRCodeGen(
                 addToResult(result, leftTr, leftTr.resultReg, -1)
                 val number = (condition.right as? PtNumber)?.number?.toInt()
                 if (number!=null) {
-                    elseBranchOpcode = when (condition.operator) {
-                        "==" -> Opcode.BNE
-                        "!=" -> Opcode.BEQ
-                        "<" -> if(signed) Opcode.BGES else Opcode.BGE
-                        ">" -> if(signed) Opcode.BLES else Opcode.BLE
-                        "<=" -> if(signed) Opcode.BGTS else Opcode.BGT
-                        ">=" -> if(signed) Opcode.BLTS else Opcode.BLT
+                    val elseBranch: Opcode
+                    var useCmpi = false     // for the branch opcodes that have been converted to CMPI + BSTxx form already
+                    when (condition.operator) {
+                        "==" -> {
+                            elseBranch = Opcode.BSTNE
+                            useCmpi = true
+                        }
+                        "!=" -> {
+                            elseBranch = Opcode.BSTEQ
+                            useCmpi = true
+                        }
+                        "<" -> elseBranch = if(signed) Opcode.BGES else Opcode.BGE
+                        ">" -> elseBranch = if(signed) Opcode.BLES else Opcode.BLE
+                        "<=" -> elseBranch = if(signed) Opcode.BGTS else Opcode.BGT
+                        ">=" -> elseBranch = if(signed) Opcode.BLTS else Opcode.BLT
                         else -> throw AssemblyError("invalid comparison operator")
                     }
                     if (ifElse.elseScope.children.isNotEmpty()) {
                         // if and else parts
                         val elseLabel = createLabelName()
                         val afterIfLabel = createLabelName()
-                        addInstr(
-                            result, IRInstruction(
-                                elseBranchOpcode, branchDt,
-                                reg1 = leftTr.resultReg, immediate = number,
-                                labelSymbol = elseLabel
-                            ), null
-                        )
+                        if(useCmpi) {
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.CMPI, branchDt, reg1 = leftTr.resultReg, immediate = number)
+                                it += IRInstruction(elseBranch, labelSymbol = elseLabel)
+                            }
+                        } else
+                            addInstr(result, IRInstruction(elseBranch, branchDt, reg1 = leftTr.resultReg, immediate = number, labelSymbol = elseLabel), null)
                         result += translateNode(ifElse.ifScope)
                         addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
                         result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
@@ -1279,52 +1389,56 @@ class IRCodeGen(
                     } else {
                         // only if part
                         val afterIfLabel = createLabelName()
-                        addInstr(
-                            result, IRInstruction(
-                                elseBranchOpcode, branchDt,
-                                reg1 = leftTr.resultReg, immediate = number,
-                                labelSymbol = afterIfLabel
-                            ), null
-                        )
+                        if(useCmpi) {
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.CMPI, branchDt, reg1 = leftTr.resultReg, immediate = number)
+                                it += IRInstruction(elseBranch, labelSymbol = afterIfLabel)
+                            }
+                        } else
+                            addInstr(result, IRInstruction(elseBranch, branchDt, reg1 = leftTr.resultReg, immediate = number, labelSymbol = afterIfLabel), null)
                         result += translateNode(ifElse.ifScope)
                         result += IRCodeChunk(afterIfLabel, null)
                     }
                 } else {
                     val rightTr = expressionEval.translateExpression(condition.right)
+                    val elseBranch: Opcode
+                    var useCmp = false
                     addToResult(result, rightTr, rightTr.resultReg, -1)
                     when (condition.operator) {
                         "==" -> {
-                            elseBranchOpcode = Opcode.BNER
+                            useCmp = true
+                            elseBranch = Opcode.BSTNE
                             elseBranchFirstReg = leftTr.resultReg
                             elseBranchSecondReg = rightTr.resultReg
                         }
                         "!=" -> {
-                            elseBranchOpcode = Opcode.BEQR
+                            useCmp = true
+                            elseBranch = Opcode.BSTEQ
                             elseBranchFirstReg = leftTr.resultReg
                             elseBranchSecondReg = rightTr.resultReg
                         }
                         "<" -> {
                             // else part when left >= right
-                            elseBranchOpcode = if (signed) Opcode.BGESR else Opcode.BGER
+                            elseBranch = if (signed) Opcode.BGESR else Opcode.BGER
                             elseBranchFirstReg = leftTr.resultReg
                             elseBranchSecondReg = rightTr.resultReg
                         }
                         ">" -> {
                             // else part when left <= right --> right >= left
-                            elseBranchOpcode = if (signed) Opcode.BGESR else Opcode.BGER
+                            elseBranch = if (signed) Opcode.BGESR else Opcode.BGER
                             elseBranchFirstReg = rightTr.resultReg
                             elseBranchSecondReg = leftTr.resultReg
                         }
                         "<=" -> {
                             // else part when left > right
-                            elseBranchOpcode = if (signed) Opcode.BGTSR else Opcode.BGTR
+                            elseBranch = if (signed) Opcode.BGTSR else Opcode.BGTR
                             elseBranchFirstReg = leftTr.resultReg
                             elseBranchSecondReg = rightTr.resultReg
                         }
 
                         ">=" -> {
                             // else part when left < right --> right > left
-                            elseBranchOpcode = if (signed) Opcode.BGTSR else Opcode.BGTR
+                            elseBranch = if (signed) Opcode.BGTSR else Opcode.BGTR
                             elseBranchFirstReg = rightTr.resultReg
                             elseBranchSecondReg = leftTr.resultReg
                         }
@@ -1334,13 +1448,14 @@ class IRCodeGen(
                         // if and else parts
                         val elseLabel = createLabelName()
                         val afterIfLabel = createLabelName()
-                        addInstr(
-                            result, IRInstruction(
-                                elseBranchOpcode, branchDt,
-                                reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg,
-                                labelSymbol = elseLabel
-                            ), null
-                        )
+                        if(useCmp) {
+                            result += IRCodeChunk(null,null).also {
+                                it += IRInstruction(Opcode.CMP, branchDt, reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg)
+                                it += IRInstruction(elseBranch, labelSymbol = elseLabel)
+                            }
+                        } else {
+                            addInstr(result, IRInstruction(elseBranch, branchDt, reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg, labelSymbol = elseLabel), null)
+                        }
                         result += translateNode(ifElse.ifScope)
                         addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = afterIfLabel), null)
                         result += labelFirstChunk(translateNode(ifElse.elseScope), elseLabel)
@@ -1348,13 +1463,14 @@ class IRCodeGen(
                     } else {
                         // only if part
                         val afterIfLabel = createLabelName()
-                        addInstr(
-                            result, IRInstruction(
-                                elseBranchOpcode, branchDt,
-                                reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg,
-                                labelSymbol = afterIfLabel
-                            ), null
-                        )
+                        if(useCmp) {
+                            result += IRCodeChunk(null,null).also {
+                                it += IRInstruction(Opcode.CMP, branchDt, reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg)
+                                it += IRInstruction(elseBranch, labelSymbol = afterIfLabel)
+                            }
+                        } else {
+                            addInstr(result, IRInstruction(elseBranch, branchDt, reg1 = elseBranchFirstReg, reg2 = elseBranchSecondReg, labelSymbol = afterIfLabel), null)
+                        }
                         result += translateNode(ifElse.ifScope)
                         result += IRCodeChunk(afterIfLabel, null)
                     }
@@ -1525,12 +1641,15 @@ class IRCodeGen(
         addToResult(result, countTr, countTr.resultReg, -1)
         if(constIntValue(repeat.count)==null) {
             // check if the counter is already zero
-            addInstr(result, IRInstruction(Opcode.BEQ, irDt, reg1=countTr.resultReg, immediate = 0, labelSymbol = skipRepeatLabel), null)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.CMPI, irDt, reg1=countTr.resultReg, immediate = 0)
+                it += IRInstruction(Opcode.BSTEQ, labelSymbol = skipRepeatLabel)
+            }
         }
         result += labelFirstChunk(translateNode(repeat.statements), repeatLabel)
         result += IRCodeChunk(null, null).also {
-            it += IRInstruction(Opcode.DEC, irDt, reg1 = countTr.resultReg)
-            it += IRInstruction(Opcode.BNE, irDt, reg1 = countTr.resultReg, immediate = 0, labelSymbol = repeatLabel)
+            it += IRInstruction(Opcode.DEC, irDt, reg1 = countTr.resultReg)  // sets status bits
+            it += IRInstruction(Opcode.BSTNE, labelSymbol = repeatLabel)
         }
         result += IRCodeChunk(skipRepeatLabel, null)
         return result

codeGenIntermediate/src/prog8/codegen/intermediate/IRPeepholeOptimizer.kt

@@ -45,6 +45,7 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
                         val indexedInstructions = chunk1.instructions.withIndex()
                             .map { IndexedValue(it.index, it.value) }
                         val changed = removeNops(chunk1, indexedInstructions)
+                                || replaceConcatZeroMsbWithExt(chunk1, indexedInstructions)
                                 || removeDoubleLoadsAndStores(chunk1, indexedInstructions)       // TODO not yet implemented
                                 || removeUselessArithmetic(chunk1, indexedInstructions)
                                 || removeNeedlessCompares(chunk1, indexedInstructions)
@@ -63,6 +64,22 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
         irprog.linkChunks()  // re-link
     }
 
+    private fun replaceConcatZeroMsbWithExt(chunk: IRCodeChunk, indexedInstructions: List<IndexedValue<IRInstruction>>): Boolean {
+        var changed = false
+        indexedInstructions.reversed().forEach { (idx, ins) ->
+            if (ins.opcode == Opcode.CONCAT) {
+                // if the previous instruction loads a zero in the msb, this can be turned into EXT.B instead
+                val prev = indexedInstructions[idx-1].value
+                if(prev.opcode==Opcode.LOAD && prev.immediate==0 && prev.reg1==ins.reg2) {
+                    chunk.instructions[idx] = IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1 = ins.reg1, reg2 = ins.reg3)
+                    chunk.instructions.removeAt(idx-1)
+                    changed = true
+                }
+            }
+        }
+        return changed
+    }
+
     private fun removeEmptyChunks(sub: IRSubroutine) {
         if(sub.chunks.isEmpty())
             return

codeGenIntermediate/src/prog8/codegen/intermediate/IRUnusedCodeRemover.kt

@@ -25,18 +25,29 @@ class IRUnusedCodeRemover(
     }
 
     private fun pruneSymboltable(blockLabel: String) {
-        // we could clean up the SymbolTable as well, but ONLY if these symbols aren't referenced somewhere still in an instruction
+        // we could clean up the SymbolTable as well, but ONLY if these symbols aren't referenced somewhere still in an instruction or variable initializer value
         val prefix = "$blockLabel."
         val blockVars = irprog.st.allVariables().filter { it.name.startsWith(prefix) }
         blockVars.forEach { stVar ->
-            irprog. allSubs().flatMap { it.chunks }.forEach { chunk ->
+            irprog.allSubs().flatMap { it.chunks }.forEach { chunk ->
                 chunk.instructions.forEach { ins ->
                     if(ins.labelSymbol == stVar.name) {
-                        return
+                        return  // symbol occurs in an instruction
                     }
                 }
             }
+
+            irprog.st.allVariables().forEach { stVar->
+                val initValue = stVar.onetimeInitializationArrayValue
+                if(initValue!=null && !initValue.isEmpty()) {
+                    if(initValue.any {
+                        it.addressOfSymbol?.startsWith(blockLabel)==true
+                    })
+                        return   // symbol occurs in an initializer value (address-of this symbol)_
+                }
+            }
         }
+
         irprog.st.removeTree(blockLabel)
     }
 
@@ -207,9 +218,7 @@ class IRUnusedCodeRemover(
         return removeUnlinkedChunks(linkedChunks)
     }
 
-    private fun removeUnlinkedChunks(
-        linkedChunks: Set<IRCodeChunkBase>
-    ): Int {
+    private fun removeUnlinkedChunks(linkedChunks: Set<IRCodeChunkBase>): Int {
         var numRemoved = 0
         irprog.foreachSub { sub ->
             sub.chunks.withIndex().reversed().forEach { (index, chunk) ->

codeGenIntermediate/test/TestIRPeepholeOpt.kt

@@ -18,6 +18,7 @@ class TestIRPeepholeOpt: FunSpec({
             CbmPrgLauncherType.NONE,
             ZeropageType.DONTUSE,
             emptyList(),
+            CompilationOptions.AllZeropageAllowed,
             floats = false,
             noSysInit = true,
             compTarget = target,

codeGenIntermediate/test/TestVmCodeGen.kt

@@ -19,6 +19,7 @@ class TestVmCodeGen: FunSpec({
             CbmPrgLauncherType.NONE,
             ZeropageType.DONTUSE,
             zpReserved = emptyList(),
+            zpAllowed = CompilationOptions.AllZeropageAllowed,
             floats = true,
             noSysInit = false,
             compTarget = target,
@@ -41,7 +42,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("pi", DataType.UBYTE, ZeropageWish.DONTCARE, PtNumber(DataType.UBYTE, 0.0, Position.DUMMY), null, Position.DUMMY))
         sub.add(PtVariable("particleX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
@@ -91,7 +92,7 @@ class TestVmCodeGen: FunSpec({
         program.add(block)
 
         // define the "cx16.r0" virtual register
-        val cx16block = PtBlock("cx16", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val cx16block = PtBlock("cx16", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         cx16block.add(PtMemMapped("r0", DataType.UWORD, 100u, null, Position.DUMMY))
         program.add(cx16block)
 
@@ -120,7 +121,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -183,7 +184,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -242,7 +243,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -289,7 +290,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -352,7 +353,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -411,7 +412,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         sub.add(PtVariable("ub1", DataType.UBYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
         val if1 = PtIfElse(Position.DUMMY)
@@ -451,7 +452,7 @@ class TestVmCodeGen: FunSpec({
 //}
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
-        val block = PtBlock("main", null, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
+        val block = PtBlock("main", null, false, false, false, false, PtBlock.BlockAlignment.NONE, SourceCode.Generated("test"), Position.DUMMY)
         val romsub = PtAsmSub("routine", 0x5000u, setOf(CpuRegister.Y), emptyList(), emptyList(), false, Position.DUMMY)
         block.add(romsub)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
@@ -466,7 +467,11 @@ class TestVmCodeGen: FunSpec({
         val result = codegen.generate(program, st, options, errors) as VmAssemblyProgram
         val irChunks = (result.irProgram.blocks.first().children.single() as IRSubroutine).chunks
         irChunks.size shouldBe 1
-        val callInstr = irChunks.single().instructions.single()
+        irChunks[0].instructions.size shouldBe 2
+        val preparecallInstr = irChunks[0].instructions[0]
+        preparecallInstr.opcode shouldBe Opcode.PREPARECALL
+        preparecallInstr.immediate shouldBe 0
+        val callInstr = irChunks[0].instructions[1]
         callInstr.opcode shouldBe Opcode.CALL
         callInstr.address shouldBe 0x5000
     }

codeOptimizers/build.gradle

@@ -33,10 +33,10 @@ dependencies {
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
 }

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -19,12 +19,13 @@ import kotlin.math.floor
 class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
 
     override fun before(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
-        // @( &thing )  -->  thing
+        // @( &thing )  -->  thing  (but only if thing is a byte type!)
         val addrOf = memread.addressExpression as? AddressOf
-        return if(addrOf!=null)
-            listOf(IAstModification.ReplaceNode(memread, addrOf.identifier, parent))
-        else
-            noModifications
+        if(addrOf!=null) {
+            if(addrOf.identifier.inferType(program).isBytes)
+                return listOf(IAstModification.ReplaceNode(memread, addrOf.identifier, parent))
+        }
+        return noModifications
     }
 
     override fun after(containment: ContainmentCheck, parent: Node): Iterable<IAstModification> {
@@ -86,7 +87,8 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
 
         if(expr.operator=="==" && rightconst!=null) {
             val leftExpr = expr.left as? BinaryExpression
-            if(leftExpr!=null) {
+            // only do this shuffling when the LHS is not a constant itself (otherwise problematic nested replacements)
+            if(leftExpr?.constValue(program) != null) {
                 val leftRightConst = leftExpr.right.constValue(program)
                 if(leftRightConst!=null) {
                     when (leftExpr.operator) {
@@ -628,5 +630,4 @@ class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
             return null
         }
     }
-
 }

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -251,7 +251,16 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
             return noModifications
         }
 
-        if(decl.type== VarDeclType.CONST || decl.type== VarDeclType.VAR) {
+        if(decl.isArray && decl.type==VarDeclType.MEMORY) {
+            val memaddr = decl.value?.constValue(program)
+            if(memaddr!=null && memaddr !== decl.value) {
+                return listOf(IAstModification.SetExpression(
+                    { decl.value = it }, memaddr, decl
+                ))
+            }
+        }
+
+        if(decl.type==VarDeclType.CONST || decl.type==VarDeclType.VAR) {
             if(decl.isArray){
                 val arraysize = decl.arraysize
                 if(arraysize==null) {
@@ -276,107 +285,10 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                         return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
                     }
                 }
-                DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W, DataType.ARRAY_W_SPLIT, DataType.ARRAY_UW_SPLIT -> {
-                    val rangeExpr = decl.value as? RangeExpression
-                    if(rangeExpr!=null) {
-                        // convert the initializer range expression to an actual array
-                        val constRange = rangeExpr.toConstantIntegerRange()
-                        if(constRange?.isEmpty()==true) {
-                            if(constRange.first>constRange.last && constRange.step>=0)
-                                errors.err("descending range with positive step", decl.value?.position!!)
-                            else if(constRange.first<constRange.last && constRange.step<=0)
-                                errors.err("ascending range with negative step", decl.value?.position!!)
-                        }
-                        val declArraySize = decl.arraysize?.constIndex()
-                        if(declArraySize!=null && declArraySize!=rangeExpr.size())
-                            errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
-                        if(constRange!=null) {
-                            val eltType = rangeExpr.inferType(program).getOr(DataType.UBYTE)
-                            val newValue = if(eltType in ByteDatatypes) {
-                                ArrayLiteral(InferredTypes.InferredType.known(decl.datatype),
-                                        constRange.map { NumericLiteral(eltType, it.toDouble(), decl.value!!.position) }.toTypedArray(),
-                                        position = decl.value!!.position)
-                            } else {
-                                ArrayLiteral(InferredTypes.InferredType.known(decl.datatype),
-                                        constRange.map { NumericLiteral(eltType, it.toDouble(), decl.value!!.position) }.toTypedArray(),
-                                        position = decl.value!!.position)
-                            }
-                            return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
-                        }
-                    }
-                    val numericLv = decl.value as? NumericLiteral
-                    if(numericLv!=null && numericLv.type== DataType.FLOAT)
-                        errors.err("arraysize requires only integers here", numericLv.position)
-                    val size = decl.arraysize?.constIndex() ?: return noModifications
-                    if (rangeExpr==null && numericLv!=null) {
-                        // arraysize initializer is empty or a single int, and we know the size; create the arraysize.
-                        val fillvalue = numericLv.number.toInt()
-                        when(decl.datatype){
-                            DataType.ARRAY_UB -> {
-                                if(fillvalue !in 0..255)
-                                    errors.err("ubyte value overflow", numericLv.position)
-                            }
-                            DataType.ARRAY_B -> {
-                                if(fillvalue !in -128..127)
-                                    errors.err("byte value overflow", numericLv.position)
-                            }
-                            DataType.ARRAY_UW -> {
-                                if(fillvalue !in 0..65535)
-                                    errors.err("uword value overflow", numericLv.position)
-                            }
-                            DataType.ARRAY_W -> {
-                                if(fillvalue !in -32768..32767)
-                                    errors.err("word value overflow", numericLv.position)
-                            }
-                            else -> {}
-                        }
-                        // create the array itself, filled with the fillvalue.
-                        val array = Array(size) {fillvalue}.map { NumericLiteral(ArrayToElementTypes.getValue(decl.datatype), it.toDouble(), numericLv.position) }.toTypedArray<Expression>()
-                        val refValue = ArrayLiteral(InferredTypes.InferredType.known(decl.datatype), array, position = numericLv.position)
-                        return listOf(IAstModification.ReplaceNode(decl.value!!, refValue, decl))
-                    }
-                }
-                DataType.ARRAY_F -> {
-                    val rangeExpr = decl.value as? RangeExpression
-                    if(rangeExpr!=null) {
-                        // convert the initializer range expression to an actual array of floats
-                        val declArraySize = decl.arraysize?.constIndex()
-                        if(declArraySize!=null && declArraySize!=rangeExpr.size())
-                            errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
-                        val constRange = rangeExpr.toConstantIntegerRange()
-                        if(constRange!=null) {
-                            val newValue = ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_F),
-                                    constRange.map { NumericLiteral(DataType.FLOAT, it.toDouble(), decl.value!!.position) }.toTypedArray(),
-                                    position = decl.value!!.position)
-                            return listOf(IAstModification.ReplaceNode(decl.value!!, newValue, decl))
-                        }
-                    }
-
-                    val numericLv = decl.value as? NumericLiteral
-                    val size = decl.arraysize?.constIndex() ?: return noModifications
-                    if(rangeExpr==null && numericLv!=null) {
-                        // arraysize initializer is a single int, and we know the size.
-                        val fillvalue = numericLv.number
-                        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 { NumericLiteral(DataType.FLOAT, it, numericLv.position) }.toTypedArray<Expression>()
-                            val refValue = ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_F), array, position = numericLv.position)
-                            return listOf(IAstModification.ReplaceNode(decl.value!!, refValue, decl))
-                        }
-                    }
-                }
-                DataType.ARRAY_BOOL -> {
-                    val numericLv = decl.value as? NumericLiteral
-                    val size = decl.arraysize?.constIndex() ?: return noModifications
-                    if(numericLv!=null) {
-                        // arraysize initializer is a single int, and we know the size.
-                        val fillvalue = if(numericLv.number==0.0) 0.0 else 1.0
-                        val array = Array(size) {fillvalue}.map { NumericLiteral(DataType.UBYTE, fillvalue, numericLv.position) }.toTypedArray<Expression>()
-                        val refValue = ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_BOOL), array, position = numericLv.position)
-                        return listOf(IAstModification.ReplaceNode(decl.value!!, refValue, decl))
-                    }
+                in ArrayDatatypes -> {
+                    val replacedArrayInitializer = createConstArrayInitializerValue(decl)
+                    if(replacedArrayInitializer!=null)
+                        return listOf(IAstModification.ReplaceNode(decl.value!!, replacedArrayInitializer, decl))
                 }
                 else -> {
                     // nothing to do for this type
@@ -386,4 +298,112 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
 
         return noModifications
     }
+
+    private fun createConstArrayInitializerValue(decl: VarDecl): ArrayLiteral? {
+
+        if(decl.type==VarDeclType.MEMORY)
+            return null     // memory mapped arrays can never have an initializer value other than the address where they're mapped.
+
+        // convert the initializer range expression from a range or int, to an actual array.
+        when(decl.datatype) {
+            DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W, DataType.ARRAY_W_SPLIT, DataType.ARRAY_UW_SPLIT -> {
+                val rangeExpr = decl.value as? RangeExpression
+                if(rangeExpr!=null) {
+                    val constRange = rangeExpr.toConstantIntegerRange()
+                    if(constRange?.isEmpty()==true) {
+                        if(constRange.first>constRange.last && constRange.step>=0)
+                            errors.err("descending range with positive step", decl.value?.position!!)
+                        else if(constRange.first<constRange.last && constRange.step<=0)
+                            errors.err("ascending range with negative step", decl.value?.position!!)
+                    }
+                    val declArraySize = decl.arraysize?.constIndex()
+                    if(declArraySize!=null && declArraySize!=rangeExpr.size())
+                        errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
+                    if(constRange!=null) {
+                        val eltType = rangeExpr.inferType(program).getOr(DataType.UBYTE)
+                        return if(eltType in ByteDatatypes) {
+                            ArrayLiteral(InferredTypes.InferredType.known(decl.datatype),
+                                constRange.map { NumericLiteral(eltType, it.toDouble(), decl.value!!.position) }.toTypedArray(),
+                                position = decl.value!!.position)
+                        } else {
+                            ArrayLiteral(InferredTypes.InferredType.known(decl.datatype),
+                                constRange.map { NumericLiteral(eltType, it.toDouble(), decl.value!!.position) }.toTypedArray(),
+                                position = decl.value!!.position)
+                        }
+                    }
+                }
+                val numericLv = decl.value as? NumericLiteral
+                if(numericLv!=null && numericLv.type== DataType.FLOAT)
+                    errors.err("arraysize requires only integers here", numericLv.position)
+                val size = decl.arraysize?.constIndex() ?: return null
+                if (rangeExpr==null && numericLv!=null) {
+                    // arraysize initializer is empty or a single int, and we know the size; create the arraysize.
+                    val fillvalue = numericLv.number.toInt()
+                    when(decl.datatype){
+                        DataType.ARRAY_UB -> {
+                            if(fillvalue !in 0..255)
+                                errors.err("ubyte value overflow", numericLv.position)
+                        }
+                        DataType.ARRAY_B -> {
+                            if(fillvalue !in -128..127)
+                                errors.err("byte value overflow", numericLv.position)
+                        }
+                        DataType.ARRAY_UW -> {
+                            if(fillvalue !in 0..65535)
+                                errors.err("uword value overflow", numericLv.position)
+                        }
+                        DataType.ARRAY_W -> {
+                            if(fillvalue !in -32768..32767)
+                                errors.err("word value overflow", numericLv.position)
+                        }
+                        else -> {}
+                    }
+                    // create the array itself, filled with the fillvalue.
+                    val array = Array(size) {fillvalue}.map { NumericLiteral(ArrayToElementTypes.getValue(decl.datatype), it.toDouble(), numericLv.position) }.toTypedArray<Expression>()
+                    return ArrayLiteral(InferredTypes.InferredType.known(decl.datatype), array, position = numericLv.position)
+                }
+            }
+            DataType.ARRAY_F -> {
+                val rangeExpr = decl.value as? RangeExpression
+                if(rangeExpr!=null) {
+                    // convert the initializer range expression to an actual array of floats
+                    val declArraySize = decl.arraysize?.constIndex()
+                    if(declArraySize!=null && declArraySize!=rangeExpr.size())
+                        errors.err("range expression size (${rangeExpr.size()}) doesn't match declared array size ($declArraySize)", decl.value?.position!!)
+                    val constRange = rangeExpr.toConstantIntegerRange()
+                    if(constRange!=null) {
+                        return ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_F),
+                            constRange.map { NumericLiteral(DataType.FLOAT, it.toDouble(), decl.value!!.position) }.toTypedArray(),
+                            position = decl.value!!.position)
+                    }
+                }
+
+                val numericLv = decl.value as? NumericLiteral
+                val size = decl.arraysize?.constIndex() ?: return null
+                if(rangeExpr==null && numericLv!=null) {
+                    // arraysize initializer is a single int, and we know the size.
+                    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 { NumericLiteral(DataType.FLOAT, it, numericLv.position) }.toTypedArray<Expression>()
+                        return ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_F), array, position = numericLv.position)
+                    }
+                }
+            }
+            DataType.ARRAY_BOOL -> {
+                val numericLv = decl.value as? NumericLiteral
+                val size = decl.arraysize?.constIndex() ?: return null
+                if(numericLv!=null) {
+                    // arraysize initializer is a single int, and we know the size.
+                    val fillvalue = if(numericLv.number==0.0) 0.0 else 1.0
+                    val array = Array(size) {fillvalue}.map { NumericLiteral(DataType.UBYTE, fillvalue, numericLv.position) }.toTypedArray<Expression>()
+                    return ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_BOOL), array, position = numericLv.position)
+                }
+            }
+            else -> return null
+        }
+        return null
+    }
 }

codeOptimizers/src/prog8/optimizer/Inliner.kt

@@ -114,6 +114,11 @@ class Inliner(private val program: Program, private val options: CompilationOpti
                                 else -> false
                             }
                     }
+
+                    if(subroutine.inline && subroutine.statements.size>1) {
+                        require(subroutine.statements.size==2 && isEmptyReturn(subroutine.statements[1]))
+                        subroutine.statements.removeLast()      // get rid of the Return, to be able to inline the (single) statement preceding it.
+                    }
                 }
             }
             super.visit(subroutine)

codeOptimizers/src/prog8/optimizer/StatementOptimizer.kt

@@ -30,7 +30,7 @@ class StatementOptimizer(private val program: Program,
         if(functionCallStatement.target.nameInSource==listOf("txt", "print")) {
             val arg = functionCallStatement.args.single()
             val stringVar: IdentifierReference? = if(arg is AddressOf) {
-                arg.identifier
+                if(arg.arrayIndex==null) arg.identifier else null
             } else {
                 arg as? IdentifierReference
             }
@@ -384,7 +384,8 @@ class StatementOptimizer(private val program: Program,
     }
 
     override fun before(unrollLoop: UnrollLoop, parent: Node): Iterable<IAstModification> {
-        return if(unrollLoop.iterations<1)
+        val iterations = unrollLoop.iterations.constValue(program)?.number?.toInt()
+        return if(iterations!=null && iterations<1)
             listOf(IAstModification.Remove(unrollLoop, parent as IStatementContainer))
         else
             noModifications
@@ -429,6 +430,27 @@ class StatementOptimizer(private val program: Program,
                 }
             }
         }
+
+        val constantValue = whenStmt.condition.constValue(program)?.number
+        if(constantValue!=null) {
+            // when condition is a constant
+            var matchingChoice: WhenChoice? = null
+            loop@ for(choice in whenStmt.choices) {
+                for(value in choice.values ?: emptyList()) {
+                    if(value.constValue(program)?.number == constantValue) {
+                        matchingChoice = choice
+                        break@loop
+                    }
+                }
+            }
+            if(matchingChoice==null)
+                matchingChoice = whenStmt.choices.singleOrNull { it.values==null }
+            if(matchingChoice!=null) {
+                // get rid of the whole when-statement and just leave the matching choice
+                return listOf(IAstModification.ReplaceNode(whenStmt, matchingChoice.statements, parent))
+            }
+        }
+
         return noModifications
     }
 

compiler/build.gradle

@@ -33,14 +33,14 @@ dependencies {
     implementation project(':codeGenIntermediate')
     implementation project(':codeGenExperimental')
     implementation project(':virtualmachine')
-    implementation "org.antlr:antlr4-runtime:4.12.0"
+    implementation "org.antlr:antlr4-runtime:4.13.1"
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.5'
+    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.6'
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:1.1.18"
 
     testImplementation project(':intermediate')
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.6.2'
+    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.7.2'
 }
 
 configurations.all {
@@ -59,10 +59,10 @@ configurations {
 sourceSets {
     main {
         java {
-            srcDirs = ["${project.projectDir}/src"]
+            srcDir "${project.projectDir}/src"
         }
         resources {
-            srcDirs = ["${project.projectDir}/res"]
+            srcDir "${project.projectDir}/res"
         }
     }
     test {

compiler/res/prog8lib/atari/syslib.p8

@@ -198,6 +198,14 @@ _longcopy
         }}
     }
 
+    sub disable_caseswitch() {
+        ; no-op
+    }
+
+    sub enable_caseswitch() {
+        ; no-op
+    }
+
     inline asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         ;    TODO

compiler/res/prog8lib/atari/textio.p8

@@ -24,15 +24,27 @@ sub spc() {
     txt.chrout(' ')
 }
 
-asmsub column(ubyte col @A) clobbers(A, X, Y) {
+const uword COLCRS = 85
+const uword ROWCRS = 84
+sub column(ubyte col) {
     ; ---- set the cursor on the given column (starting with 0) on the current line
-    ;      TODO
-    %asm {{
-        rts
-    }}
+    pokew(COLCRS, col as uword)
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+sub get_column() -> ubyte {
+    return peekw(COLCRS) as ubyte
+}
+
+sub row(ubyte rownum) {
+    ; ---- set the cursor on the given row (starting with 0) on the current line
+    @(ROWCRS) = rownum
+}
+
+sub get_row() -> ubyte {
+    return @(ROWCRS)
+}
+
+asmsub  fill_screen (ubyte character @ 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)
     ;      TODO
@@ -42,7 +54,7 @@ asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
     }}
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	; TODO
@@ -114,10 +126,10 @@ asmsub  scroll_down  (bool alsocolors @ Pc) clobbers(A)  {
 }
 
 
-romsub $F2B0 = outchar(ubyte char @ A)
+romsub $F2B0 = outchar(ubyte character @ A)
 romsub $F2Fd = waitkey()
 
-asmsub chrout(ubyte char @ A) {
+asmsub chrout(ubyte character @ A) {
 	%asm {{
 		sta  _tmp_outchar+1
 		txa
@@ -364,7 +376,7 @@ asmsub  getclr  (ubyte col @A, ubyte row @Y) clobbers(Y) -> ubyte @ A {
         }}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte char, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	; TODO
 	%asm {{
@@ -372,12 +384,14 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
         }}
 }
 
-asmsub  plot  (ubyte col @ Y, ubyte row @ A) {
-	; ---- set cursor at specific position
-	; TODO
-	%asm  {{
-	    rts
-	}}
+sub  plot(ubyte col, ubyte rownum) {
+  column(col)
+  row(rownum)
+}
+
+sub get_cursor(uword colptr, uword rowptr) {
+  @(colptr) = get_column()
+  @(rowptr) = get_row()
 }
 
 asmsub width() clobbers(X,Y) -> ubyte @A {

compiler/res/prog8lib/c128/graphics.p8

@@ -1,62 +0,0 @@
-%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 {
-    %option no_symbol_prefixing
-
-    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

@@ -87,7 +87,7 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 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 $FFD2 = CHROUT(ubyte character @ A)                           ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
 romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ 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
@@ -335,7 +335,7 @@ asmsub  init_system()  {
     ; 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.
+    ; Uppercase charset is activated.
     %asm {{
         sei
         cld
@@ -734,6 +734,20 @@ _longcopy
         }}
     }
 
+    inline asmsub disable_caseswitch() {
+        %asm {{
+            lda  #$80
+            sta  247
+        }}
+    }
+
+    inline asmsub enable_caseswitch() {
+        %asm {{
+            lda  #0
+            sta  247
+        }}
+    }
+
     inline asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         %asm {{

compiler/res/prog8lib/c128/textio.p8

@@ -1,4 +1,6 @@
-; Prog8 definitions for the Text I/O and Screen routines for the Commodore-64
+; Prog8 definitions for the Text I/O and Screen routines for the Commodore-128
+; All routines work with Screencode character encoding, except `print`, `chrout` and `input_chars`,
+; these work with PETSCII encoding instead.
 
 %import syslib
 %import conv
@@ -38,6 +40,45 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
+
+asmsub get_column() -> ubyte @Y {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+asmsub row(ubyte rownum @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given row (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  cbm.PLOT
+        tax
+        clc
+        jmp  cbm.PLOT
+    }}
+}
+
+asmsub get_row() -> ubyte @X {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+sub get_cursor(uword colptr, uword rowptr) {
+    %asm {{
+        sec
+        jsr cbm.PLOT
+        tya
+        ldy #$00
+        sta (colptr),y
+        txa
+        sta (rowptr),y
+    }}
+}
+
+
 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)
@@ -238,10 +279,10 @@ _scroll_screen  ; scroll only the screen memory
 	}}
 }
 
-romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
 
 asmsub  print (str text @ AY) clobbers(A,Y)  {
-	; ---- print null terminated string from A/Y
+	; ---- print null terminated string, in PETSCII encoding, 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 cbm.CHROUT of that single char.
@@ -427,7 +468,8 @@ asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 }
 
 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)
+	; ---- Input a string (max. 80 chars) from the keyboard, in PETSCII encoding.
+	;      Returns length in Y. (string is terminated with a 0 byte as well)
 	;      It assumes the keyboard is selected as I/O channel!
 
 	%asm {{
@@ -466,7 +508,7 @@ asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)
 _mod		sta  $ffff		; modified
 		rts
 
-_screenrows	.word  $0400 + range(0, 1000, 40)
+_screenrows	.word  cbm.Screen + range(0, 1000, 40)
 	}}
 }
 
@@ -533,7 +575,7 @@ _mod		lda  $ffff		; modified
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	%asm {{
 		lda  row
@@ -545,13 +587,13 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
 		sta  _colormod+2
 		lda  setchr._screenrows,y
 		clc
-		adc  column
+		adc  col
 		sta  _charmod+1
 		sta  _colormod+1
 		bcc  +
 		inc  _charmod+2
 		inc  _colormod+2
-+		lda  char
++		lda  character
 _charmod	sta  $ffff		; modified
 		lda  charcolor
 _colormod	sta  $ffff		; modified
@@ -584,4 +626,11 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub waitkey() {
+    %asm {{
+-       jsr cbm.GETIN
+        beq -
+        rts
+    }}
+}
 }

compiler/res/prog8lib/c64/floats.asm

@@ -418,8 +418,9 @@ pushFAC1    .proc
 	rts
 	.pend
 
-popFAC1 .proc
-	; -- pop floating point value from cpu stack into FAC1
+popFAC .proc
+	; -- pop floating point value from cpu stack into FAC1 or FAC2 (
+	;    carry flag clear=FAC1, carry set=FAC2
 	; save return address
 	pla
 	sta  P8ZP_SCRATCH_W2
@@ -437,8 +438,11 @@ popFAC1 .proc
 	sta  floats.floats_temp_var
 	lda  #<floats.floats_temp_var
 	ldy  #>floats.floats_temp_var
+	bcs  +
 	jsr  floats.MOVFM
-	; re-push return address
+	jmp  ++
++       jsr  floats.CONUPK
++	; re-push return address
 	lda  P8ZP_SCRATCH_W2+1
 	pha
 	lda  P8ZP_SCRATCH_W2

compiler/res/prog8lib/c64/floats.p8

@@ -46,7 +46,7 @@ romsub $b391 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
 
 romsub $b3a2 = FREADUY(ubyte value @ Y) clobbers(A,X,Y)     ; 8 bit unsigned Y -> float in fac1
 romsub $bc3c = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
-romsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length
+romsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length.  Also see parse_f()
 romsub $aabc = FPRINTLN() clobbers(A,X,Y)                   ; print string of fac1, on one line (= with newline) destroys fac1.  (consider FOUT + STROUT as well)
 romsub $bddd = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY ($0100)
 
@@ -165,6 +165,16 @@ sub rndf() -> float {
     }}
 }
 
+asmsub parse_f(str value @AY) -> float @FAC1 {
+    %asm {{
+        sta  $22
+        sty  $23
+        jsr  prog8_lib.strlen
+        tya
+        jmp  FREADSTR
+    }}
+}
+
 
 %asminclude "library:c64/floats.asm"
 %asminclude "library:c64/floats_funcs.asm"

compiler/res/prog8lib/c64/graphics.p8

@@ -1,35 +1,46 @@
-%import textio
-
 ; bitmap pixel graphics module for the C64
 ; only black/white monochrome 320x200 for now
-; assumes bitmap screen memory is $2000-$3fff
+;
+; NOTE: For sake of speed, NO BOUNDS CHECKING is performed in most routines!
+;       You'll have to make sure yourself that you're not writing outside of bitmap boundaries!
+;
+; Sets character matrix and bitmap screen memory at a higher memory location $5c00-$7fff
+; so that the program itself can be larger without starting to overwrite the graphics memory.
 
 graphics {
     %option no_symbol_prefixing
 
-    const uword BITMAP_ADDRESS = $2000
     const uword WIDTH = 320
     const ubyte HEIGHT = 200
 
+    const uword BITMAP_ADDRESS = $6000      ; MUST BE IN REGULAR RAM if you are not messing with ROM/RAM banking. (and $2000-aligned)
+                                            ; note: this constant is also used in a asm multiplication table below!
+    const uword CHARS_ADDRESS = $5c00       ; must be in same vic memory bank as the bitmap.
+
     sub enable_bitmap_mode() {
         ; enable bitmap screen, erase it and set colors to black/white.
-        c64.SCROLY = %00111011
-        c64.SCROLX = %00001000
-        c64.VMCSB = (c64.VMCSB & %11110000) | %00001000   ; $2000-$3fff
         clear_screen(1, 0)
+        c64.SCROLY = %00111011      ; enable bitmap graphics mode
+        c64.SCROLX = %00001000      ; 40 column mode, no scrolling, multicolor bitmap off
+        c64.VMCSB = (lsb(CHARS_ADDRESS >> 6) & $F0) | (lsb((BITMAP_ADDRESS & $3fff) / $0800) << 1)     ; set bitmap address
+        c64.CIA2DDRA |= %11
+        c64.CIA2PRA = lsb(BITMAP_ADDRESS >> 14) ^ 3     ; set VIC bank.
     }
 
     sub disable_bitmap_mode() {
-        ; enables text mode, erase the text screen, color white
-        c64.SCROLY = %00011011
-        c64.SCROLX = %00001000
-        c64.VMCSB = (c64.VMCSB & %11110000) | %00000100   ; $1000-$2fff
-        txt.fill_screen(' ', 1)
+        ; enables erase the text screen, text mode
+        sys.memset(CHARS_ADDRESS, 40*25, sc:' ')
+        c64.SCROLY = %00011011      ; disable bitmap graphics mode
+        c64.SCROLX = %00001000      ; 40 column mode, no scrolling
+        c64.VMCSB  = %00010100      ; screen addresses back to defaults
+        c64.CIA2DDRA |= %11
+        c64.CIA2PRA = %11   ; back to VIC bank 0.
     }
 
     sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
         sys.memset(BITMAP_ADDRESS, 320*200/8, 0)
-        txt.fill_screen(pixelcolor << 4 | bgcolor, 0)
+        sys.memset(CHARS_ADDRESS, 40*25, pixelcolor << 4 | bgcolor)
+        sys.memset(cbm.Colors, 40*25, 255)
     }
 
     sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
@@ -127,46 +138,46 @@ graphics {
         }
     }
 
-    sub rect(uword x, ubyte y, uword width, ubyte height) {
+    sub rect(uword xx, ubyte yy, uword width, ubyte height) {
         if width==0 or height==0
             return
-        horizontal_line(x, y, width)
+        horizontal_line(xx, yy, width)
         if height==1
             return
-        horizontal_line(x, y+height-1, width)
-        vertical_line(x, y+1, height-2)
+        horizontal_line(xx, yy+height-1, width)
+        vertical_line(xx, yy+1, height-2)
         if width==1
             return
-        vertical_line(x+width-1, y+1, height-2)
+        vertical_line(xx+width-1, yy+1, height-2)
     }
 
-    sub fillrect(uword x, ubyte y, uword width, ubyte height) {
+    sub fillrect(uword xx, ubyte yy, uword width, ubyte height) {
         if width==0
             return
         repeat height {
-            horizontal_line(x, y, width)
-            y++
+            horizontal_line(xx, yy, width)
+            yy++
         }
     }
 
-    sub horizontal_line(uword x, ubyte y, uword length) {
+    sub horizontal_line(uword xx, ubyte yy, uword length) {
         if length<8 {
-            internal_plotx=x
+            internal_plotx=xx
             repeat lsb(length) {
-                internal_plot(y)
+                internal_plot(yy)
                 internal_plotx++
             }
             return
         }
 
-        ubyte separate_pixels = lsb(x) & 7
-        uword addr = get_y_lookup(y) + (x&$fff8)
+        ubyte separate_pixels = lsb(xx) & 7
+        uword pixaddr = get_y_lookup(yy) + (xx&$fff8)
 
         if separate_pixels {
             %asm {{
-                lda  addr
+                lda  pixaddr
                 sta  P8ZP_SCRATCH_W1
-                lda  addr+1
+                lda  pixaddr+1
                 sta  P8ZP_SCRATCH_W1+1
                 ldy  separate_pixels
                 lda  hline_filled_right,y
@@ -175,7 +186,7 @@ graphics {
                 ora  (P8ZP_SCRATCH_W1),y
                 sta  (P8ZP_SCRATCH_W1),y
             }}
-            addr += 8
+            pixaddr += 8
             length += separate_pixels
             length -= 8
         }
@@ -191,9 +202,9 @@ graphics {
                 ror  length
                 lsr  length+1
                 ror  length
-                lda  addr
+                lda  pixaddr
                 sta  _modified+1
-                lda  addr+1
+                lda  pixaddr+1
                 sta  _modified+2
                 lda  length
                 ora  length+1
@@ -227,16 +238,17 @@ hline_zero2
         }
     }
 
-    sub vertical_line(uword x, ubyte y, ubyte height) {
-        internal_plotx = x
+    sub vertical_line(uword xx, ubyte yy, ubyte height) {
+        internal_plotx = xx
         repeat height {
-            internal_plot(y)
-            y++
+            internal_plot(yy)
+            yy++
         }
     }
 
     sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
-        ; Midpoint algorithm
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm.
         if radius==0
             return
         ubyte @zp ploty
@@ -265,16 +277,17 @@ hline_zero2
             internal_plotx = xcenter - yy
             internal_plot(ploty)
             yy++
-            if decisionOver2<=0
-                decisionOver2 += (yy as word)*2+1
-            else {
+            if decisionOver2>=0 {
                 radius--
-                decisionOver2 += (yy as word -radius)*2+1
+                decisionOver2 -= radius*$0002
             }
+            decisionOver2 += yy*$0002
+            decisionOver2++
         }
     }
 
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
         ; Midpoint algorithm, filled
         if radius==0
             return
@@ -287,12 +300,12 @@ hline_zero2
             horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
             horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
             yy++
-            if decisionOver2<=0
-                decisionOver2 += (yy as word)*2+1
-            else {
+            if decisionOver2>=0 {
                 radius--
-                decisionOver2 += (yy as word -radius)*2+1
+                decisionOver2 -= radius*$0002
             }
+            decisionOver2 += yy*$0002
+            decisionOver2++
         }
     }
 
@@ -300,8 +313,8 @@ hline_zero2
 ; here is the non-asm code for the plot routine below:
 ;    sub plot_nonasm(uword px, ubyte py) {
 ;        ubyte[] ormask = [128, 64, 32, 16, 8, 4, 2, 1]
-;        uword addr = BITMAP_ADDRESS + 320*(py>>3) + (py & 7) + (px & %0000000111111000)
-;        @(addr) |= ormask[lsb(px) & 7]
+;        uword pixaddr = BITMAP_ADDRESS + 320*(py>>3) + (py & 7) + (px & %0000000111111000)
+;        @(pixaddr) |= ormask[lsb(px) & 7]
 ;    }
 
     inline asmsub  plot(uword plotx @XY, ubyte ploty @A) clobbers (A, X, Y) {
@@ -352,7 +365,7 @@ _ormask     .byte 128, 64, 32, 16, 8, 4, 2, 1
 ; the y lookup tables encodes this formula:  BITMAP_ADDRESS + 320*(py>>3) + (py & 7)    (y from 0..199)
 ; We use the 64tass syntax for range expressions to calculate this table on assembly time.
 
-_plot_y_values := $2000 + 320*(range(200)>>3) + (range(200) & 7)
+_plot_y_values := BITMAP_ADDRESS + 320*(range(200)>>3) + (range(200) & 7)
 
 _y_lookup_lo    .byte  <_plot_y_values
 _y_lookup_hi    .byte  >_plot_y_values
@@ -360,7 +373,7 @@ _y_lookup_hi    .byte  >_plot_y_values
         }}
     }
 
-    asmsub get_y_lookup(ubyte y @Y) -> uword @AY {
+    asmsub get_y_lookup(ubyte yy @Y) -> uword @AY {
         %asm {{
             lda  internal_plot._y_lookup_lo,y
             pha

compiler/res/prog8lib/c64/syslib.p8

@@ -91,7 +91,7 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 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 $FFD2 = CHROUT(ubyte character @ A)                      ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
 romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ 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
@@ -136,6 +136,8 @@ c64 {
     %option no_symbol_prefixing
 
         ; the default locations of the 8 sprite pointers (store address of sprite / 64)
+        ; (depending on the VIC bank and screen ram address selection these can be shifted around though,
+        ; see the two routines after this for a dynamic way of determining the correct memory location)
         &ubyte  SPRPTR0         = 2040
         &ubyte  SPRPTR1         = 2041
         &ubyte  SPRPTR2         = 2042
@@ -284,6 +286,39 @@ c64 {
 
 ; ---- end of SID registers ----
 
+
+    sub get_vic_memory_base() -> uword {
+        ; one of the 4 possible banks. $0000/$4000/$8000/$c000.
+        c64.CIA2DDRA |= %11
+        return ((c64.CIA2PRA & 3) ^ 3) as uword << 14
+    }
+
+    sub get_char_matrix_ptr() -> uword {
+        ; Usually the character screen matrix is at 1024-2039 (see above)
+        ; However the vic memory configuration can be altered and this moves these registers with it.
+        ; So this routine determines it dynamically from the VIC memory setup.
+        uword chars_matrix_offset = (c64.VMCSB & $f0) as uword << 6
+        return get_vic_memory_base() + chars_matrix_offset
+    }
+
+    sub get_bitmap_ptr() -> uword {
+        return get_vic_memory_base() + ((c64.VMCSB & %00001000) as uword << 10)
+    }
+
+    sub get_sprite_addr_ptrs() -> uword {
+        ; Usually the sprite address pointers are at addresses 2040-2047 (see above)
+        ; However the vic memory configuration can be altered and this moves these registers with it.
+        ; So this routine determines it dynamically from the VIC memory setup.
+        return get_char_matrix_ptr() + 1016
+    }
+
+    sub set_sprite_ptr(ubyte sprite_num, uword sprite_data_address) {
+        ; Sets the sprite data pointer to the given address.
+        ; Because it takes some time to calculate things based on the vic memory setup,
+        ; its only suitable if you're not continuously changing the data address.
+        ; Otherwise store the correct sprite data pointer location somewhere yourself and reuse it.
+        @(get_sprite_addr_ptrs() + sprite_num) = lsb(sprite_data_address / 64)
+    }
 }
 
 sys {
@@ -300,7 +335,7 @@ asmsub  init_system()  {
     ; 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.
+    ; Uppercase charset is activated.
     %asm {{
         sei
         cld
@@ -702,6 +737,20 @@ _longcopy
         }}
     }
 
+    inline asmsub disable_caseswitch() {
+        %asm {{
+            lda  #$80
+            sta  657
+        }}
+    }
+
+    inline asmsub enable_caseswitch() {
+        %asm {{
+            lda  #0
+            sta  657
+        }}
+    }
+
     inline asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         %asm {{

compiler/res/prog8lib/c64/textio.p8

@@ -1,4 +1,6 @@
 ; Prog8 definitions for the Text I/O and Screen routines for the Commodore-64
+; All routines work with Screencode character encoding, except `print`, `chrout` and `input_chars`,
+; these work with PETSCII encoding instead.
 
 %import syslib
 %import conv
@@ -39,7 +41,46 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
+
+asmsub get_column() -> ubyte @Y {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+asmsub row(ubyte rownum @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given row (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  cbm.PLOT
+        tax
+        clc
+        jmp  cbm.PLOT
+    }}
+}
+
+asmsub get_row() -> ubyte @X {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+sub get_cursor(uword colptr, uword rowptr) {
+    %asm {{
+        sec
+        jsr cbm.PLOT
+        tya
+        ldy #$00
+        sta (colptr),y
+        txa
+        sta (rowptr),y
+    }}
+}
+
+
+asmsub  fill_screen (ubyte character @ 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)
 
@@ -54,7 +95,7 @@ asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
 
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	%asm {{
@@ -237,10 +278,10 @@ _scroll_screen  ; scroll only the screen memory
 	}}
 }
 
-romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
 
 asmsub  print (str text @ AY) clobbers(A,Y)  {
-	; ---- print null terminated string from A/Y
+	; ---- print null terminated string, in PESCII encoding, 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 cbm.CHROUT of that single char.
@@ -426,7 +467,8 @@ asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 }
 
 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)
+	; ---- Input a string (max. 80 chars) from the keyboard, in PETSCII encoding.
+	;      Returns length in Y. (string is terminated with a 0 byte as well)
 	;      It assumes the keyboard is selected as I/O channel!
 
 	%asm {{
@@ -465,7 +507,7 @@ asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A, Y)
 _mod		sta  $ffff		; modified
 		rts
 
-_screenrows	.word  $0400 + range(0, 1000, 40)
+_screenrows	.word  cbm.Screen + range(0, 1000, 40)
 	}}
 }
 
@@ -532,7 +574,7 @@ _mod		lda  $ffff		; modified
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
 	%asm {{
 		lda  row
@@ -544,13 +586,13 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
 		sta  _colormod+2
 		lda  setchr._screenrows,y
 		clc
-		adc  column
+		adc  col
 		sta  _charmod+1
 		sta  _colormod+1
 		bcc  +
 		inc  _charmod+2
 		inc  _colormod+2
-+		lda  char
++		lda  character
 _charmod	sta  $ffff		; modified
 		lda  charcolor
 _colormod	sta  $ffff		; modified
@@ -583,4 +625,11 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub waitkey() {
+    %asm {{
+-       jsr cbm.GETIN
+        beq -
+        rts
+    }}
+}
 }

compiler/res/prog8lib/cx16/diskio.p8

@@ -1,6 +1,16 @@
 ; Commander X16 disk drive I/O routines.
 ; Largely compatible with the default C64 ones, but adds more stuff specific to the X16 as well.
 
+; NOTE: If you experience weird behavior with these routines and you are using them
+;       in the X16 emulator using HostFs, please try again with an SD-card image instead first.
+;       It is possible that there are still small differences between HostFS and actual CBM DOS in the emulator.
+;
+; About the secondary addresses:
+; for writes (new files or overwrites), you can use 1 (without the mode string) or 2-14 (with the mode string)
+; for reads (existing files) you can use 0 or 2-14 (mode string is optional)
+; for modify mode (new files* or existing files), you must use 2-14, and the mode string ,s,m is required
+
+
 %import textio
 %import string
 %import syslib
@@ -8,22 +18,33 @@
 diskio {
     %option no_symbol_prefixing
 
+    const ubyte READ_IO_CHANNEL=12
+    const ubyte WRITE_IO_CHANNEL=13
+
     ubyte drivenumber = 8
 
     sub set_drive(ubyte number) {
         drivenumber = number
     }
 
+    sub reset_read_channel() {
+        void cbm.CHKIN(READ_IO_CHANNEL)
+    }
+
+    sub reset_write_channel() {
+        cbm.CHKOUT(WRITE_IO_CHANNEL)
+    }
+
     sub directory() -> bool {
         ; -- Prints the directory contents to the screen. Returns success.
 
         cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         ubyte status = 1
         void cbm.OPEN()          ; open 12,8,0,"$"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -43,12 +64,12 @@ diskio {
             ubyte high = cbm.CHRIN()
             txt.print_uw(mkword(high, low))
             txt.spc()
-            ubyte @zp char
+            ubyte @zp character
             repeat {
-                char = cbm.CHRIN()
-                if char==0
+                character = cbm.CHRIN()
+                if character==0
                     break
-                txt.chrout(char)
+                txt.chrout(character)
             }
             txt.nl()
             void cbm.CHRIN()     ; skip 2 bytes
@@ -61,7 +82,7 @@ diskio {
 
 io_error:
         cbm.CLRCHN()        ; restore default i/o devices
-        cbm.CLOSE(12)
+        cbm.CLOSE(READ_IO_CHANNEL)
 
         if status and status & $40 == 0 {            ; bit 6=end of file
             txt.print("\ni/o error, status: ")
@@ -76,12 +97,12 @@ io_error:
     sub diskname() -> uword {
         ; returns disk label name or 0 if error
         cbm.SETNAM(3, "$")
-        cbm.SETLFS(12, diskio.drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         ubyte status = 1
         void cbm.OPEN()          ; open 12,8,0,"$=c"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -89,12 +110,12 @@ io_error:
             ; skip up to entry name
         }
 
-        cx16.r0 = &diskio.list_filename
+        cx16.r0 = &list_filename
         repeat {
             @(cx16.r0) = cbm.CHRIN()
             if @(cx16.r0)=='"' {
                 @(cx16.r0) = ' '
-                while @(cx16.r0)==' ' and cx16.r0>=&diskio.list_filename {
+                while @(cx16.r0)==' ' and cx16.r0>=&list_filename {
                     @(cx16.r0) = 0
                     cx16.r0--
                 }
@@ -106,10 +127,10 @@ io_error:
 
 io_error:
         cbm.CLRCHN()
-        cbm.CLOSE(12)
+        cbm.CLOSE(READ_IO_CHANNEL)
         if status and status & $40 == 0
             return 0
-        return diskio.list_filename
+        return list_filename
     }
 
     ; internal variables for the iterative file lister / loader
@@ -124,7 +145,7 @@ io_error:
 
     sub list_filenames(uword pattern_ptr, uword filenames_buffer, uword filenames_buf_size) -> ubyte {
         ; -- fill the provided buffer with the names of the files on the disk (until buffer is full).
-        ;    Files in the buffer are separeted by a 0 byte. You can provide an optional pattern to match against.
+        ;    Files in the buffer are separated by a 0 byte. You can provide an optional pattern to match against.
         ;    After the last filename one additional 0 byte is placed to indicate the end of the list.
         ;    Returns number of files (it skips 'dir' entries i.e. subdirectories).
         ;    Also sets carry on exit: Carry clear = all files returned, Carry set = directory has more files that didn't fit in the buffer.
@@ -150,7 +171,7 @@ io_error:
         return files_found
     }
 
-    ; ----- iterative file lister functions (uses io channel 12) -----
+    ; ----- iterative file lister functions (uses the read io channel) -----
 
     sub lf_start_list(uword pattern_ptr) -> bool {
         ; -- start an iterative file listing with optional pattern matching.
@@ -161,11 +182,11 @@ io_error:
         iteration_in_progress = true
 
         cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         void cbm.OPEN()          ; open 12,8,0,"$"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -190,7 +211,7 @@ io_error:
             return false
 
         repeat {
-            void cbm.CHKIN(12)        ; use #12 as input channel again
+            reset_read_channel()        ; use the input io channel again
 
             uword nameptr = &list_filename
             ubyte blocks_lsb = cbm.CHRIN()
@@ -209,12 +230,12 @@ io_error:
 
             ; read the filename
             repeat {
-                ubyte char = cbm.CHRIN()
-                if char==0
+                ubyte character = cbm.CHRIN()
+                if character==0
                     break
-                if char=='\"'
+                if character=='\"'
                     break
-                @(nameptr) = char
+                @(nameptr) = character
                 nameptr++
             }
 
@@ -251,32 +272,34 @@ close_end:
         ; -- end an iterative file listing session (close channels).
         if iteration_in_progress {
             cbm.CLRCHN()
-            cbm.CLOSE(12)
+            cbm.CLOSE(READ_IO_CHANNEL)
             iteration_in_progress = false
         }
     }
 
 
-    ; ----- iterative file loader functions (uses io channel 12) -----
+    ; ----- iterative file loader functions (uses the read io channel) -----
 
-    sub f_open(uword filenameptr) -> bool {
+    sub f_open(str filenameptr) -> bool {
         ; -- open a file for iterative reading with f_read
         ;    note: only a single iteration loop can be active at a time!
+        ;    Returns true if the file is successfully opened and readable.
+        ;    No need to check status(), unlike f_open_w() !
         f_close()
 
         cbm.SETNAM(string.length(filenameptr), filenameptr)
-        cbm.SETLFS(12, drivenumber, 12)     ; note: has to be 12,x,12 because otherwise f_seek doesn't work
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, READ_IO_CHANNEL)     ; note: has to be Channel,x,Channel because otherwise f_seek doesn't work
         void cbm.OPEN()          ; open 12,8,12,"filename"
         if_cc {
             if cbm.READST()==0 {
                 iteration_in_progress = true
-                void cbm.CHKIN(12)          ; use #12 as input channel
+                reset_read_channel()
                 if_cc {
                     void cbm.CHRIN()        ; read first byte to test for file not found
                     if not cbm.READST() {
-                        cbm.CLOSE(12)           ; close file because we already consumed first byte
+                        cbm.CLOSE(READ_IO_CHANNEL)           ; close file because we already consumed first byte
                         void cbm.OPEN()         ; re-open the file
-                        void cbm.CHKIN(12)
+                        cbm.CLRCHN()            ; reset default i/o channels
                         return true
                     }
                 }
@@ -290,26 +313,26 @@ close_end:
     sub f_read(uword bufferpointer, uword num_bytes) -> uword {
         ; -- read from the currently open file, up to the given number of bytes.
         ;    returns the actual number of bytes read.  (checks for End-of-file and error conditions)
-        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.macptr() yourself with the vera data register as address.
+        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.MACPTR() yourself with the vera data register as address.
         if not iteration_in_progress or not num_bytes
             return 0
 
+        reset_read_channel()
         list_blocks = 0     ; we reuse this variable for the total number of bytes read
 
-        ; commander X16 supports fast block-read via macptr() kernal call
-        uword size
+        uword readsize
         while num_bytes {
-            size = 255
-            if num_bytes<size
-                size = num_bytes
-            size = cx16.macptr(lsb(size), bufferpointer, false)
+            readsize = 255
+            if num_bytes<readsize
+                readsize = num_bytes
+            readsize = cx16.MACPTR(lsb(readsize), bufferpointer, false)     ; fast block reads
             if_cs
-                goto byte_read_loop     ; macptr block read not supported, do fallback loop
-            list_blocks += size
-            bufferpointer += size
+                goto byte_read_loop     ; MACPTR block read not supported, do fallback loop
+            list_blocks += readsize
+            bufferpointer += readsize
             if msb(bufferpointer) == $c0
                 bufferpointer = mkword($a0, lsb(bufferpointer))  ; wrap over bank boundary
-            num_bytes -= size
+            num_bytes -= readsize
             if cbm.READST() & $40 {
                 f_close()       ; end of file, close it
                 break
@@ -317,7 +340,7 @@ close_end:
         }
         return list_blocks  ; number of bytes read
 
-byte_read_loop:         ; fallback if macptr() isn't supported on the device
+byte_read_loop:         ; fallback if MACPTR isn't supported on the device
         %asm {{
             lda  bufferpointer
             sta  m_in_buffer+1
@@ -348,10 +371,11 @@ m_in_buffer     sta  $ffff
     ; optimized for Commander X16 to use MACPTR block read kernal call
     sub f_read_all(uword bufferpointer) -> uword {
         ; -- read the full contents of the file, returns number of bytes read.
-        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.macptr() yourself with the vera data register as address.
+        ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.MACPTR() yourself with the vera data register as address.
         if not iteration_in_progress
             return 0
 
+        reset_read_channel()
         uword total_read = 0
         while not cbm.READST() {
             cx16.r0 = f_read(bufferpointer, 256)
@@ -370,8 +394,7 @@ m_in_buffer     sta  $ffff
         %asm {{
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
-            ldx  #12
-            jsr  cbm.CHKIN              ; use channel 12 again for input
+            jsr  reset_read_channel
             ldy  #0
 _loop       jsr  cbm.CHRIN
             sta  (P8ZP_SCRATCH_W1),y
@@ -392,23 +415,41 @@ _end        rts
         ; -- end an iterative file loading session (close channels).
         if iteration_in_progress {
             cbm.CLRCHN()
-            cbm.CLOSE(12)
+            cbm.CLOSE(READ_IO_CHANNEL)
             iteration_in_progress = false
         }
     }
 
 
-    ; ----- iterative file writing functions (uses io channel 13) -----
+    ; ----- iterative file writing functions (uses write io channel) -----
 
-    sub f_open_w(uword filenameptr) -> bool {
-        ; -- open a file for iterative writing with f_write
+    sub f_open_w(str filename) -> bool {
+        ; -- Open a file for iterative writing with f_write
+        ;    WARNING: returns true if the open command was received by the device,
+        ;    but this can still mean the file wasn't successfully opened for writing!
+        ;    (for example, if it already exists). This is different than f_open()!
+        ;    To be 100% sure if this call was successful, you have to use status()
+        ;    and check the drive's status message!
+        return internal_f_open_w(filename, false)
+    }
+
+    sub f_open_w_seek(str filename) -> bool {
+        ; -- Open an existing file for iterative writing with f_write, and seeking with f_seek_w.
+        return internal_f_open_w(filename, true)
+    }
+
+    sub internal_f_open_w(str filename, bool open_for_seeks) -> bool {
         f_close_w()
-
-        cbm.SETNAM(string.length(filenameptr), filenameptr)
-        cbm.SETLFS(13, drivenumber, 1)
-        void cbm.OPEN()             ; open 13,8,1,"filename"
+        list_filename = filename
+        str modifier = ",s,?"
+        modifier[3] = 'w'
+        if open_for_seeks
+            modifier[3] = 'm'
+        cx16.r0L = string.append(list_filename, modifier)   ; secondary 13 requires a mode suffix to signal we're writing/modifying
+        cbm.SETNAM(cx16.r0L, list_filename)
+        cbm.SETLFS(WRITE_IO_CHANNEL, drivenumber, WRITE_IO_CHANNEL)
+        void cbm.OPEN()             ; open 13,8,13,"filename"
         if_cc {
-            cbm.CHKOUT(13)          ; use #13 as output channel
             return not cbm.READST()
         }
         f_close_w()
@@ -418,7 +459,22 @@ _end        rts
     sub f_write(uword bufferpointer, uword num_bytes) -> bool {
         ; -- write the given number of bytes to the currently open file
         if num_bytes!=0 {
-            cbm.CHKOUT(13)        ; use #13 as output channel again
+            reset_write_channel()
+            do {
+                cx16.r0 = cx16.MCIOUT(lsb(num_bytes), bufferpointer, false)     ; fast block writes
+                if_cs
+                    goto no_mciout
+                num_bytes -= cx16.r0
+                bufferpointer += cx16.r0
+                if msb(bufferpointer) == $c0
+                    bufferpointer = mkword($a0, lsb(bufferpointer))  ; wrap over bank boundary
+                if cbm.READST()!=0
+                    return false
+            } until num_bytes==0
+            return not cbm.READST()
+
+no_mciout:
+            ; the device doesn't support MCIOUT, use a normal per-byte write loop
             repeat num_bytes {
                 cbm.CHROUT(@(bufferpointer))
                 bufferpointer++
@@ -431,7 +487,7 @@ _end        rts
     sub f_close_w() {
         ; -- end an iterative file writing session (close channels).
         cbm.CLRCHN()
-        cbm.CLOSE(13)
+        cbm.CLOSE(WRITE_IO_CHANNEL)
     }
 
 
@@ -470,25 +526,25 @@ io_error:
 
 
     ; saves a block of memory to disk, including the default 2 byte prg header.
-    sub save(uword filenameptr, uword address, uword size) -> bool {
-        return internal_save_routine(filenameptr, address, size, false)
+    sub save(uword filenameptr, uword startaddress, uword savesize) -> bool {
+        return internal_save_routine(filenameptr, startaddress, savesize, false)
     }
 
     ; like save() but omits the 2 byte prg header.
-    sub save_raw(uword filenameptr, uword address, uword size) -> bool {
-        return internal_save_routine(filenameptr, address, size, true)
+    sub save_raw(uword filenameptr, uword startaddress, uword savesize) -> bool {
+        return internal_save_routine(filenameptr, startaddress, savesize, true)
     }
 
-    sub internal_save_routine(uword filenameptr, uword address, uword size, bool headerless) -> bool {
+    sub internal_save_routine(uword filenameptr, uword startaddress, uword savesize, bool headerless) -> bool {
         cbm.SETNAM(string.length(filenameptr), filenameptr)
         cbm.SETLFS(1, drivenumber, 0)
-        uword @shared end_address = address + size
+        uword @shared end_address = startaddress + savesize
         cx16.r0L = 0
 
         %asm {{
-            lda  address
+            lda  startaddress
             sta  P8ZP_SCRATCH_W1
-            lda  address+1
+            lda  startaddress+1
             sta  P8ZP_SCRATCH_W1+1
             ldx  end_address
             ldy  end_address+1
@@ -506,9 +562,6 @@ io_error:
         if_cc
             cx16.r0L = cbm.READST()==0
 
-        cbm.CLRCHN()
-        cbm.CLOSE(1)
-
         return cx16.r0L
     }
 
@@ -532,8 +585,8 @@ io_error:
     ; Identical to load(), but DOES INCLUDE the first 2 bytes in the file.
     ; No program header is assumed in the file. Everything is loaded.
     ; See comments on load() for more details.
-    sub load_raw(uword filenameptr, uword address) -> uword {
-        return internal_load_routine(filenameptr, address, true)
+    sub load_raw(uword filenameptr, uword startaddress) -> uword {
+        return internal_load_routine(filenameptr, startaddress, true)
     }
 
 
@@ -557,8 +610,6 @@ io_error:
 +
         }}
 
-        cbm.CLRCHN()
-        cbm.CLOSE(1)
         return cx16.r1
     }
 
@@ -606,7 +657,7 @@ io_error:
         return $2000 * (cx16.getrambank() - startbank) + endaddress - startaddress
     }
 
-    asmsub vload(str name @R0, ubyte bank @A, uword address @R1) clobbers(X, Y) -> ubyte @A {
+    asmsub vload(str name @R0, ubyte bank @A, uword startaddress @R1) clobbers(X, Y) -> ubyte @A {
         ; -- like the basic command VLOAD "filename",drivenumber,bank,address
         ;    loads a file into Vera's video memory in the given bank:address, returns success in A
         ;    the file has to have the usual 2 byte header (which will be skipped)
@@ -645,7 +696,7 @@ internal_vload:
         }}
     }
 
-    asmsub vload_raw(str name @R0, ubyte bank @A, uword address @R1) clobbers(X, Y) -> ubyte @A {
+    asmsub vload_raw(str name @R0, ubyte bank @A, uword startaddress @R1) clobbers(X, Y) -> ubyte @A {
         ; -- like the basic command BVLOAD "filename",drivenumber,bank,address
         ;    loads a file into Vera's video memory in the given bank:address, returns success in A
         ;    the file is read fully including the first two bytes.
@@ -693,11 +744,11 @@ internal_vload:
         cx16.r12 = reversebuffer + MAX_PATH_LEN-1
         @(cx16.r12)=0
         cbm.SETNAM(3, "$=c")
-        cbm.SETLFS(12, diskio.drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         void cbm.OPEN()          ; open 12,8,0,"$=c"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -734,7 +785,7 @@ internal_vload:
 
 io_error:
         cbm.CLRCHN()
-        cbm.CLOSE(12)
+        cbm.CLOSE(READ_IO_CHANNEL)
         if status and status & $40 == 0
             return 0
         if @(cx16.r12)==0 {
@@ -767,7 +818,7 @@ io_error:
     sub f_seek(uword pos_hiword, uword pos_loword) {
         ; -- seek in the reading file opened with f_open, to the given 32-bits position
         ubyte[6] command = ['p',0,0,0,0,0]
-        command[1] = 12       ; f_open uses channel 12
+        command[1] = READ_IO_CHANNEL       ; f_open uses this secondary address
         command[2] = lsb(pos_loword)
         command[3] = msb(pos_loword)
         command[4] = lsb(pos_hiword)
@@ -777,18 +828,22 @@ io_error:
         cbm.SETLFS(15, drivenumber, 15)
         void cbm.OPEN()
         cbm.CLOSE(15)
+        reset_read_channel()       ; back to the read io channel
     }
 
 
-    ; NOTE: f_seek_w() doesn't work reliably right now. I only manage to corrupt the fat32 filesystem on the sdcard with it...
-;    sub f_seek_w(uword pos_hiword, uword pos_loword) {
-;        ; -- seek in the output file opened with f_open_w, to the given 32-bits position
-;        diskio.f_seek.command[1] = 13       ; f_open_w uses channel 13
-;        diskio.f_seek.command[2] = lsb(pos_loword)
-;        diskio.f_seek.command[3] = msb(pos_loword)
-;        diskio.f_seek.command[4] = lsb(pos_hiword)
-;        diskio.f_seek.command[5] = msb(pos_hiword)
-;        goto diskio.f_seek.send_command
-;    }
+    sub f_seek_w(uword pos_hiword, uword pos_loword) {
+        ; -- seek in the output file opened with f_open_w_seek, to the given 32-bits position
+        diskio.f_seek.command[1] = WRITE_IO_CHANNEL       ; f_open_w uses this secondary address
+        diskio.f_seek.command[2] = lsb(pos_loword)
+        diskio.f_seek.command[3] = msb(pos_loword)
+        diskio.f_seek.command[4] = lsb(pos_hiword)
+        diskio.f_seek.command[5] = msb(pos_hiword)
+        cbm.SETNAM(sizeof(diskio.f_seek.command), &diskio.f_seek.command)
+        cbm.SETLFS(15, drivenumber, 15)
+        void cbm.OPEN()
+        cbm.CLOSE(15)
+        reset_write_channel()    ; back to the write io channel
+    }
 
 }

compiler/res/prog8lib/cx16/emudbg.p8

@@ -0,0 +1,62 @@
+; Emulator debug interface.
+; Docs: https://github.com/X16Community/x16-emulator#emulator-io-registers
+
+emudbg {
+
+    const uword EMU_BASE = $9fb0
+
+    &ubyte EMU_DBG_HOTKEY_ENABLED = EMU_BASE + 0
+    &ubyte EMU_LOG_VIDEO          = EMU_BASE + 1
+    &ubyte EMU_LOG_KEYBOARD       = EMU_BASE + 2
+    &ubyte EMU_ECHO_MODE          = EMU_BASE + 3
+    &ubyte EMU_SAVE_ON_EXIT       = EMU_BASE + 4
+    &ubyte EMU_RECORD_GIF         = EMU_BASE + 5
+    &ubyte EMU_RECORD_WAV         = EMU_BASE + 6
+    &ubyte EMU_CMDKEYS_DISABLED   = EMU_BASE + 7
+    &ubyte EMU_CPUCLK_L           = EMU_BASE + 8     ; write: reset cpu clock to 0
+    &ubyte EMU_CPUCLK_M           = EMU_BASE + 9     ; write: outputs "User debug 1: $xx"
+    &ubyte EMU_CPUCLK_H           = EMU_BASE + 10    ; write: outputs "User debug 2: $xx"
+    &ubyte EMU_CPUCLK_U           = EMU_BASE + 11    ; write: outputs as character to console
+    ; 12 is unused for now
+    &ubyte EMU_KEYMAP             = EMU_BASE + 13
+    &ubyte EMU_EMU_DETECT1        = EMU_BASE + 14
+    &ubyte EMU_EMU_DETECT2        = EMU_BASE + 15
+
+    sub is_emulator() -> bool {
+        ; Test for emulator presence.
+        ; It is recommended to only use the other debug routines if this returns true.
+        return EMU_EMU_DETECT1=='1' and EMU_EMU_DETECT2=='6'
+    }
+
+    asmsub console_write(str isoString @R0) clobbers(Y) {
+        ; note: make sure the text is in Iso encoding.
+        %asm {{
+            ldy  #0
+-           lda  (cx16.r0),y
+            beq  +
+            sta  p8_EMU_CPUCLK_U
+            iny
+            bne  -
++
+        }}
+    }
+
+    asmsub console_chrout(ubyte char @A) {
+        ; note: make sure the character is in Iso encoding.
+        %asm {{
+            sta  p8_EMU_CPUCLK_U
+        }}
+    }
+
+    asmsub console_value1(ubyte value @A) {
+        %asm {{
+            sta  p8_EMU_CPUCLK_M
+        }}
+    }
+
+    asmsub console_value2(ubyte value @A) {
+        %asm {{
+            sta  p8_EMU_CPUCLK_H
+        }}
+    }
+}

compiler/res/prog8lib/cx16/floats.p8

@@ -28,7 +28,7 @@ romsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101
 romsub $fe03 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
 
 romsub $fe06 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
-; romsub $fe09 = VAL_1() clobbers(A,X,Y)                      ; convert ASCII string to floating point [not yet implemented!!!]
+; romsub $fe09 = VAL_1() clobbers(A,X,Y)                      ; convert ASCII string to floating point [not yet implemented!!!] see parse_f() instead
 
 ; GETADR: fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
 ; (tip: use GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
@@ -136,6 +136,19 @@ asmsub  FREADUY (ubyte value @Y) {
     }}
 }
 
+asmsub parse_f(str value @AY) -> float @FAC1 {
+    ; -- parse a string value of a number to float in FAC1
+    ;    warning: uses an internal BASIC routine that may be rom version dependent
+    ;    ($ddf2 is inside the routine for VAL at $ddef)
+    %asm {{
+        sta  $a9
+        sty  $aa
+        jsr  prog8_lib.strlen
+        tya
+        jmp  $ddf2
+    }}
+}
+
 
 &uword AYINT_facmo = $c6      ; $c6/$c7 contain result of AYINT
 

compiler/res/prog8lib/cx16/graphics.p8

@@ -5,13 +5,19 @@
 ; Wraps the graphics functions that are in ROM.
 ; Only lo-res 320x240 256 color mode for now.
 ; Unlike graphics module on the C64, you can use colors() to set new drawing colors for every draw operation.
-; For other resolutions or other color modes, use the "gfx2" module instead. (which is Cx16-specific)
+; For other resolutions or other color modes, use the "gfx2" or "monogfx" module instead. (which is Cx16-specific)
 ; Note: there is no color palette manipulation here, you have to do that yourself or use the "palette" module.
+;
+; NOTE: For sake of speed, NO BOUNDS CHECKING is performed in most routines!
+;       You'll have to make sure yourself that you're not writing outside of bitmap boundaries!
+;
 
 
 graphics {
     %option no_symbol_prefixing
 
+    romsub $feff = FB_cursor_position2()  clobbers(A,X,Y)     ; alias for the normal FB_cursor_position() call but reuses existing r0 and r1
+
     const uword WIDTH = 320
     const ubyte HEIGHT = 240
 
@@ -29,7 +35,7 @@ graphics {
     sub disable_bitmap_mode() {
         ; enables text mode, erase the text screen, color white
         void cx16.screen_mode(0, false)
-        txt.fill_screen(' ', 1)     ; doesn't seem to fully clear the text screen after returning from gfx mode
+        txt.clear_screen()
     }
 
 
@@ -50,25 +56,27 @@ graphics {
         cx16.GRAPH_draw_line(x1, y1, x2, y2)
     }
 
-    sub fillrect(uword x, uword y, uword width, uword height) {
-        cx16.GRAPH_draw_rect(x, y, width, height, 0, 1)
+    sub fillrect(uword xx, uword yy, uword width, uword height) {
+        cx16.GRAPH_draw_rect(xx, yy, width, height, 0, 1)
     }
 
-    sub rect(uword x, uword y, uword width, uword height) {
-        cx16.GRAPH_draw_rect(x, y, width, height, 0, 0)
+    sub rect(uword xx, uword yy, uword width, uword height) {
+        cx16.GRAPH_draw_rect(xx, yy, width, height, 0, 0)
     }
 
-    sub horizontal_line(uword x, uword y, uword length) {
+    sub horizontal_line(uword xx, uword yy, uword length) {
         if length
-            cx16.GRAPH_draw_line(x, y, x+length-1, y)
+            cx16.GRAPH_draw_line(xx, yy, xx+length-1, yy)
     }
 
-    sub vertical_line(uword x, uword y, uword height) {
+    sub vertical_line(uword xx, uword yy, uword height) {
         if height
-            cx16.GRAPH_draw_line(x, y, x, y+height-1)
+            cx16.GRAPH_draw_line(xx, yy, xx, yy+height-1)
     }
 
     sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+
         ;cx16.r0 = xcenter - radius/2
         ;cx16.r1 = ycenter - radius/2
         ;cx16.r2 = radius*2
@@ -85,43 +93,47 @@ graphics {
         while xx>=yy {
             cx16.r0 = xcenter + xx
             cx16.r1 = ycenter + yy
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter - xx
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter + xx
             cx16.r1 = ycenter - yy
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter - xx
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter + yy
             cx16.r1 = ycenter + xx
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter - yy
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter + yy
             cx16.r1 = ycenter - xx
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
             cx16.r0 = xcenter - yy
-            cx16.FB_cursor_position2()
+            FB_cursor_position2()
             cx16.FB_set_pixel(stroke_color)
+
             yy++
-            if decisionOver2<=0 {
-                decisionOver2 += (yy as word)*2+1
-            } else {
+            if decisionOver2>=0 {
                 xx--
-                decisionOver2 += (yy as word -xx)*2+1
+                decisionOver2 -= xx*$0002
             }
+            decisionOver2 += yy*$0002
+            decisionOver2++
         }
     }
 
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm, filled
+
         if radius==0
             return
         ubyte @zp yy = 0
@@ -132,13 +144,14 @@ graphics {
             horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
             horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
             horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
+
             yy++
-            if decisionOver2<=0
-                decisionOver2 += (yy as word)*2+1
-            else {
+            if decisionOver2>=0 {
                 radius--
-                decisionOver2 += (yy as word -radius)*2+1
+                decisionOver2 -= radius*$0002
             }
+            decisionOver2 += yy*$0002
+            decisionOver2++
         }
     }
 

compiler/res/prog8lib/cx16/monogfx.p8

@@ -0,0 +1,911 @@
+; Monochrome Bitmap pixel graphics routines for the CommanderX16
+; Using the full-screen 640x480 and 320x240 screen modes, in 1 bpp mode (black/white).
+;
+; No text layer is currently shown, but text can be drawn as part of the bitmap itself.
+; For color bitmap graphics, see the gfx2 library.
+;
+; NOTE: a lot of the code here is similar or the same to that in gfx2
+; NOTE: For sake of speed, NO BOUNDS CHECKING is performed in most routines!
+;       You'll have to make sure yourself that you're not writing outside of bitmap boundaries!
+;
+
+monogfx {
+
+    %option no_symbol_prefixing
+
+    ; read-only control variables:
+    uword width = 0
+    uword height = 0
+    bool dont_stipple_flag = true            ; set to false to enable stippling mode
+
+    sub lores() {
+        ; enable 320*240 bitmap mode
+        cx16.VERA_CTRL=0
+        cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+        cx16.VERA_DC_HSCALE = 64
+        cx16.VERA_DC_VSCALE = 64
+        cx16.VERA_L1_CONFIG = %00000100
+        cx16.VERA_L1_MAPBASE = 0
+        cx16.VERA_L1_TILEBASE = 0
+        width = 320
+        height = 240
+        clear_screen(0)
+    }
+
+    sub hires() {
+        ; enable 640*480 bitmap mode
+        cx16.VERA_CTRL=0
+        cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+        cx16.VERA_DC_HSCALE = 128
+        cx16.VERA_DC_VSCALE = 128
+        cx16.VERA_L1_CONFIG = %00000100
+        cx16.VERA_L1_MAPBASE = 0
+        cx16.VERA_L1_TILEBASE = %00000001
+        width = 640
+        height = 480
+        clear_screen(0)
+    }
+
+    sub textmode() {
+        ; back to normal text mode
+        cx16.r15L = cx16.VERA_DC_VIDEO & %00000111 ; retain chroma + output mode
+        cbm.CINT()
+        cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11111000) | cx16.r15L
+    }
+
+    sub clear_screen(ubyte color) {
+        stipple(false)
+        position(0, 0)
+        when width {
+            320 -> {
+                repeat 240/2/8
+                    cs_innerloop640(color)
+            }
+            640 -> {
+                repeat 480/8
+                    cs_innerloop640(color)
+            }
+        }
+        position(0, 0)
+    }
+
+    sub stipple(bool enable) {
+        dont_stipple_flag = not enable
+    }
+
+    sub rect(uword xx, uword yy, uword rwidth, uword rheight, bool draw) {
+        if rwidth==0 or rheight==0
+            return
+        horizontal_line(xx, yy, rwidth, draw)
+        if rheight==1
+            return
+        horizontal_line(xx, yy+rheight-1, rwidth, draw)
+        vertical_line(xx, yy+1, rheight-2, draw)
+        if rwidth==1
+            return
+        vertical_line(xx+rwidth-1, yy+1, rheight-2, draw)
+    }
+
+    sub fillrect(uword xx, uword yy, uword rwidth, uword rheight, bool draw) {
+        ; Draw a filled rectangle of the given size and color.
+        ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
+        if rwidth==0
+            return
+        repeat rheight {
+            horizontal_line(xx, yy, rwidth, draw)
+            yy++
+        }
+    }
+
+    sub horizontal_line(uword xx, uword yy, uword length, bool draw) {
+        ubyte[9] masked_starts = [ 0, %00000001, %00000011, %00000111, %00001111, %00011111, %00111111, %01111111, %11111111]
+        ubyte[9] masked_ends   = [ 0, %10000000, %11000000, %11100000, %11110000, %11111000, %11111100, %11111110, %11111111]
+
+        if length==0
+            return
+        if length<=8 {
+            ; just use 2 byte writes with shifted mask
+            position2(xx,yy,true)
+            %asm {{
+                ldy  length
+                lda  masked_ends,y
+                sta  cx16.r0L           ; save left byte
+                stz  P8ZP_SCRATCH_B1
+                lda  xx
+                and  #7
+                beq  +
+                tay
+                lda  cx16.r0L
+-               lsr  a
+                ror  P8ZP_SCRATCH_B1
+                dey
+                bne  -
+                sta  cx16.r0L           ; new left byte
++
+                lda  dont_stipple_flag
+                bne  _dontstipple
+                ; determine stipple pattern
+                lda  yy
+                and  #1
+                beq  +
+                lda  #%10101010
+                bne  ++
++               lda  #%01010101
++               sta  P8ZP_SCRATCH_REG
+                lda  cx16.r0L
+                and  P8ZP_SCRATCH_REG
+                sta  cx16.r0L
+                lda  P8ZP_SCRATCH_B1
+                and  P8ZP_SCRATCH_REG
+                sta  P8ZP_SCRATCH_B1
+_dontstipple
+                lda  draw
+                beq  _clear
+                lda  cx16.r0L           ; left byte
+                ora  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                lda  P8ZP_SCRATCH_B1    ; right byte
+                ora  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                rts
+_clear
+                lda  cx16.r0L           ; left byte
+                eor  #255
+                and  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                lda  P8ZP_SCRATCH_B1    ; right byte
+                eor  #255
+                and  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                rts
+            }}
+        }
+
+        ubyte separate_pixels = (8-lsb(xx)) & 7
+        if separate_pixels {
+            if dont_stipple_flag {
+                position(xx,yy)
+                cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
+                if draw
+                    cx16.VERA_DATA0 |= masked_starts[separate_pixels]
+                else
+                    cx16.VERA_DATA0 &= ~masked_starts[separate_pixels]
+                xx += separate_pixels
+            } else {
+                repeat separate_pixels {
+                    plot(xx, yy, draw)
+                    xx++
+                }
+            }
+            length -= separate_pixels
+        }
+        if length {
+            position(xx, yy)
+            separate_pixels = lsb(length) & 7
+            xx += length & $fff8
+            %asm {{
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lda  draw
+                bne  +
+                ldy  #0     ; black
+                bra  _loop
++               lda  dont_stipple_flag
+                beq  _stipple
+                ldy  #255       ; don't stipple
+                bra  _loop
+_stipple        lda  yy
+                and  #1         ; determine stipple pattern to use
+                bne  +
+                ldy  #%01010101
+                bra  _loop
++               ldy  #%10101010
+_loop           lda  length
+                ora  length+1
+                beq  _done
+                sty  cx16.VERA_DATA0
+                lda  length
+                bne  +
+                dec  length+1
++               dec  length
+                bra  _loop
+_done
+            }}
+
+            if dont_stipple_flag {
+                cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off
+                if draw
+                    cx16.VERA_DATA0 |= masked_ends[separate_pixels]
+                else
+                    cx16.VERA_DATA0 &= ~masked_ends[separate_pixels]
+            } else {
+                repeat separate_pixels {
+                    plot(xx, yy, draw)
+                    xx++
+                }
+            }
+        }
+        cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off again
+    }
+
+    sub safe_horizontal_line(uword xx, uword yy, uword length, bool draw) {
+        ; does bounds checking and clipping
+        if msb(yy)&$80!=0 or yy>=height
+            return
+        if msb(xx)&$80!=0 {
+            length += xx
+            xx = 0
+        }
+        if xx>=width
+            return
+        if xx+length>width
+            length = width-xx
+        if length>width
+            return
+
+        horizontal_line(xx, yy, length, draw)
+    }
+
+    sub vertical_line(uword xx, uword yy, uword lheight, bool draw) {
+        cx16.r15L = monogfx.plot.maskbits[xx as ubyte & 7]           ; bitmask
+        if draw {
+            if dont_stipple_flag {
+                ; draw continuous line.
+                position2(xx,yy,true)
+                if width==320
+                    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 lheight {
+                    %asm {{
+                        lda  cx16.VERA_DATA0
+                        ora  cx16.r15L
+                        sta  cx16.VERA_DATA1
+                    }}
+                }
+            } else {
+                ; draw stippled line.
+                if (xx ^ yy)&1==0  {
+                    yy++
+                    lheight--
+                }
+                lheight++   ; because it is divided by 2 later, don't round off the last pixel
+                position2(xx,yy,true)
+                if width==320
+                    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 lheight/2 {
+                    %asm {{
+                        lda  cx16.VERA_DATA0
+                        ora  cx16.r15L
+                        sta  cx16.VERA_DATA1
+                    }}
+                }
+            }
+        } else {
+            position2(xx,yy,true)
+            cx16.r15 = ~cx16.r15    ; erase pixels
+            if width==320
+                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 lheight {
+                %asm {{
+                    lda  cx16.VERA_DATA0
+                    and  cx16.r15L
+                    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, bool draw) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        if y1>y2 {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            cx16.r0 = x1
+            x1 = x2
+            x2 = cx16.r0
+            cx16.r0 = y1
+            y1 = y2
+            y2 = cx16.r0
+        }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as uword +1, draw)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1, draw)
+            return
+        }
+
+        word @zp d = 0
+        cx16.r1L = true      ; 'positive_ix'
+        if dx < 0 {
+            dx = -dx
+            cx16.r1L = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+        cx16.r14 = x1       ; internal plot X
+
+        if dx >= dy {
+            if cx16.r1L {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14++
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14--
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if cx16.r1L {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+    }
+
+    sub circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, bool draw) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+
+            yy++
+            if decisionOver2>=0 {
+                xx--
+                decisionOver2 -= xx*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+
+        sub plotq() {
+            ; cx16.r14 = x, cx16.r15 = y, draw=draw
+            plot(cx16.r14, cx16.r15, draw)
+        }
+    }
+
+    sub safe_circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, bool draw) {
+        ; Does bounds checking and clipping.
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+
+            yy++
+            if decisionOver2>=0 {
+                xx--
+                decisionOver2 -= xx*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+
+        sub plotq() {
+            ; cx16.r14 = x, cx16.r15 = y, draw=draw
+            safe_plot(cx16.r14, cx16.r15, draw)
+        }
+    }
+
+    sub disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+    }
+
+    sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
+        ; Does bounds checking and clipping.
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
+            safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+    }
+
+    sub plot(uword @zp xx, uword @zp yy, bool @zp draw) {
+        ubyte[8] @shared maskbits = [128, 64, 32, 16, 8, 4, 2, 1]
+        if draw {
+            ; solid color or perhaps stipple
+            %asm {{
+                lda  xx
+                eor  yy
+                ora  dont_stipple_flag
+                and  #1
+            }}
+            if_nz {
+                prepare()
+                %asm {{
+                    tsb  cx16.VERA_DATA0
+                }}
+            }
+        } else {
+            ; only erase
+            prepare()
+            %asm {{
+                trb  cx16.VERA_DATA0
+            }}
+        }
+
+        sub prepare() {
+            %asm {{
+                lda  xx
+                and  #7
+                pha     ; xbits
+            }}
+            xx /= 8
+            if width==320
+                xx += yy*(320/8)
+            else
+                xx += yy*(640/8)
+            %asm {{
+                stz  cx16.VERA_CTRL
+                stz  cx16.VERA_ADDR_H
+                lda  xx+1
+                sta  cx16.VERA_ADDR_M
+                lda  xx
+                sta  cx16.VERA_ADDR_L
+                ply     ; xbits
+                lda  maskbits,y
+            }}
+        }
+    }
+
+    sub safe_plot(uword xx, uword yy, bool draw) {
+        ; A plot that does bounds checks to see if the pixel is inside the screen.
+        if msb(xx)&$80!=0 or msb(yy)&$80!=0
+            return
+        if xx >= width or yy >= height
+            return
+        plot(xx, yy, draw)
+    }
+
+    sub pget(uword @zp xx, uword yy) -> ubyte {
+        %asm {{
+            lda  xx
+            and  #7
+            pha     ; xbits
+        }}
+        xx /= 8
+        if width==320
+            xx += yy*(320/8)
+        else
+            xx += yy*(640/8)
+
+        %asm {{
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            lda  xx+1
+            sta  cx16.VERA_ADDR_M
+            lda  xx
+            sta  cx16.VERA_ADDR_L
+            ply         ; xbits
+            lda  plot.maskbits,y
+            and  cx16.VERA_DATA0
+            beq  +
+            lda  #1
++           rts
+        }}
+    }
+
+    sub fill(uword x, uword y, bool draw) {
+        ; Non-recursive scanline flood fill.
+        ; based loosely on code found here https://www.codeproject.com/Articles/6017/QuickFill-An-efficient-flood-fill-algorithm
+        ; with the fixes applied to the seedfill_4 routine as mentioned in the comments.
+        const ubyte MAXDEPTH = 64
+        word @zp xx = x as word
+        word @zp yy = y as word
+        word[MAXDEPTH] @split @shared stack_xl
+        word[MAXDEPTH] @split @shared stack_xr
+        word[MAXDEPTH] @split @shared stack_y
+        byte[MAXDEPTH] @shared stack_dy
+        cx16.r12L = 0       ; stack pointer
+        word x1
+        word x2
+        byte dy
+        cx16.r10L = draw as ubyte
+        sub push_stack(word sxl, word sxr, word sy, byte sdy) {
+            if cx16.r12L==MAXDEPTH
+                return
+            cx16.r0s = sy+sdy
+            if cx16.r0s>=0 and cx16.r0s<=height-1 {
+;;                stack_xl[cx16.r12L] = sxl
+;;                stack_xr[cx16.r12L] = sxr
+;;                stack_y[cx16.r12L] = sy
+;;                stack_dy[cx16.r12L] = sdy
+;;                cx16.r12L++
+                %asm {{
+                    ldy  cx16.r12L
+                    lda  sxl
+                    sta  stack_xl_lsb,y
+                    lda  sxl+1
+                    sta  stack_xl_msb,y
+                    lda  sxr
+                    sta  stack_xr_lsb,y
+                    lda  sxr+1
+                    sta  stack_xr_msb,y
+                    lda  sy
+                    sta  stack_y_lsb,y
+                    lda  sy+1
+                    sta  stack_y_msb,y
+                    ldy  cx16.r12L
+                    lda  sdy
+                    sta  stack_dy,y
+                    inc  cx16.r12L
+                }}
+            }
+        }
+        sub pop_stack() {
+;;            cx16.r12L--
+;;            x1 = stack_xl[cx16.r12L]
+;;            x2 = stack_xr[cx16.r12L]
+;;            y = stack_y[cx16.r12L]
+;;            dy = stack_dy[cx16.r12L]
+            %asm {{
+                dec  cx16.r12L
+                ldy  cx16.r12L
+                lda  stack_xl_lsb,y
+                sta  x1
+                lda  stack_xl_msb,y
+                sta  x1+1
+                lda  stack_xr_lsb,y
+                sta  x2
+                lda  stack_xr_msb,y
+                sta  x2+1
+                lda  stack_y_lsb,y
+                sta  yy
+                lda  stack_y_msb,y
+                sta  yy+1
+                ldy  cx16.r12L
+                lda  stack_dy,y
+                sta  dy
+            }}
+            yy+=dy
+        }
+        cx16.r11L = pget(xx as uword, yy as uword)        ; old_color
+        if cx16.r11L == cx16.r10L
+            return
+        if xx<0 or xx>width-1 or yy<0 or yy>height-1
+            return
+        push_stack(xx, xx, yy, 1)
+        push_stack(xx, xx, yy + 1, -1)
+        word left = 0
+        while cx16.r12L {
+            pop_stack()
+            xx = x1
+            while xx >= 0 {
+                if pget(xx as uword, yy as uword) != cx16.r11L
+                    break
+                xx--
+            }
+            if x1!=xx
+                horizontal_line(xx as uword+1, yy as uword, x1-xx as uword, cx16.r10L)
+            else
+                goto skip
+
+            left = xx + 1
+            if left < x1
+                push_stack(left, x1 - 1, yy, -dy)
+            xx = x1 + 1
+
+            do {
+                cx16.r9 = xx
+                while xx <= width-1 {
+                    if pget(xx as uword, yy as uword) != cx16.r11L
+                        break
+                    xx++
+                }
+                if cx16.r9!=xx
+                    horizontal_line(cx16.r9, yy as uword, (xx as uword)-cx16.r9, cx16.r10L)
+
+                push_stack(left, xx - 1, yy, dy)
+                if xx > x2 + 1
+                    push_stack(x2 + 1, xx - 1, yy, -dy)
+skip:
+                xx++
+                while xx <= x2 {
+                    if pget(xx as uword, yy as uword) == cx16.r11L
+                        break
+                    xx++
+                }
+                left = xx
+            } until xx>x2
+        }
+    }
+
+    sub position(uword @zp xx, uword yy) {
+        if width==320
+            cx16.r0 = yy*(320/8)
+        else
+            cx16.r0 = yy*(640/8)
+        cx16.vaddr(0, cx16.r0+(xx/8), 0, 1)
+    }
+
+    sub position2(uword @zp xx, uword yy, bool also_port_1) {
+        position(xx, yy)
+        if also_port_1
+            cx16.vaddr_clone(0)
+    }
+
+    const ubyte charset_bank = $1
+    const uword charset_addr = $f000       ; in bank 1, so $1f000
+
+    sub text_charset(ubyte charset) {
+        ; -- select the text charset to use with the text() routine
+        ;    the charset number is the same as for the cx16.screen_set_charset() ROM function.
+        ;    1 = ISO charset, 2 = PETSCII uppercase+graphs, 3= PETSCII uppercase+lowercase.
+        cx16.screen_set_charset(charset, 0)
+    }
+
+    sub text(uword @zp xx, uword yy, bool draw, 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.
+        uword chardataptr
+        ubyte[8] @shared char_bitmap_bytes_left
+        ubyte[8] @shared char_bitmap_bytes_right
+
+        cx16.r3 = sctextptr
+        while @(cx16.r3) {
+            chardataptr = charset_addr + @(cx16.r3) * $0008
+            ; copy the character bitmap into RAM
+            cx16.vaddr_autoincr(charset_bank, chardataptr, 0, 1)
+            %asm {{
+                ; pre-shift the bits
+                lda  text.xx
+                and  #7
+                sta  P8ZP_SCRATCH_B1
+                ldy  #0
+-               lda  cx16.VERA_DATA0
+                stz  P8ZP_SCRATCH_REG
+                ldx  P8ZP_SCRATCH_B1
+                cpx  #0
+                beq  +
+-               lsr  a
+                ror  P8ZP_SCRATCH_REG
+                dex
+                bne  -
++               sta  char_bitmap_bytes_left,y
+                lda  P8ZP_SCRATCH_REG
+                sta  char_bitmap_bytes_right,y
+                iny
+                cpy  #8
+                bne  --
+            }}
+            ; left part of shifted char
+            position2(xx, yy, true)
+            set_autoincrs()
+            if draw {
+                %asm {{
+                    ldy  #0
+-                   lda  char_bitmap_bytes_left,y
+                    ora  cx16.VERA_DATA1
+                    sta  cx16.VERA_DATA0
+                    iny
+                    cpy  #8
+                    bne  -
+                }}
+            } else {
+                %asm {{
+                    ldy  #0
+-                   lda  char_bitmap_bytes_left,y
+                    eor  #255
+                    and  cx16.VERA_DATA1
+                    sta  cx16.VERA_DATA0
+                    iny
+                    cpy  #8
+                    bne  -
+                }}
+            }
+            ; right part of shifted char
+            if lsb(xx) & 7 {
+                position2(xx+8, yy, true)
+                set_autoincrs()
+                if draw {
+                    %asm {{
+                        ldy  #0
+-                       lda  char_bitmap_bytes_right,y
+                        ora  cx16.VERA_DATA1
+                        sta  cx16.VERA_DATA0
+                        iny
+                        cpy  #8
+                        bne  -
+                    }}
+                } else {
+                    %asm {{
+                        ldy  #0
+-                       lda  char_bitmap_bytes_right,y
+                        eor  #255
+                        and  cx16.VERA_DATA1
+                        sta  cx16.VERA_DATA0
+                        iny
+                        cpy  #8
+                        bne  -
+                    }}
+                }
+            }
+            cx16.r3++
+            xx += 8
+        }
+
+        sub set_autoincrs() {
+            ; set autoincrements to go to next pixel row (40 or 80 increment)
+            if width==320 {
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
+                cx16.VERA_CTRL = 1
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
+            } else {
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
+                cx16.VERA_CTRL = 1
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
+            }
+        }
+    }
+
+    asmsub cs_innerloop640(ubyte color @A) clobbers(Y) {
+        %asm {{
+            cmp  #0
+            beq  +
+            lda  #255
++           ldy  #80
+-           sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            dey
+            bne  -
+            rts
+        }}
+    }
+}

compiler/res/prog8lib/cx16/palette.p8

@@ -79,15 +79,13 @@ palette {
 
     sub set_grayscale() {
         vera_palette_ptr = $fa00
+        cc=0
         repeat 16 {
-            cc=0
-            repeat 16 {
-                cx16.vpoke(1, vera_palette_ptr, cc)
-                vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, cc)
-                vera_palette_ptr++
-                cc += $11
-            }
+            cx16.vpoke(1, vera_palette_ptr, cc)
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, cc)
+            vera_palette_ptr++
+            cc += $11
         }
     }
 
@@ -150,40 +148,34 @@ palette {
 
     sub set_c64pepto() {
         vera_palette_ptr = $fa00
-        repeat 16 {
-            for cc in 0 to 15 {
-                uword ccp = C64_colorpalette_pepto[cc]
-                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
-                vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
-                vera_palette_ptr++
-            }
+        for cc in 0 to 15 {
+            uword ccp = C64_colorpalette_pepto[cc]
+            cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
+            vera_palette_ptr++
         }
     }
 
     sub set_c64light() {
         vera_palette_ptr = $fa00
-        repeat 16 {
-            for cc in 0 to 15 {
-                uword ccp = C64_colorpalette_light[cc]
-                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
-                vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
-                vera_palette_ptr++
-            }
+        for cc in 0 to 15 {
+            uword ccp = C64_colorpalette_light[cc]
+            cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
+            vera_palette_ptr++
         }
     }
 
     sub set_c64dark() {
         vera_palette_ptr = $fa00
-        repeat 16 {
-            for cc in 0 to 15 {
-                uword ccp = C64_colorpalette_dark[cc]
-                cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
-                vera_palette_ptr++
-                cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
-                vera_palette_ptr++
-            }
+        for cc in 0 to 15 {
+            uword ccp = C64_colorpalette_dark[cc]
+            cx16.vpoke(1, vera_palette_ptr, lsb(ccp))     ; G, B
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, msb(ccp))     ; R
+            vera_palette_ptr++
         }
     }
 

compiler/res/prog8lib/cx16/psg.p8

@@ -16,11 +16,11 @@ psg {
     const ubyte LEFT     = %01000000
     const ubyte RIGHT    = %10000000
 
-    sub voice(ubyte voice_num, ubyte channel, ubyte volume, ubyte waveform, ubyte pulsewidth) {
+    sub voice(ubyte voice_num, ubyte channel, ubyte vol, ubyte waveform, ubyte pulsewidth) {
         ; -- Enables a 'voice' on the PSG.
         ;    voice_num = 0-15, the voice number.
         ;    channel = either LEFT or RIGHT or (LEFT|RIGHT). Specifies the stereo channel(s) to use.
-        ;    volume = 0-63, the starting volume for the voice
+        ;    vol = 0-63, the starting volume for the voice
         ;    waveform = one of PULSE,SAWTOOTH,TRIANGLE,NOISE.
         ;    pulsewidth = 0-63.  Specifies the pulse width for waveform=PULSE.
         envelope_states[voice_num] = 255
@@ -30,11 +30,11 @@ psg {
         cx16.VERA_ADDR_L = lsb(cx16.r0)
         cx16.VERA_ADDR_M = msb(cx16.r0)
         cx16.VERA_ADDR_H = 1
-        cx16.VERA_DATA0 = channel | volume
+        cx16.VERA_DATA0 = channel | vol
         cx16.VERA_ADDR_L++
         cx16.VERA_DATA0 = waveform | pulsewidth
-        envelope_volumes[voice_num] = mkword(volume, 0)
-        envelope_maxvolumes[voice_num] = volume
+        envelope_volumes[voice_num] = mkword(vol, 0)
+        envelope_maxvolumes[voice_num] = vol
         sys.irqsafe_clear_irqd()
     }
 

compiler/res/prog8lib/cx16/sprites.p8

@@ -0,0 +1,109 @@
+; Simple routines to control sprites.
+; They're not written for high performance, but for simplicity.
+; That's why they control 1 sprite at a time. The exception is pos_batch().
+; which is quite efficient to update sprite positions of multiple sprites in one go.
+
+; note: sprites z-order will be in front of all layers.
+; note: collision mask is not supported here yet.
+
+sprites {
+    const ubyte SIZE_8  = 0
+    const ubyte SIZE_16 = 1
+    const ubyte SIZE_32 = 2
+    const ubyte SIZE_64 = 3
+    const ubyte COLORS_16 = 0
+    const ubyte COLORS_256 = 128
+    const uword VERA_SPRITEREGS = $fc00     ; $1fc00
+    uword @zp sprite_reg
+
+    sub init(ubyte spritenum,
+             ubyte databank, uword dataaddr,
+             ubyte width_flag, ubyte height_flag,
+             ubyte colors_flag, ubyte palette_offset_idx) {
+        hide(spritenum)
+        cx16.VERA_DC_VIDEO |= %01000000             ; enable sprites globally
+        dataaddr >>= 5
+        dataaddr |= (databank as uword)<<11
+        sprite_reg = VERA_SPRITEREGS + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(dataaddr))                    ; address 12:5
+        cx16.vpoke(1, sprite_reg+1, colors_flag | msb(dataaddr))    ; 4 bpp + address 16:13
+        cx16.vpoke(1, sprite_reg+6, %00001100)                      ; z depth %11 = in front of both layers, no flips
+        cx16.vpoke(1, sprite_reg+7, height_flag<<6 | width_flag<<4 | palette_offset_idx>>4) ; 64x64 pixels, palette offset
+    }
+
+    sub data(ubyte spritenum, ubyte bank, uword addr) {
+        addr >>= 5
+        addr |= (bank as uword)<<11
+        sprite_reg = VERA_SPRITEREGS + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(addr))                    ; address 12:5
+        cx16.vpoke_mask(1, sprite_reg+1, %11110000, msb(addr))    ; address 16:13
+    }
+
+    sub pos(ubyte spritenum, word xpos, word ypos) {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(xpos))
+        cx16.vpoke(1, sprite_reg+1, msb(xpos))
+        cx16.vpoke(1, sprite_reg+2, lsb(ypos))
+        cx16.vpoke(1, sprite_reg+3, msb(ypos))
+    }
+
+    sub pos_batch(ubyte first_spritenum, ubyte num_sprites, uword xpositions_ptr, uword ypositions_ptr) {
+        ; -- note: the x and y positions word arrays must be regular arrays, they cannot be @split arrays!
+        sprite_reg = VERA_SPRITEREGS + 2 + first_spritenum*$0008
+        cx16.vaddr_autoincr(1, sprite_reg, 0, 8)
+        cx16.vaddr_autoincr(1, sprite_reg+1, 1, 8)
+        repeat num_sprites {
+            cx16.VERA_DATA0 = @(xpositions_ptr)
+            xpositions_ptr ++
+            cx16.VERA_DATA1 = @(xpositions_ptr)
+            xpositions_ptr ++
+        }
+        sprite_reg += 2
+        cx16.vaddr_autoincr(1, sprite_reg, 0, 8)
+        cx16.vaddr_autoincr(1, sprite_reg+1, 1, 8)
+        repeat num_sprites {
+            cx16.VERA_DATA0 = @(ypositions_ptr)
+            ypositions_ptr ++
+            cx16.VERA_DATA1 = @(ypositions_ptr)
+            ypositions_ptr ++
+        }
+    }
+
+    sub setx(ubyte spritenum, word xpos) {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(xpos))
+        cx16.vpoke(1, sprite_reg+1, msb(xpos))
+    }
+
+    sub sety(ubyte spritenum, word ypos) {
+        sprite_reg = VERA_SPRITEREGS + 4 + spritenum*$0008
+        cx16.vpoke(1, sprite_reg, lsb(ypos))
+        cx16.vpoke(1, sprite_reg+1, msb(ypos))
+    }
+
+    sub getx(ubyte spritenum) -> word {
+        sprite_reg = VERA_SPRITEREGS + 2 + spritenum*$0008
+        return mkword(cx16.vpeek(1, sprite_reg+1), cx16.vpeek(1, sprite_reg)) as word
+    }
+
+    sub gety(ubyte spritenum) -> word {
+        sprite_reg = VERA_SPRITEREGS + 4 + spritenum*$0008
+        return mkword(cx16.vpeek(1, sprite_reg+1), cx16.vpeek(1, sprite_reg)) as word
+    }
+
+    sub hide(ubyte spritenum) {
+        pos(spritenum, -64, -64)
+    }
+
+    sub flipx(ubyte spritenum, bool flipped) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11111110, flipped)
+    }
+
+    sub flipy(ubyte spritenum, bool flipped) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11111101, flipped<<1)
+    }
+
+    sub set_palette_offset(ubyte spritenum, ubyte offset) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 7 + spritenum*$0008, %11110000, offset>>4)
+    }
+}

compiler/res/prog8lib/cx16/syslib.p8

@@ -11,9 +11,9 @@ cbm {
 ; CLEARSCR -> use txt.clear_screen
 ; 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)
-romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
+romsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip, including resetting to the default color palette
+romsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, IRQ, ...)
+romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, screen
 romsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
 romsub $FF8D = VECTOR(uword userptr @ XY, bool dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
 romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
@@ -38,9 +38,9 @@ romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798
 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 $FFD2 = CHROUT(ubyte character @ A)                           ; (via 806 ($326)) output a character
 romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
-romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) clobbers (X, Y) -> bool @ Pc, ubyte @ A       ; (via 818 ($332)) save to a device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) clobbers (X, Y) -> bool @ Pc, ubyte @ A       ; (via 818 ($332)) save to a device.  See also BSAVE
 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) -> bool @ Pz, ubyte @ A      ; (via 808 ($328)) check the STOP key (and some others in A)
@@ -120,6 +120,9 @@ cx16 {
     &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
 
+    &uword  edkeyvec    = $ac03     ; for intercepting BASIN/CHRIN key strokes. See set_basin_handler()
+    &uword  edkeybk     = $ac05     ; ...the RAM bank of this routine if not in low ram
+
 
 ; the sixteen virtual 16-bit registers in both normal unsigned mode and signed mode (s)
     &uword r0  = $0002
@@ -267,6 +270,17 @@ cx16 {
     &ubyte  VERA_AUDIO_DATA     = VERA_BASE + $001D
     &ubyte  VERA_SPI_DATA       = VERA_BASE + $001E
     &ubyte  VERA_SPI_CTRL       = VERA_BASE + $001F
+
+    ; experimental Vera FX registers: (depends on particular value set in VERA_CTRL!!!)
+    &ubyte VERA_FX_CTRL         = VERA_BASE + $0009
+    &ubyte VERA_FX_MULT         = VERA_BASE + $000C
+    &ubyte VERA_FX_CACHE_L      = VERA_BASE + $0009
+    &ubyte VERA_FX_CACHE_M      = VERA_BASE + $000A
+    &ubyte VERA_FX_CACHE_H      = VERA_BASE + $000B
+    &ubyte VERA_FX_CACHE_U      = VERA_BASE + $000C
+    &ubyte VERA_FX_ACCUM_RESET  = VERA_BASE + $0009     ; (DCSEL=6)
+
+
 ; VERA_PSG_BASE     = $1F9C0
 ; VERA_PALETTE_BASE = $1FA00
 ; VERA_SPRITES_BASE = $1FC00
@@ -320,17 +334,16 @@ cx16 {
 ; spelling of the names is taken from the Commander X-16 rom sources
 
 ; supported C128 additions
-romsub $ff4a = close_all(ubyte device @A)  clobbers(A,X,Y)
-romsub $ff59 = lkupla(ubyte la @A)  clobbers(A,X,Y)
-romsub $ff5c = lkupsa(ubyte sa @Y)  clobbers(A,X,Y)
+romsub $ff4a = CLOSE_ALL(ubyte device @A)  clobbers(A,X,Y)
+romsub $ff59 = LKUPLA(ubyte la @A)  clobbers(A,X,Y)
+romsub $ff5c = LKUPSA(ubyte sa @Y)  clobbers(A,X,Y)
 romsub $ff5f = screen_mode(ubyte mode @A, bool getCurrent @Pc)  clobbers(X, Y) -> ubyte @A, bool @Pc        ; note: X,Y size result is not supported, use SCREEN or get_screen_mode routine for that
 romsub $ff62 = screen_set_charset(ubyte charset @A, uword charsetptr @XY)  clobbers(A,X,Y)      ; incompatible with C128  dlchr()
 ; not yet supported: romsub $ff65 = pfkey()  clobbers(A,X,Y)
-romsub $ff6e = jsrfar()  ; following word = address to call, byte after that=rom/ram bank it is in
+romsub $ff6e = JSRFAR()  ; following word = address to call, byte after that=rom/ram bank it is in
 romsub $ff74 = fetch(ubyte bank @X, ubyte index @Y)  clobbers(X) -> ubyte @A
 romsub $ff77 = stash(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)
-romsub $ff7a = cmpare(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)
-romsub $ff7d = primm()
+romsub $ff7d = PRIMM()
 
 ; It's not documented what registers are clobbered, so we assume the worst for all following kernal routines...:
 
@@ -346,15 +359,14 @@ romsub $ff35 = GRAPH_draw_oval(uword x @R0, uword y @R1, uword width @R2, uword
 romsub $ff38 = GRAPH_draw_image(uword x @R0, uword y @R1, uword ptr @R2, uword width @R3, uword height @R4)  clobbers(A,X,Y)
 romsub $ff3b = GRAPH_set_font(uword fontptr @R0)  clobbers(A,X,Y)
 romsub $ff3e = GRAPH_get_char_size(ubyte baseline @A, ubyte width @X, ubyte height_or_style @Y, bool is_control @Pc)  clobbers(A,X,Y)
-romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte char @A)  clobbers(A,X,Y)
-romsub $ff41 = GRAPH_put_next_char(ubyte char @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
+romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte character @A)  clobbers(A,X,Y)
+romsub $ff41 = GRAPH_put_next_char(ubyte character @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
 
 ; framebuffer
 romsub $fef6 = FB_init()  clobbers(A,X,Y)
 romsub $fef9 = FB_get_info()  clobbers(X,Y) -> byte @A, uword @R0, uword @R1    ; width=r0, height=r1
 romsub $fefc = FB_set_palette(uword pointer @R0, ubyte index @A, ubyte colorcount @X)  clobbers(A,X,Y)
 romsub $feff = FB_cursor_position(uword x @R0, uword y @R1)  clobbers(A,X,Y)
-romsub $feff = FB_cursor_position2()  clobbers(A,X,Y)           ;  alias for the previous routine, but avoiding having to respecify both x and y every time
 romsub $ff02 = FB_cursor_next_line(uword x @R0)  clobbers(A,X,Y)
 romsub $ff05 = FB_get_pixel()  clobbers(X,Y) -> ubyte @A
 romsub $ff08 = FB_get_pixels(uword pointer @R0, uword count @R1)  clobbers(A,X,Y)
@@ -377,14 +389,15 @@ romsub $fee7 = memory_copy(uword source @R0, uword target @R1, uword num_bytes @
 romsub $feea = memory_crc(uword address @R0, uword num_bytes @R1)  clobbers(A,X,Y) -> uword @R2
 romsub $feed = memory_decompress(uword input @R0, uword output @R1)  clobbers(A,X,Y) -> uword @R1       ; last address +1 is result in R1
 romsub $fedb = console_init(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
-romsub $fede = console_put_char(ubyte char @A, bool wrapping @Pc)  clobbers(A,X,Y)
+romsub $fede = console_put_char(ubyte character @A, bool wrapping @Pc)  clobbers(A,X,Y)
 romsub $fee1 = console_get_char()  clobbers(X,Y) -> ubyte @A
 romsub $fed8 = console_put_image(uword pointer @R0, uword width @R1, uword height @R2)  clobbers(A,X,Y)
 romsub $fed5 = console_set_paging_message(uword msgptr @R0)  clobbers(A,X,Y)
 romsub $fecf = entropy_get() -> ubyte @A, ubyte @X, ubyte @Y
 romsub $fecc = monitor()  clobbers(A,X,Y)
 
-romsub $ff44 = macptr(ubyte length @A, uword buffer @XY, bool dontAdvance @Pc)  clobbers(A) -> bool @Pc, uword @XY
+romsub $ff44 = MACPTR(ubyte length @A, uword buffer @XY, bool dontAdvance @Pc)  clobbers(A) -> bool @Pc, uword @XY
+romsub $feb1 = MCIOUT(ubyte length @A, uword buffer @XY, bool dontAdvance @Pc)  clobbers(A) -> bool @Pc, uword @XY
 romsub $ff47 = enter_basic(bool cold_or_warm @Pc)  clobbers(A,X,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()
@@ -395,7 +408,8 @@ romsub $febd = kbdbuf_peek() -> ubyte @A, ubyte @X     ; key in A, queue length
 romsub $febd = kbdbuf_peek2() -> uword @AX             ; alternative to above to not have the hassle to deal with multiple return values
 romsub $fec0 = kbdbuf_get_modifiers() -> ubyte @A
 romsub $fec3 = kbdbuf_put(ubyte key @A) clobbers(X)
-romsub $ff68 = mouse_config(ubyte shape @A, ubyte resX @X, ubyte resY @Y)  clobbers (A, X, Y)
+romsub $fed2 = keymap(uword identifier @XY, bool read @Pc) -> bool @Pc
+romsub $ff68 = mouse_config(byte shape @A, ubyte resX @X, ubyte resY @Y)  clobbers (A, X, Y)
 romsub $ff6b = mouse_get(ubyte zpdataptr @X) -> ubyte @A
 romsub $ff71 = mouse_scan()  clobbers(A, X, Y)
 romsub $ff53 = joystick_scan()  clobbers(A, X, Y)
@@ -403,10 +417,10 @@ romsub $ff56 = joystick_get(ubyte joynr @A) -> ubyte @A, ubyte @X, ubyte @Y
 romsub $ff56 = joystick_get2(ubyte joynr @A) clobbers(Y) -> uword @AX   ; alternative to above to not have the hassle to deal with multiple return values
 
 ; Audio (rom bank 10)
-romsub $C04B = psg_init() clobbers(A,X,Y)
+romsub $C04B = psg_init() clobbers(A,X,Y)                              ; (re)init Vera PSG
 romsub $C063 = ym_init() clobbers(A,X,Y) -> bool @Pc                   ; (re)init YM chip
 romsub $C066 = ym_loaddefpatches() clobbers(A,X,Y) -> bool @Pc         ; load default YM patches
-romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc                ; (re)initialize PSG and YM audio chips
+romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc                ; (re)initialize both vera PSG and YM audio chips
 ; TODO: add more of the audio routines?
 
 
@@ -438,7 +452,7 @@ asmsub kbdbuf_clear() {
     }}
 }
 
-asmsub mouse_config2(ubyte shape @A) clobbers (A, X, Y) {
+asmsub mouse_config2(byte shape @A) clobbers (A, X, Y) {
     ; -- convenience wrapper function that handles the screen resolution for mouse_config() for you
     %asm {{
         pha                         ; save shape
@@ -514,15 +528,12 @@ asmsub vpeek(ubyte bank @A, uword address @XY) -> ubyte @A {
         ; -- get a byte from VERA's video memory
         ;    note: inefficient when reading multiple sequential bytes!
         %asm {{
-                pha
-                lda  #1
-                sta  cx16.VERA_CTRL
-                pla
+                stz  cx16.VERA_CTRL
                 and  #1
                 sta  cx16.VERA_ADDR_H
                 sty  cx16.VERA_ADDR_M
                 stx  cx16.VERA_ADDR_L
-                lda  cx16.VERA_DATA1
+                lda  cx16.VERA_DATA0
                 rts
             }}
 }
@@ -819,6 +830,24 @@ asmsub restore_vera_context() clobbers(A) {
     }}
 }
 
+
+    asmsub set_chrin_keyhandler(ubyte handlerbank @A, uword handler @XY) clobbers(A) {
+        ; Install a custom CHRIN (BASIN) key handler. Call this before each line you want to read.
+        ; See https://github.com/X16Community/x16-docs/blob/master/X16%20Reference%20-%2002%20-%20Editor.md#custom-basin-petscii-code-override-handler
+        %asm {{
+            sei
+            sta  cx16.edkeybk
+            lda  $00
+            pha
+            stz  $00
+            stx  cx16.edkeyvec
+            sty  cx16.edkeyvec+1
+            pla
+            sta  $00
+            cli
+            rts
+        }}
+    }
 }
 
 sys {
@@ -860,6 +889,8 @@ asmsub  init_system()  {
         jsr  cbm.CHROUT
         lda  #147       ; clear screen
         jsr  cbm.CHROUT
+        lda  #8         ; disable charset case switch
+        jsr  cbm.CHROUT
         lda  #PROG8_VARSHIGH_RAMBANK
         sta  $00    ; select ram bank
         lda  #0
@@ -1237,6 +1268,20 @@ _longcopy
         }}
     }
 
+    inline asmsub disable_caseswitch() {
+        %asm {{
+            lda  #8
+            jsr  cbm.CHROUT
+        }}
+    }
+
+    inline asmsub enable_caseswitch() {
+        %asm {{
+            lda  #9
+            jsr  cbm.CHROUT
+        }}
+    }
+
     inline asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         %asm {{

compiler/res/prog8lib/cx16/textio.p8

@@ -1,4 +1,6 @@
 ; Prog8 definitions for the Text I/O and Screen routines for the CommanderX16
+; All routines work with Screencode character encoding, except `print`, `chrout` and `input_chars`,
+; these work with PETSCII encoding instead.
 
 %import syslib
 %import conv
@@ -42,7 +44,44 @@ asmsub column(ubyte col @A) clobbers(A, X, Y) {
     }}
 }
 
-asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A, X)  {
+asmsub get_column() -> ubyte @Y {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+asmsub row(ubyte rownum @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given row (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  cbm.PLOT
+        tax
+        clc
+        jmp  cbm.PLOT
+    }}
+}
+
+asmsub get_row() -> ubyte @X {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+sub get_cursor(uword colptr, uword rowptr) {
+    %asm {{
+        sec
+        jsr cbm.PLOT
+        tya
+        ldy #$00
+        sta (colptr),y
+        txa
+        sta (rowptr),y
+    }}
+}
+
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A, X)  {
 	; ---- fill the character screen with the given fill character and character color.
 	%asm {{
         sty  _ly+1
@@ -88,7 +127,7 @@ set_vera_textmatrix_addresses:
         }}
 }
 
-asmsub  clear_screenchars (ubyte char @ A) clobbers(X, Y)  {
+asmsub  clear_screenchars (ubyte character @ A) clobbers(X, Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	%asm {{
@@ -384,10 +423,10 @@ _nextline
 	}}
 }
 
-romsub $FFD2 = chrout(ubyte char @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse.
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
 
 asmsub  print (str text @ AY) clobbers(A,Y)  {
-	; ---- print null terminated string from A/Y
+	; ---- print null terminated string, in PETSCII encoding, 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 cbm.CHROUT of that single char.
@@ -572,7 +611,8 @@ asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
 }
 
 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)
+	; ---- Input a string (max. 80 chars) from the keyboard, in PETSCII encoding.
+	;      Returns length in Y. (string is terminated with a 0 byte as well)
 	;      It assumes the keyboard is selected as I/O channel!
 
 	%asm {{
@@ -674,12 +714,12 @@ asmsub  getclr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
 	}}
 }
 
-sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor) {
+sub  setcc  (ubyte col, ubyte row, ubyte character, 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 and leaves the background color as is.
 	;            Use setcc2 if you want Cx-16 specific feature of setting both Bg+Fg colors (is faster as well).
 	%asm {{
-            lda  column
+            lda  col
             asl  a
             tax
             ldy  row
@@ -691,7 +731,7 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor) {
             ;clc
             adc  #>VERA_TEXTMATRIX_ADDR
             sta  cx16.VERA_ADDR_M
-            lda  char
+            lda  character
             sta  cx16.VERA_DATA0
             inc  cx16.VERA_ADDR_L
             lda  charcolor
@@ -705,12 +745,12 @@ sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor) {
     }}
 }
 
-sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
+sub  setcc2  (ubyte col, ubyte row, ubyte character, 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! Is a bit faster than setcc() too.
 	%asm {{
-            lda  column
+            lda  col
             asl  a
             tax
             ldy  row
@@ -722,7 +762,7 @@ sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
             ; clc
             adc  #>VERA_TEXTMATRIX_ADDR
             sta  cx16.VERA_ADDR_M
-            lda  char
+            lda  character
             sta  cx16.VERA_DATA0
             inc  cx16.VERA_ADDR_L
             lda  colors
@@ -756,4 +796,11 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub waitkey() {
+    %asm {{
+-       jsr cbm.GETIN
+        beq -
+        rts
+    }}
+}
 }

compiler/res/prog8lib/cx16/verafx.p8

@@ -0,0 +1,131 @@
+; Experimental Vera FX support.
+; Docs:
+; https://github.com/X16Community/x16-docs/blob/master/VERA%20FX%20Reference.md
+; https://docs.google.com/document/d/1q34uWOiM3Be2pnaHRVgSdHySI-qsiQWPTo_gfE54PTg/edit
+
+verafx {
+    %option no_symbol_prefixing
+
+    sub available() -> bool {
+        ; returns true if Vera FX is available (Vera V0.3.1 or later), false if not.
+        cx16.r1L = 0
+        cx16.r0L = cx16.VERA_CTRL
+        cx16.VERA_CTRL = $7e
+        if cx16.VERA_DC_VER0 == $56 {
+            ; Vera version number is valid.
+            ; Vera fx is available on Vera version 0.3.1 and later,
+            ; so no need to even check VERA_DC_VER1, which contains 0 (or higher)
+            cx16.r1L = mkword(cx16.VERA_DC_VER2, cx16.VERA_DC_VER3) >= $0301
+        }
+        cx16.VERA_CTRL = cx16.r0L
+        return cx16.r1L
+    }
+
+    sub clear(ubyte vbank, uword vaddr, ubyte data, uword amountof32bits) {
+        ; use cached 4-byte write to quickly clear a portion of the video memory to a given byte value
+        ; this routine is around 3 times faster as gfx2.clear_screen()
+        cx16.VERA_CTRL = 0
+        cx16.VERA_ADDR_H = vbank | %00110000       ; 4-byte increment
+        cx16.VERA_ADDR_M = msb(vaddr)
+        cx16.VERA_ADDR_L = lsb(vaddr)
+        cx16.VERA_CTRL = 6<<1       ; dcsel = 6, fill the 32 bits cache
+        cx16.VERA_FX_CACHE_L = data
+        cx16.VERA_FX_CACHE_M = data
+        cx16.VERA_FX_CACHE_H = data
+        cx16.VERA_FX_CACHE_U = data
+        cx16.VERA_CTRL = 2<<1       ; dcsel = 2
+        cx16.VERA_FX_MULT = 0
+        cx16.VERA_FX_CTRL = %01000000    ; cache write enable
+
+        if (amountof32bits & %1111110000000011) == 0 {
+            repeat lsb(amountof32bits >> 2)
+                unroll 4 cx16.VERA_DATA0=0       ; write 4 bytes at a time, unrolled
+        }
+        else if (amountof32bits & %1111111000000001) == 0 {
+            repeat lsb(amountof32bits >> 1)
+                unroll 2 cx16.VERA_DATA0=0       ; write 4 bytes at a time, unrolled
+        }
+        else if (lsb(amountof32bits) & 3) == 0 {
+            repeat amountof32bits >> 2
+                unroll 4 cx16.VERA_DATA0=0       ; write 4 bytes at a time, unrolled
+        }
+        else if (lsb(amountof32bits) & 1) == 0 {
+            repeat amountof32bits >> 1
+                unroll 2 cx16.VERA_DATA0=0       ; write 4 bytes at a time, unrolled
+        }
+        else {
+            repeat amountof32bits
+                cx16.VERA_DATA0=0       ; write 4 bytes at a time
+        }
+
+        cx16.VERA_FX_CTRL = 0       ; cache write disable
+        cx16.VERA_CTRL = 0
+    }
+
+    ; unsigned multiplication just passes the values as signed to muls
+    ; if you do this yourself in your call to muls, it will save a few instructions.
+    sub mult(uword value1, uword value2) -> uword {
+        ; Returns the lower 16 bits of the 32 bits result,
+        ; the upper 16 bits are stored in cx16.r0 so you can access those separately.
+        ; It's not part of the subroutine's signature to avoid awkward use of multiple returnvalues.
+        return muls(value1 as word, value2 as word) as uword
+    }
+
+    asmsub muls(word value1 @R0, word value2 @R1) clobbers(X) -> word @AY {
+        ; Returns the lower 16 bits of the 32 bits result in AY,
+        ; the upper 16 bits are stored in cx16.r0 so you can access those separately.
+        ; It's not part of the subroutine's signature to avoid awkward use of multiple returnvalues.
+        %asm {{
+            lda  #(2 << 1)
+            sta  cx16.VERA_CTRL        ; $9F25
+            stz  cx16.VERA_FX_CTRL     ; $9F29 (mainly to reset Addr1 Mode to 0)
+            lda  #%00010000
+            sta  cx16.VERA_FX_MULT     ; $9F2C
+            lda  #(6 << 1)
+            sta  cx16.VERA_CTRL        ; $9F25
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sta  cx16.VERA_FX_CACHE_L  ; $9F29
+            sty  cx16.VERA_FX_CACHE_M  ; $9F2A
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            sta  cx16.VERA_FX_CACHE_H  ; $9F2B
+            sty  cx16.VERA_FX_CACHE_U  ; $9F2C
+            lda  cx16.VERA_FX_ACCUM_RESET   ; $9F29 (DCSEL=6)
+
+            ; Set the ADDR0 pointer to $1f9bc and write our multiplication result there
+            ; (these are the 4 bytes just before the PSG registers start)
+            lda  #(2 << 1)
+            sta  cx16.VERA_CTRL
+            lda  #%01000000           ; Cache Write Enable
+            sta  cx16.VERA_FX_CTRL
+            lda  #$bc
+            sta  cx16.VERA_ADDR_L
+            lda  #$f9
+            sta  cx16.VERA_ADDR_M
+            lda  #$01
+            sta  cx16.VERA_ADDR_H     ; no increment
+            stz  cx16.VERA_DATA0      ; multiply and write out result
+            lda  #%00010001           ; $01 with Increment 1
+            sta  cx16.VERA_ADDR_H     ; so we can read out the result
+            lda  cx16.VERA_DATA0
+            ldy  cx16.VERA_DATA0
+            ldx  cx16.VERA_DATA0      ; store the upper 16 bits of the result in r0
+            stx  cx16.r0
+            ldx  cx16.VERA_DATA0
+            stx  cx16.r0+1
+            stz  cx16.VERA_FX_CTRL    ; Cache write disable
+            stz  cx16.VERA_CTRL       ; reset DCSEL
+            rts
+        }}
+    }
+
+    sub transparency(bool enable) {
+        cx16.VERA_CTRL = 2<<1       ; dcsel = 2
+        if enable
+            cx16.VERA_FX_CTRL |= %10000000
+        else
+            cx16.VERA_FX_CTRL &= %01111111
+        cx16.VERA_CTRL = 0
+    }
+}

compiler/res/prog8lib/diskio.p8

@@ -7,22 +7,33 @@
 diskio {
     %option no_symbol_prefixing
 
+    const ubyte READ_IO_CHANNEL=12
+    const ubyte WRITE_IO_CHANNEL=13
+
     ubyte drivenumber = 8
 
     sub set_drive(ubyte number) {
         drivenumber = number
     }
 
+    sub reset_read_channel() {
+        void cbm.CHKIN(READ_IO_CHANNEL)
+    }
+
+    sub reset_write_channel() {
+        cbm.CHKOUT(WRITE_IO_CHANNEL)
+    }
+
     sub directory() -> bool {
         ; -- Prints the directory contents to the screen. Returns success.
 
         cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         ubyte status = 1
         void cbm.OPEN()          ; open 12,8,0,"$"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -42,12 +53,12 @@ diskio {
             ubyte high = cbm.CHRIN()
             txt.print_uw(mkword(high, low))
             txt.spc()
-            ubyte @zp char
+            ubyte @zp character
             repeat {
-                char = cbm.CHRIN()
-                if char==0
+                character = cbm.CHRIN()
+                if character==0
                     break
-                txt.chrout(char)
+                txt.chrout(character)
             }
             txt.nl()
             void cbm.CHRIN()     ; skip 2 bytes
@@ -60,7 +71,7 @@ diskio {
 
 io_error:
         cbm.CLRCHN()        ; restore default i/o devices
-        cbm.CLOSE(12)
+        cbm.CLOSE(READ_IO_CHANNEL)
 
         if status and status & $40 == 0 {            ; bit 6=end of file
             txt.print("\ni/o error, status: ")
@@ -76,12 +87,12 @@ io_error:
         ; -- Returns pointer to disk name string or 0 if failure.
 
         cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         ubyte okay = false
         void cbm.OPEN()          ; open 12,8,0,"$"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -96,7 +107,7 @@ io_error:
             @(cx16.r0) = cbm.CHRIN()
             if @(cx16.r0)=='"' {
                 @(cx16.r0) = ' '
-                while @(cx16.r0)==' ' and cx16.r0>=&diskio.list_filename {
+                while @(cx16.r0)==' ' and cx16.r0>=&list_filename {
                     @(cx16.r0) = 0
                     cx16.r0--
                 }
@@ -108,7 +119,7 @@ io_error:
 
 io_error:
         cbm.CLRCHN()        ; restore default i/o devices
-        cbm.CLOSE(12)
+        cbm.CLOSE(READ_IO_CHANNEL)
         if okay
             return &list_filename
         return 0
@@ -126,7 +137,7 @@ io_error:
 
     sub list_filenames(uword pattern_ptr, uword filenames_buffer, uword filenames_buf_size) -> ubyte {
         ; -- fill the provided buffer with the names of the files on the disk (until buffer is full).
-        ;    Files in the buffer are separeted by a 0 byte. You can provide an optional pattern to match against.
+        ;    Files in the buffer are separated by a 0 byte. You can provide an optional pattern to match against.
         ;    After the last filename one additional 0 byte is placed to indicate the end of the list.
         ;    Returns number of files (it skips 'dir' entries i.e. subdirectories).
         ;    Also sets carry on exit: Carry clear = all files returned, Carry set = directory has more files that didn't fit in the buffer.
@@ -152,7 +163,7 @@ io_error:
         return files_found
     }
 
-    ; ----- iterative file lister functions (uses io channel 12) -----
+    ; ----- iterative file lister functions (uses the read io channel) -----
 
     sub lf_start_list(uword pattern_ptr) -> bool {
         ; -- start an iterative file listing with optional pattern matching.
@@ -163,11 +174,11 @@ io_error:
         iteration_in_progress = true
 
         cbm.SETNAM(1, "$")
-        cbm.SETLFS(12, drivenumber, 0)
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, 0)
         void cbm.OPEN()          ; open 12,8,0,"$"
         if_cs
             goto io_error
-        void cbm.CHKIN(12)        ; use #12 as input channel
+        reset_read_channel()
         if_cs
             goto io_error
 
@@ -192,7 +203,7 @@ io_error:
             return false
 
         repeat {
-            void cbm.CHKIN(12)        ; use #12 as input channel again
+            reset_read_channel()        ; use the input io channel again
 
             uword nameptr = &list_filename
             ubyte blocks_lsb = cbm.CHRIN()
@@ -211,12 +222,12 @@ io_error:
 
             ; read the filename
             repeat {
-                ubyte char = cbm.CHRIN()
-                if char==0
+                ubyte character = cbm.CHRIN()
+                if character==0
                     break
-                if char=='\"'
+                if character=='\"'
                     break
-                @(nameptr) = char
+                @(nameptr) = character
                 nameptr++
             }
 
@@ -253,13 +264,13 @@ close_end:
         ; -- end an iterative file listing session (close channels).
         if iteration_in_progress {
             cbm.CLRCHN()
-            cbm.CLOSE(12)
+            cbm.CLOSE(READ_IO_CHANNEL)
             iteration_in_progress = false
         }
     }
 
 
-    ; ----- iterative file loader functions (uses io channel 12) -----
+    ; ----- iterative file loader functions (uses the input io channel) -----
 
     sub f_open(uword filenameptr) -> bool {
         ; -- open a file for iterative reading with f_read
@@ -267,18 +278,18 @@ close_end:
         f_close()
 
         cbm.SETNAM(string.length(filenameptr), filenameptr)
-        cbm.SETLFS(12, drivenumber, 12)     ; note: has to be 12,x,12 because otherwise f_seek doesn't work
+        cbm.SETLFS(READ_IO_CHANNEL, drivenumber, READ_IO_CHANNEL)     ; note: has to be Channel,x,Channel because otherwise f_seek doesn't work
         void cbm.OPEN()          ; open 12,8,12,"filename"
         if_cc {
             if cbm.READST()==0 {
                 iteration_in_progress = true
-                void cbm.CHKIN(12)          ; use #12 as input channel
+                reset_read_channel()
                 if_cc {
                     void cbm.CHRIN()        ; read first byte to test for file not found
                     if not cbm.READST() {
-                        cbm.CLOSE(12)           ; close file because we already consumed first byte
+                        cbm.CLOSE(READ_IO_CHANNEL)           ; close file because we already consumed first byte
                         void cbm.OPEN()         ; re-open the file
-                        void cbm.CHKIN(12)
+                        cbm.CLRCHN()            ; reset default i/o channels
                         return true
                     }
                 }
@@ -294,6 +305,7 @@ close_end:
         if not iteration_in_progress or not num_bytes
             return 0
 
+        reset_read_channel()
         list_blocks = 0     ; we reuse this variable for the total number of bytes read
 
         %asm {{
@@ -328,6 +340,7 @@ m_in_buffer     sta  $ffff
         if not iteration_in_progress
             return 0
 
+        reset_read_channel()
         uword total_read = 0
         while not cbm.READST() {
             cx16.r0 = f_read(bufferpointer, 256)
@@ -346,8 +359,7 @@ m_in_buffer     sta  $ffff
         %asm {{
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
-            ldx  #12
-            jsr  cbm.CHKIN              ; use channel 12 again for input
+            jsr  reset_read_channel
             ldy  #0
 _loop       jsr  cbm.CHRIN
             sta  (P8ZP_SCRATCH_W1),y
@@ -368,25 +380,23 @@ _end        rts
         ; -- end an iterative file loading session (close channels).
         if iteration_in_progress {
             cbm.CLRCHN()
-            cbm.CLOSE(12)
+            cbm.CLOSE(READ_IO_CHANNEL)
             iteration_in_progress = false
         }
     }
 
 
-    ; ----- iterative file writing functions (uses io channel 13) -----
+    ; ----- iterative file writing functions (uses write io channel) -----
 
     sub f_open_w(uword filenameptr) -> bool {
         ; -- open a file for iterative writing with f_write
         f_close_w()
 
         cbm.SETNAM(string.length(filenameptr), filenameptr)
-        cbm.SETLFS(13, drivenumber, 1)
+        cbm.SETLFS(WRITE_IO_CHANNEL, drivenumber, 1)
         void cbm.OPEN()             ; open 13,8,1,"filename"
-        if_cc {
-            cbm.CHKOUT(13)          ; use #13 as output channel
+        if_cc
             return not cbm.READST()
-        }
         f_close_w()
         return false
     }
@@ -394,7 +404,7 @@ _end        rts
     sub f_write(uword bufferpointer, uword num_bytes) -> bool {
         ; -- write the given number of bytes to the currently open file
         if num_bytes!=0 {
-            cbm.CHKOUT(13)        ; use #13 as output channel again
+            reset_write_channel()
             repeat num_bytes {
                 cbm.CHROUT(@(bufferpointer))
                 bufferpointer++
@@ -407,7 +417,7 @@ _end        rts
     sub f_close_w() {
         ; -- end an iterative file writing session (close channels).
         cbm.CLRCHN()
-        cbm.CLOSE(13)
+        cbm.CLOSE(WRITE_IO_CHANNEL)
     }
 
 
@@ -444,16 +454,16 @@ io_error:
         goto done
     }
 
-    sub save(uword filenameptr, uword address, uword size) -> bool {
+    sub save(uword filenameptr, uword start_address, uword savesize) -> bool {
         cbm.SETNAM(string.length(filenameptr), filenameptr)
         cbm.SETLFS(1, drivenumber, 0)
-        uword @shared end_address = address + size
+        uword @shared end_address = start_address + savesize
         cx16.r0L = 0
 
         %asm {{
-            lda  address
+            lda  start_address
             sta  P8ZP_SCRATCH_W1
-            lda  address+1
+            lda  start_address+1
             sta  P8ZP_SCRATCH_W1+1
             lda  #<P8ZP_SCRATCH_W1
             ldx  end_address
@@ -466,9 +476,6 @@ io_error:
         if_cc
             cx16.r0L = cbm.READST()==0
 
-        cbm.CLRCHN()
-        cbm.CLOSE(1)
-
         return cx16.r0L
     }
 
@@ -496,24 +503,22 @@ io_error:
 +
         }}
 
-        cbm.CLRCHN()
-        cbm.CLOSE(1)
         return cx16.r1
     }
 
     ; Identical to load(), but DOES INCLUDE the first 2 bytes in the file.
     ; No program header is assumed in the file. Everything is loaded.
     ; See comments on load() for more details.
-    sub load_raw(uword filenameptr, uword address) -> uword {
+    sub load_raw(uword filenameptr, uword start_address) -> uword {
         ; read the 2 header bytes separately to skip them
         if not f_open(filenameptr)
             return 0
-        cx16.r1 = f_read(address, 2)
+        cx16.r1 = f_read(start_address, 2)
         f_close()
         if cx16.r1!=2
             return 0
-        address += 2
-        return load(filenameptr, address)
+        start_address += 2
+        return load(filenameptr, start_address)
     }
 
     sub delete(uword filenameptr) {

compiler/res/prog8lib/math.asm

@@ -4,78 +4,183 @@
 ;  some more interesting routines can be found here:
 ;	http://6502org.wikidot.com/software-math
 ;	http://codebase64.org/doku.php?id=base:6502_6510_maths
-;
+;       https://github.com/TobyLobster/multiply_test
+;       https://github.com/TobyLobster/sqrt_test
 
 
 multiply_bytes	.proc
 	; -- multiply 2 bytes A and Y, result as byte in A  (signed or unsigned)
-		sta  P8ZP_SCRATCH_B1         ; num1
-		sty  P8ZP_SCRATCH_REG        ; num2
-		lda  #0
-		beq  _enterloop
-_doAdd		clc
-		adc  P8ZP_SCRATCH_B1
-_loop		asl  P8ZP_SCRATCH_B1
-_enterloop	lsr  P8ZP_SCRATCH_REG
-		bcs  _doAdd
-		bne  _loop
-		rts
-		.pend
+	; https://github.com/TobyLobster/multiply_test/blob/main/tests/mult29.a
 
+_multiplicand    = P8ZP_SCRATCH_B1
+_multiplier      = P8ZP_SCRATCH_REG
 
-multiply_bytes_into_word	.proc
-	; -- multiply 2 bytes A and Y, result as word in A/Y (unsigned)
-		sta  P8ZP_SCRATCH_B1
-		sty  P8ZP_SCRATCH_REG
-		lda  #0
-		ldx  #8
-		lsr  P8ZP_SCRATCH_B1
--		bcc  +
-		clc
-		adc  P8ZP_SCRATCH_REG
-+		ror  a
-		ror  P8ZP_SCRATCH_B1
-		dex
-		bne  -
-		tay
-		lda  P8ZP_SCRATCH_B1
-		rts
+    sty  _multiplicand
+    lsr  a
+    sta  _multiplier
+    lda  #0
+    ldx  #2
+-
+    bcc  +
+    clc
+    adc  _multiplicand
++
+    ror  a
+    ror  _multiplier
+    bcc  +
+    clc
+    adc  _multiplicand
++
+    ror  a
+    ror  _multiplier
+
+    bcc  +
+    clc
+    adc  _multiplicand
++
+    ror  a
+    ror  _multiplier
+    bcc  +
+    clc
+    adc  _multiplicand
++
+    ror  a
+    ror  _multiplier
+    dex
+    bne  -
+    ; tay       ; if you want 16 bits result in AY, enable this again
+    lda  _multiplier
+    rts
 		.pend
 
 
 multiply_words	.proc
 	; -- multiply two 16-bit words into a 32-bit result  (signed and unsigned)
-	;      input: A/Y = first 16-bit number, P8ZP_SCRATCH_W1 in ZP = second 16-bit number
-	;      output: multiply_words.result  4-bytes/32-bits product, LSB order (low-to-high)  low 16 bits also in AY.
+	;      input: A/Y = first 16-bit number, multiply_words.multiplier = second 16-bit number
+	;      output: multiply_words.result, 4-bytes/32-bits product, LSB order (low-to-high)  low 16 bits also in AY.
 
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
+	; NOTE: the result (which includes the multiplier parameter on entry) is a 4-byte array.
+	;       this routine could be faster if we could stick that into zeropage,
+	;       but there currently is no way to use 4 consecutive bytes in ZP (without disabling irq and saving/restoring them)...
 
-mult16		lda  #0
-		sta  result+2	; clear upper bits of product
-		sta  result+3
-		ldx  #16			; for all 16 bits...
--	 	lsr  P8ZP_SCRATCH_W1+1	; divide multiplier by 2
-		ror  P8ZP_SCRATCH_W1
-		bcc  +
-		lda  result+2	; get upper half of product and add multiplicand
-		clc
-		adc  P8ZP_SCRATCH_W2
-		sta  result+2
-		lda  result+3
-		adc  P8ZP_SCRATCH_W2+1
-+ 		ror  a				; rotate partial product
-		sta  result+3
-		ror  result+2
-		ror  result+1
-		ror  result
-		dex
-		bne  -
-		lda  result
-		ldy  result+1
-		rts
+; mult62.a
+; from: https://github.com/TobyLobster/multiply_test/blob/main/tests/mult62.a
+; based on Dr Jefyll, http://forum.6502.org/viewtopic.php?f=9&t=689&start=0#p19958
+; - adjusted to use fixed zero page addresses
+; - removed 'decrement to avoid clc' as this is slower on average
+; - rearranged memory use to remove final memory copy and give LSB first order to result
+; - removed temp zp storage bytes
+; - unrolled the outer loop
+; - unrolled the two inner loops once
+;
+; 16 bit x 16 bit unsigned multiply, 32 bit result
+; Average cycles: ~442 ?
+; 93 bytes
+
+_multiplicand    = P8ZP_SCRATCH_W2   ; 2 bytes
+multiplier      = result
+
+; 16 bit x 16 bit unsigned multiply, 32 bit result
+;
+; On Entry:
+;   (multiplier, multiplier+1): two byte multiplier, four bytes needed for result
+;   (multiplicand, multiplicand+1): two byte multiplicand
+; On Exit:
+;   (result, result+1, result+2, result+3): product
+
+    sta  _multiplicand
+    sty  _multiplicand+1
+
+    lda  #0              ;
+    sta  result+2        ; 16 bits of zero in A, result+2
+                        ;  Note:    First 8 shifts are  A -> result+2 -> result
+                        ;           Final 8 shifts are  A -> result+2 -> result+1
+
+    ; --- 1st byte ---
+    ldy  #4              ; count for inner loop
+    lsr  result
+
+    ; inner loop (8 times)
+_inner_loop
+    ; first time
+    bcc +
+    tax                 ; retain A
+    lda  result+2
+    clc
+    adc  _multiplicand
+    sta  result+2
+    txa                 ; recall A
+    adc  _multiplicand+1
+
++
+    ror  a                ; shift
+    ror  result+2
+    ror  result
+
+    ; second time
+    bcc +
+    tax                 ; retain A
+    lda  result+2
+    clc
+    adc  _multiplicand
+    sta  result+2
+    txa                 ; recall A
+    adc  _multiplicand+1
+
++
+    ror  a                 ; shift
+    ror  result+2
+    ror  result
+
+    dey
+    bne  _inner_loop      ; go back for 1 more shift?
+
+    ; --- 2nd byte ---
+    ldy  #4              ; count for inner loop
+    lsr  result+1
+
+    ; inner loop (8 times)
+_inner_loop2
+    ; first time
+    bcc  +
+    tax                 ; retain A
+    lda  result+2
+    clc
+    adc  _multiplicand
+    sta  result+2
+    txa                 ; recall A
+    adc  _multiplicand+1
+
++
+    ror  a                ; shift
+    ror  result+2
+    ror  result+1
+
+    ; second time
+    bcc  +
+    tax                 ; retain A
+    lda  result+2
+    clc
+    adc  _multiplicand
+    sta  result+2
+    txa                 ; recall A
+    adc  _multiplicand+1
+
++
+    ror  a                ; shift
+    ror  result+2
+    ror  result+1
+    dey
+    bne  _inner_loop2     ; go back for 1 more shift?
+
+    sta  result+3        ; ms byte of hi-word of result
+
+    lda  result
+    ldy  result+1
+    rts
 
 result		.byte  0,0,0,0
+
 		.pend
 
 
@@ -749,74 +854,86 @@ mul_word_640	.proc
 ; support for bit shifting that is too large to be unrolled:
 
 lsr_byte_A	.proc
-		; -- lsr signed byte in A times the value in Y (>1)
+		; -- lsr signed byte in A times the value in Y
+		cpy  #0
+		beq  +
 		cmp  #0
 		bpl  lsr_ubyte_A
 -       	sec
 		ror  a
 		dey
 		bne  -
-		rts
++		rts
 		.pend
 
 lsr_ubyte_A	.proc
-		; -- lsr unsigned byte in A times the value in Y (>1)
+		; -- lsr unsigned byte in A times the value in Y
+		cpy  #0
+		beq  +
 -		lsr  a
 		dey
 		bne  -
-		rts
++		rts
 		.pend
 
 asl_byte_A      .proc
-		; -- asl any byte in A times the value in Y (>1)
+		; -- asl any byte in A times the value in Y
+		cpy  #0
+		beq  +
 -		asl  a
 		dey
 		bne  -
-		rts
++		rts
 		.pend
 
 
 lsr_word_AY     .proc
-		; -- lsr signed word in AY times the value in X (>1)
+		; -- lsr signed word in AY times the value in X
+		cpx  #0
+		beq  +
 		cpy  #0
 		bpl  lsr_uword_AY
 		sty  P8ZP_SCRATCH_B1
-_negative       sec
+-		sec
 		ror  P8ZP_SCRATCH_B1
 		ror  a
 		dex
-		bne  _negative
+		bne  -
 		ldy  P8ZP_SCRATCH_B1
-		rts
++		rts
 		.pend
 
 lsr_uword_AY    .proc
-		; -- lsr unsigned word in AY times the value in X (>1)
+		; -- lsr unsigned word in AY times the value in X
+		cpx  #0
+		beq  +
 		sty  P8ZP_SCRATCH_B1
 -		lsr  P8ZP_SCRATCH_B1
 		ror  a
 		dex
 		bne  -
 		ldy  P8ZP_SCRATCH_B1
-		rts
++		rts
 		.pend
 
 asl_word_AY     .proc
-		; -- asl any word in AY times the value in X (>1)
+		; -- asl any word in AY times the value in X
+		cpx  #0
+		beq  +
 		sty  P8ZP_SCRATCH_B1
 -               asl  a
 		rol  P8ZP_SCRATCH_B1
 		dex
 		bne  -
 		ldy  P8ZP_SCRATCH_B1
-		rts
++		rts
 		.pend
 
 
 square          .proc
-; -- calculate square root of signed word in AY, result in AY
-; routine by Lee Davsion, source: http://6502.org/source/integers/square.htm
-; using this routine is about twice as fast as doing a regular multiplication.
+; -- calculate square of signed word (actually -255..255) in AY, result in AY
+; routine by Lee Davison, source: http://6502.org/source/integers/square.htm
+; using this routine is a lot faster as doing a regular multiplication (for words)
 ;
 ; Calculates the 16 bit unsigned integer square of the signed 16 bit integer in
 ; Numberl/Numberh.  The result is always in the range 0 to 65025 and is held in

compiler/res/prog8lib/math.p8

@@ -1,4 +1,6 @@
 ; Internal Math library routines - always included by the compiler
+; note: some functions you might expect here are builtin functions,
+;       such as abs, sqrt, clamp, min, max for example.
 
 math {
     %option no_symbol_prefixing
@@ -98,6 +100,62 @@ _sinecosR8	.char  trunc(127.0 * sin(range(180+45) * rad(360.0/180.0)))
         }}
     }
 
+    asmsub log2(ubyte value @A) -> ubyte @Y {
+        %asm {{
+            sta  P8ZP_SCRATCH_B1
+            lda  #$80
+            ldy  #7
+-           bit  P8ZP_SCRATCH_B1
+            beq  +
+            rts
++           dey
+            bne  +
+            rts
++           lsr  a
+            bne  -
+        }}
+    }
+
+    asmsub log2w(uword value @AY) -> ubyte @Y {
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  #<$8000
+            sta  cx16.r0
+            lda  #>$8000
+            sta  cx16.r0+1
+            ldy  #15
+-           lda  P8ZP_SCRATCH_W1
+            and  cx16.r0
+            sta  P8ZP_SCRATCH_B1
+            lda  P8ZP_SCRATCH_W1+1
+            and  cx16.r0+1
+            ora  P8ZP_SCRATCH_B1
+            beq  +
+            rts
++           dey
+            bne  +
+            rts
++           lsr  cx16.r0+1
+            ror  cx16.r0
+            jmp  -
+        }}
+    }
+
+    asmsub mul16_last_upper() -> uword @AY {
+        ; This routine peeks into the internal 32 bits multiplication result buffer of the
+        ; 16*16 bits multiplication routine, to fetch the upper 16 bits of the last calculation.
+        ; Notes:
+        ;   - to avoid interference it's best to fetch and store this value immediately after the multiplication expression.
+        ;     for instance, simply printing a number may already result in new multiplication calls being performed
+        ;   - not all multiplications in the source code result in an actual multiplication call:
+        ;     some simpler multiplications will be optimized away into faster routines. These will not set the upper 16 bits at all!
+        %asm {{
+            lda  multiply_words.result+2
+            ldy  multiply_words.result+3
+            rts
+        }}
+    }
 
 sub direction_sc(byte x1, byte y1, byte x2, byte y2) -> ubyte {
     ; From a pair of signed coordinates around the origin, calculate discrete direction between 0 and 23 into A.
@@ -389,4 +447,41 @@ log2_tab
     }}
 }
 
+    asmsub diff(ubyte v1 @A, ubyte v2 @Y) -> ubyte @A {
+        ; -- returns the (absolute) difference, or distance, between the two bytes
+        %asm {{
+            sty  P8ZP_SCRATCH_REG
+            sec
+            sbc  P8ZP_SCRATCH_REG
+            bcs  +
+            eor  #255
+            inc  a
++           rts
+        }}
+    }
+
+    asmsub diffw(uword w1 @R0, uword w2 @AY) -> uword @AY {
+        ; -- returns the (absolute) difference, or distance, between the two words
+        %asm {{
+            sec
+            sbc  cx16.r0L
+            sta  cx16.r0L
+            tya
+            sbc  cx16.r0H
+            sta  cx16.r0H
+            bcs  +
+            eor  #255
+            sta  cx16.r0H
+            lda  cx16.r0L
+            eor  #255
+            inc  a
+            sta  cx16.r0L
+            bne  +
+            inc  cx16.r0H
++           lda  cx16.r0L
+            ldy  cx16.r0H
+            rts
+        }}
+    }
+
 }

compiler/res/prog8lib/pet32/syslib.p8

@@ -0,0 +1,460 @@
+; Prog8 definitions for the Commodore PET
+; Including memory registers, I/O registers, Basic and Kernal subroutines.
+; see: https://www.pagetable.com/?p=926  ,  http://www.zimmers.net/cbmpics/cbm/PETx/petmem.txt
+
+cbm {
+    ; Commodore (CBM) common variables, vectors and kernal routines
+    %option no_symbol_prefixing
+
+        &ubyte  TIME_HI         = $8d       ; software jiffy clock, hi byte
+        &ubyte  TIME_MID        = $8e       ;  .. mid byte
+        &ubyte  TIME_LO         = $8f       ;    .. lo byte. Updated by IRQ every 1/60 sec
+        &ubyte  STATUS          = $96       ; kernal status variable for I/O
+
+        &uword  CINV            = $0090     ; IRQ vector (in ram)
+        &uword  CBINV           = $0092     ; BRK vector (in ram)
+        &uword  NMINV           = $0094     ; 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    = $8000     ; to have this as an array[40*25] the compiler would have to support array size > 255
+
+
+romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; define an input channel
+romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; define an output channel
+romsub $FFCC = CLRCHN() clobbers(A,X)                           ; restore default devices
+romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A              ; input a character (for keyboard, read a whole line from the screen) A=byte read.
+romsub $FFD2 = CHROUT(ubyte character @ A)                           ; output a character
+romsub $FFE1 = STOP() clobbers(X) -> bool @ Pz, ubyte @ A       ; check the STOP key (and some others in A)
+romsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; get a character
+romsub $FFE7 = CLALL() clobbers(A,X)                            ; close all files
+romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+
+asmsub STOP2() clobbers(X) -> 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 {{
+        jsr  cbm.STOP
+        beq  +
+        lda  #0
+        rts
++       lda  #1
+        rts
+    }}
+}
+
+asmsub SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y) {
+    ; PET stub to set the software clock
+    %asm {{
+        sty  TIME_HI
+        stx  TIME_MID
+        sta  TIME_LO
+        rts
+    }}
+}
+
+asmsub RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y {
+    ; PET stub to read the software clock (A=lo,X=mid,Y=high)
+    %asm {{
+        ldy  TIME_HI
+        ldx  TIME_MID
+        lda  TIME_LO
+        rts
+    }}
+}
+
+asmsub RDTIM16() clobbers(X) -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits in AY for convenience
+    %asm {{
+        lda  TIME_LO
+        ldy  TIME_MID
+        rts
+    }}
+}
+
+
+}
+
+sys {
+    ; ------- lowlevel system routines --------
+    %option no_symbol_prefixing
+
+    const ubyte target = 32         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 32=PET
+
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    ; Uppercase charset is activated.
+    %asm {{
+        sei
+        cld
+        lda  #142
+        jsr  cbm.CHROUT     ; uppercase
+        lda  #147
+        jsr  cbm.CHROUT     ; clear screen
+        clc
+        clv
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        rts     ; no phase 2 steps on the PET
+    }}
+}
+
+asmsub  cleanup_at_exit() {
+    ; executed when the main subroutine does rts
+    %asm {{
+        rts
+    }}
+}
+
+    asmsub  reset_system()  {
+        ; Soft-reset the system back to initial power-on Basic prompt.
+        %asm {{
+            sei
+            jmp  (cbm.RESET_VEC)
+        }}
+    }
+
+    asmsub waitvsync() clobbers(A) {
+        ; --- busy wait till the next vsync has occurred (approximately), without depending on custom irq handling.
+        ;     Note: on PET this simply waits until the next jiffy clock update, I don't know if a true vsync is possible there
+        %asm {{
+            lda  #1
+            ldy  #0
+            jmp  wait
+        }}
+    }
+
+    asmsub wait(uword jiffies @AY) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds) (N or N+1)
+        ;     note: the system irq handler has to be active for this to work as it depends on the system jiffy clock
+        %asm {{
+            stx  P8ZP_SCRATCH_B1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+
+_loop       lda  P8ZP_SCRATCH_W1
+            ora  P8ZP_SCRATCH_W1+1
+            bne  +
+            ldx  P8ZP_SCRATCH_B1
+            rts
+
++           lda  cbm.TIME_LO
+            sta  P8ZP_SCRATCH_B1
+-           lda  cbm.TIME_LO
+            cmp  P8ZP_SCRATCH_B1
+            beq  -
+
+            lda  P8ZP_SCRATCH_W1
+            bne  +
+            dec  P8ZP_SCRATCH_W1+1
++           dec  P8ZP_SCRATCH_W1
+            jmp  _loop
+        }}
+    }
+
+    asmsub internal_stringcopy(uword source @R0, uword target @AY) clobbers (A,Y) {
+        ; Called when the compiler wants to assign a string value to another string.
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcpy
+        }}
+    }
+
+    asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        ; note: only works for NON-OVERLAPPING memory regions!
+        ; note: can't be inlined because is called from asm as well
+        %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 irqsafe_set_irqd() {
+        %asm {{
+        php
+        sei
+        }}
+    }
+
+    inline asmsub irqsafe_clear_irqd() {
+        %asm {{
+        plp
+        }}
+    }
+
+    sub disable_caseswitch() {
+        ; PET doesn't have a key to swap case, so no-op
+    }
+
+    sub enable_caseswitch() {
+        ; PET doesn't have a key to swap case, so no-op
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            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 {
+    %option no_symbol_prefixing
+
+    ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
+    ; they are simulated on the PET as well but their location in memory is different
+    ; (because there's no room for them in the zeropage)
+    ; we select the top page of RAM (assume 32Kb)
+    &uword r0  = $7fe0
+    &uword r1  = $7fe2
+    &uword r2  = $7fe4
+    &uword r3  = $7fe6
+    &uword r4  = $7fe8
+    &uword r5  = $7fea
+    &uword r6  = $7fec
+    &uword r7  = $7fee
+    &uword r8  = $7ff0
+    &uword r9  = $7ff2
+    &uword r10 = $7ff4
+    &uword r11 = $7ff6
+    &uword r12 = $7ff8
+    &uword r13 = $7ffa
+    &uword r14 = $7ffc
+    &uword r15 = $7ffe
+
+    &word r0s  = $7fe0
+    &word r1s  = $7fe2
+    &word r2s  = $7fe4
+    &word r3s  = $7fe6
+    &word r4s  = $7fe8
+    &word r5s  = $7fea
+    &word r6s  = $7fec
+    &word r7s  = $7fee
+    &word r8s  = $7ff0
+    &word r9s  = $7ff2
+    &word r10s = $7ff4
+    &word r11s = $7ff6
+    &word r12s = $7ff8
+    &word r13s = $7ffa
+    &word r14s = $7ffc
+    &word r15s = $7ffe
+
+    &ubyte r0L  = $7fe0
+    &ubyte r1L  = $7fe2
+    &ubyte r2L  = $7fe4
+    &ubyte r3L  = $7fe6
+    &ubyte r4L  = $7fe8
+    &ubyte r5L  = $7fea
+    &ubyte r6L  = $7fec
+    &ubyte r7L  = $7fee
+    &ubyte r8L  = $7ff0
+    &ubyte r9L  = $7ff2
+    &ubyte r10L = $7ff4
+    &ubyte r11L = $7ff6
+    &ubyte r12L = $7ff8
+    &ubyte r13L = $7ffa
+    &ubyte r14L = $7ffc
+    &ubyte r15L = $7ffe
+
+    &ubyte r0H  = $7fe1
+    &ubyte r1H  = $7fe3
+    &ubyte r2H  = $7fe5
+    &ubyte r3H  = $7fe7
+    &ubyte r4H  = $7fe9
+    &ubyte r5H  = $7feb
+    &ubyte r6H  = $7fed
+    &ubyte r7H  = $7fef
+    &ubyte r8H  = $7ff1
+    &ubyte r9H  = $7ff3
+    &ubyte r10H = $7ff5
+    &ubyte r11H = $7ff7
+    &ubyte r12H = $7ff9
+    &ubyte r13H = $7ffb
+    &ubyte r14H = $7ffd
+    &ubyte r15H = $7fff
+
+    &byte r0sL  = $7fe0
+    &byte r1sL  = $7fe2
+    &byte r2sL  = $7fe4
+    &byte r3sL  = $7fe6
+    &byte r4sL  = $7fe8
+    &byte r5sL  = $7fea
+    &byte r6sL  = $7fec
+    &byte r7sL  = $7fee
+    &byte r8sL  = $7ff0
+    &byte r9sL  = $7ff2
+    &byte r10sL = $7ff4
+    &byte r11sL = $7ff6
+    &byte r12sL = $7ff8
+    &byte r13sL = $7ffa
+    &byte r14sL = $7ffc
+    &byte r15sL = $7ffe
+
+    &byte r0sH  = $7fe1
+    &byte r1sH  = $7fe3
+    &byte r2sH  = $7fe5
+    &byte r3sH  = $7fe7
+    &byte r4sH  = $7fe9
+    &byte r5sH  = $7feb
+    &byte r6sH  = $7fed
+    &byte r7sH  = $7fef
+    &byte r8sH  = $7ff1
+    &byte r9sH  = $7ff3
+    &byte r10sH = $7ff5
+    &byte r11sH = $7ff7
+    &byte r12sH = $7ff9
+    &byte r13sH = $7ffb
+    &byte r14sH = $7ffd
+    &byte r15sH = $7fff
+
+    asmsub save_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  cx16.r0,y
+            sta  _cx16_vreg_storage,y
+            dey
+            bpl  -
+            rts
+
+    _cx16_vreg_storage
+            .word 0,0,0,0,0,0,0,0
+            .word 0,0,0,0,0,0,0,0
+        }}
+    }
+
+    asmsub restore_virtual_registers() clobbers(A,Y) {
+        %asm {{
+            ldy  #31
+    -       lda  save_virtual_registers._cx16_vreg_storage,y
+            sta  cx16.r0,y
+            dey
+            bpl  -
+            rts
+        }}
+    }
+}

compiler/res/prog8lib/pet32/textio.p8

@@ -0,0 +1,505 @@
+; Prog8 definitions for the Text I/O and Screen routines for the Commodore PET
+; All routines work with Screencode character encoding, except `print`, `chrout` and `input_chars`,
+; these work with PETSCII encoding instead.
+
+%import syslib
+%import conv
+
+
+txt {
+    %option no_symbol_prefixing
+
+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 get_column() -> ubyte @Y {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+asmsub row(ubyte rownum @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given row (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  cbm.PLOT
+        tax
+        clc
+        jmp  cbm.PLOT
+    }}
+}
+
+asmsub get_row() -> ubyte @X {
+    %asm {{
+        sec
+        jmp cbm.PLOT
+    }}
+}
+
+sub get_cursor(uword colptr, uword rowptr) {
+    %asm {{
+        sec
+        jsr cbm.PLOT
+        tya
+        ldy #$00
+        sta (colptr),y
+        txa
+        sta (rowptr),y
+    }}
+}
+
+
+asmsub  fill_screen (ubyte character @ A, ubyte color @ Y) clobbers(A)  {
+	; ---- fill the character screen with the given fill character. color is ignored on PET
+	%asm {{
+		jmp  clear_screenchars
+        }}
+
+}
+
+asmsub  clear_screenchars (ubyte character @ A) clobbers(Y)  {
+	; ---- clear the character screen with the given fill character
+	;      (assumes screen matrix is at the default address)
+	%asm {{
+		ldy  #250
+-		sta  cbm.Screen+250*0-1,y
+		sta  cbm.Screen+250*1-1,y
+		sta  cbm.Screen+250*2-1,y
+		sta  cbm.Screen+250*3-1,y
+		dey
+		bne  -
+		rts
+        }}
+}
+
+sub  clear_screencolors (ubyte color)  {
+    ; --- dummy function on PET
+}
+
+sub color (ubyte txtcol) {
+    ; --- dummy function on PET
+}
+
+
+sub lowercase() {
+    txt.chrout(14)
+}
+
+sub uppercase() {
+    txt.chrout(142)
+}
+
+asmsub  scroll_left  () clobbers(A, X, Y)  {
+	; ---- scroll the whole screen 1 character to the left
+	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
+
+	%asm {{
+		ldx  #0
+		ldy  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  cbm.Screen + 40*row + 1,x
+            sta  cbm.Screen + 40*row + 0,x
+        .next
+		inx
+		dey
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_right  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character to the right
+	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
+	%asm {{
+		ldx  #38
+-
+        .for row=0, row<=24, row+=1
+            lda  cbm.Screen + 40*row + 0,x
+            sta  cbm.Screen + 40*row + 1,x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_up  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character up
+	;      contents of the bottom row are unchanged, you should refill/clear this yourself
+	%asm {{
+		ldx #39
+-
+        .for row=1, row<=24, row+=1
+            lda  cbm.Screen + 40*row,x
+            sta  cbm.Screen + 40*(row-1),x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+asmsub  scroll_down  () clobbers(A,X)  {
+	; ---- scroll the whole screen 1 character down
+	;      contents of the top row are unchanged, you should refill/clear this yourself
+	%asm {{
+		ldx #39
+-
+        .for row=23, row>=0, row-=1
+            lda  cbm.Screen + 40*row,x
+            sta  cbm.Screen + 40*(row+1),x
+        .next
+		dex
+		bpl  -
+
+		rts
+	}}
+}
+
+romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
+
+asmsub  print (str text @ AY) clobbers(A,Y)  {
+	; ---- print null terminated string, in PETSCII encoding, from A/Y
+	; note: the compiler contains an optimization that will replace
+	;       a call to this subroutine with a string argument of just one char,
+	;       by just one call to cbm.CHROUT of that single char.
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		sty  P8ZP_SCRATCH_REG
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_B1),y
+		beq  +
+		jsr  cbm.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_ub0  (ubyte value @ A) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in decimal form, with left padding 0s (3 positions total)
+	%asm {{
+		jsr  conv.ubyte2decimal
+		pha
+		tya
+		jsr  cbm.CHROUT
+		pla
+		jsr  cbm.CHROUT
+		txa
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_ub  (ubyte value @ A) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in decimal form, without left padding 0s
+	%asm {{
+		jsr  conv.ubyte2decimal
+_print_byte_digits
+		pha
+		cpy  #'0'
+		beq  +
+		tya
+		jsr  cbm.CHROUT
+		pla
+		jsr  cbm.CHROUT
+		jmp  _ones
++       pla
+        cmp  #'0'
+        beq  _ones
+        jsr  cbm.CHROUT
+_ones   txa
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_b  (byte value @ A) clobbers(A,X,Y)  {
+	; ---- print the byte in A in decimal form, without left padding 0s
+	%asm {{
+		pha
+		cmp  #0
+		bpl  +
+		lda  #'-'
+		jsr  cbm.CHROUT
++		pla
+		jsr  conv.byte2decimal
+		jmp  print_ub._print_byte_digits
+	}}
+}
+
+asmsub  print_ubhex  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in hex form (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		bcc  +
+		pha
+		lda  #'$'
+		jsr  cbm.CHROUT
+		pla
++		jsr  conv.ubyte2hex
+		jsr  cbm.CHROUT
+		tya
+		jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_ubbin  (ubyte value @ A, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the ubyte in A in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		sta  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'%'
+		jsr  cbm.CHROUT
++		ldy  #8
+-		lda  #'0'
+		asl  P8ZP_SCRATCH_B1
+		bcc  +
+		lda  #'1'
++		jsr  cbm.CHROUT
+		dey
+		bne  -
+		rts
+	}}
+}
+
+asmsub  print_uwbin  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in binary form (if Carry is set, a radix prefix '%' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubbin
+		pla
+		clc
+		jmp  print_ubbin
+	}}
+}
+
+asmsub  print_uwhex  (uword value @ AY, bool prefix @ Pc) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in hexadecimal form (4 digits)
+	;      (if Carry is set, a radix prefix '$' is printed as well)
+	%asm {{
+		pha
+		tya
+		jsr  print_ubhex
+		pla
+		clc
+		jmp  print_ubhex
+	}}
+}
+
+asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
+	%asm {{
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+        beq  +
+		jsr  cbm.CHROUT
+		iny
+		bne  -
++		rts
+	}}
+}
+
+asmsub  print_uw  (uword value @ AY) clobbers(A,X,Y)  {
+	; ---- print the uword in A/Y in decimal form, without left padding 0s
+	%asm {{
+		jsr  conv.uword2decimal
+		ldy  #0
+-		lda  conv.uword2decimal.decTenThousands,y
+		beq  _allzero
+		cmp  #'0'
+		bne  _gotdigit
+		iny
+		bne  -
+
+_gotdigit
+		jsr  cbm.CHROUT
+		iny
+		lda  conv.uword2decimal.decTenThousands,y
+		bne  _gotdigit
+		rts
+_allzero
+        lda  #'0'
+        jmp  cbm.CHROUT
+	}}
+}
+
+asmsub  print_w  (word value @ AY) clobbers(A,X,Y)  {
+	; ---- print the (signed) word in A/Y in decimal form, without left padding 0's
+	%asm {{
+		cpy  #0
+		bpl  +
+		pha
+		lda  #'-'
+		jsr  cbm.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, in PETSCII encoding.
+	;      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  cbm.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  cbm.Screen + 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
+	}}
+}
+
+sub  setclr  (ubyte col, ubyte row, ubyte color)  {
+    ; --- dummy function on PET
+}
+
+
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
+	; ---- set char at the given position on the screen. charcolor is ignored on PET
+	%asm {{
+  		lda  row
+		asl  a
+		tay
+		lda  setchr._screenrows+1,y
+		sta  _charmod+2
+		lda  setchr._screenrows,y
+		clc
+		adc  col
+		sta  _charmod+1
+		bcc  +
+		inc  _charmod+2
++		lda  character
+_charmod	sta  $ffff		; modified
+		rts
+	}}
+}
+
+asmsub  plot  (ubyte col @ Y, ubyte row @ X) {
+	%asm  {{
+	    jsr  home
+	    cpy  #0
+	    beq  +
+-	    lda  #17
+	    jsr  chrout
+	    dey
+	    bne  -
++	    cpx  #0
+	    beq  +
+-       lda  #29
+        jsr  chrout
+        dex
+        bne  -
++		rts
+	}}
+}
+
+asmsub width() clobbers(X,Y) -> ubyte @A {
+    ; -- returns the text screen width (number of columns)
+    %asm {{
+        lda  $d5
+        clc
+        adc  #1
+        rts
+    }}
+}
+
+asmsub height() clobbers(X, Y) -> ubyte @A {
+    ; -- returns the text screen height (number of rows)
+    %asm {{
+        lda  #25
+        rts
+    }}
+}
+
+asmsub waitkey() {
+    %asm {{
+-       jsr cbm.GETIN
+        beq -
+        rts
+    }}
+}
+}

compiler/res/prog8lib/prog8_funcs.asm

@@ -127,33 +127,42 @@ _possibly_zero	cmp  #0
 
 
 func_sqrt16_into_A	.proc
-		; integer square root from  http://6502org.wikidot.com/software-math-sqrt
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		lda  #0
-		sta  P8ZP_SCRATCH_B1
-		sta  P8ZP_SCRATCH_REG
-		ldx  #8
--		sec
-		lda  P8ZP_SCRATCH_W1+1
-		sbc  #$40
-		tay
-		lda  P8ZP_SCRATCH_REG
-		sbc  P8ZP_SCRATCH_B1
-		bcc  +
-		sty  P8ZP_SCRATCH_W1+1
-		sta  P8ZP_SCRATCH_REG
-+		rol  P8ZP_SCRATCH_B1
-		asl  P8ZP_SCRATCH_W1
-		rol  P8ZP_SCRATCH_W1+1
-		rol  P8ZP_SCRATCH_REG
-		asl  P8ZP_SCRATCH_W1
-		rol  P8ZP_SCRATCH_W1+1
-		rol  P8ZP_SCRATCH_REG
-		dex
-		bne  -
-		lda  P8ZP_SCRATCH_B1
-		rts
+		; integer square root
+		; http://6502org.wikidot.com/software-math-sqrt
+		; https://github.com/TobyLobster/sqrt_test/blob/main/sqrt/sqrt7.a
+		; Tweaked by TobyLobster and 0xC0DE to be smaller and faster
+_numl = P8ZP_SCRATCH_W1
+_numh = P8ZP_SCRATCH_W1+1
+_loop_counter = P8ZP_SCRATCH_REG
+_root = P8ZP_SCRATCH_B1
+            sta  _numl
+            sty  _numh
+            ldx  #$ff
+            stx  _loop_counter
+            inx
+            stx  _root
+            sec
+_loop       lda  _numh
+            sbc  #$40
+            tay
+            txa
+            sbc  _root
+            bcc  +
+            sty  _numh
+            bcs  ++
++           txa
++           rol  _root
+            asl  _numl
+            rol  _numh
+            rol  a
+            asl  _numl
+            rol  _numh
+            rol  a
+            tax
+            lsr  _loop_counter
+            bne  _loop
+            lda  _root
+            rts
 		.pend
 
 

compiler/res/prog8lib/string.p8

@@ -146,11 +146,11 @@ _startloop	dey
 		iny
 		bne  -
 _notfound	lda  #0
-                clc
+        clc
 		rts
-_found		tya
-                sec
-                rts
+_found	tya
+        sec
+        rts
         }}
     }
 
@@ -168,6 +168,31 @@ _found		tya
         }}
     }
 
+    asmsub append(uword target @R0, uword suffix @R1) clobbers(Y) -> ubyte @A {
+        ; Append the suffix string to the target. (make sure the buffer is large enough!)
+        ; Returns the length of the resulting string.
+        %asm {{
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            jsr  length
+            sty  P8ZP_SCRATCH_B1
+            tya
+            clc
+            adc  cx16.r0
+            sta  P8ZP_SCRATCH_W1
+            lda  cx16.r0+1
+            adc  #0
+            sta  P8ZP_SCRATCH_W1+1
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            jsr  prog8_lib.strcpy
+            tya
+            clc
+            adc  P8ZP_SCRATCH_B1
+            rts
+        }}
+    }
+
     asmsub compare(uword string1 @R0, uword string2 @AY) clobbers(Y) -> byte @A {
         ; Compares two strings for sorting.
         ; Returns -1 (255), 0 or 1 depending on wether string1 sorts before, equal or after string2.
@@ -225,7 +250,7 @@ _done       rts
         }}
     }
 
-    asmsub lowerchar(ubyte char @A) -> ubyte @A {
+    asmsub lowerchar(ubyte character @A) -> ubyte @A {
         %asm {{
             and  #$7f
             cmp  #97
@@ -237,7 +262,7 @@ _done       rts
         }}
     }
 
-    asmsub upperchar(ubyte char @A) -> ubyte @A {
+    asmsub upperchar(ubyte character @A) -> ubyte @A {
         %asm {{
             cmp  #65
             bcc  +
@@ -280,10 +305,10 @@ _done       rts
 ;
 ; see http://6502.org/source/strings/patmatch.htm
 
-str = P8ZP_SCRATCH_W1
+strptr = P8ZP_SCRATCH_W1
 
-	sta  str
-	sty  str+1
+	sta  strptr
+	sty  strptr+1
 	lda  cx16.r0
 	sta  modify_pattern1+1
 	sta  modify_pattern2+1
@@ -306,9 +331,9 @@ next    lda $ffff,x   ; look at next pattern character    MODIFIED
 	iny             ; no, let's look at the string
 	cmp #'?'     ; is the pattern caracter a ques?
 	bne reg         ; no, it's a regular character
-	lda (str),y     ; yes, so it will match anything
+	lda (strptr),y     ; yes, so it will match anything
 	beq fail        ;  except the end of string
-reg     cmp (str),y     ; are both characters the same?
+reg     cmp (strptr),y     ; are both characters the same?
 	bne fail        ; no, so no match
 	inx             ; yes, keep checking
 	cmp #0          ; are we at end of string?
@@ -330,7 +355,7 @@ stloop  txa             ; we first try to match with * = ""
 	tax
 	bcs found       ; we found a match, return with c=1
 	iny             ; no match yet, try to grow * string
-	lda (str),y     ; are we at the end of string?
+	lda (strptr),y     ; are we at the end of string?
 	bne stloop      ; not yet, add a character
 fail    clc             ; yes, no match found, return with c=0
 	rts

compiler/res/prog8lib/virtual/floats.p8

@@ -16,6 +16,15 @@ sub print_f(float value) {
     }}
 }
 
+sub parse_f(str value) -> float {
+    ; -- parse a string value of a number to float
+    %ir {{
+        loadm.w  r65535,floats.parse_f.value
+        syscall 45 (r65535.w): fr0.f
+        returnr.f fr0
+    }}
+}
+
 sub pow(float value, float power) -> float {
     %ir {{
         loadm.f fr0,floats.pow.value

compiler/res/prog8lib/virtual/math.p8

@@ -1,4 +1,6 @@
 ; Internal Math library routines - always included by the compiler
+; note: some functions you might expect here are builtin functions,
+;       such as abs, sqrt, clamp, min, max for example.
 
 math {
 
@@ -183,60 +185,111 @@ math {
         }}
     }
 
-
-sub direction(ubyte x1, ubyte y1, ubyte x2, ubyte y2) -> ubyte {
-    ; From a pair of positive coordinates, calculate discrete direction between 0 and 23 into A.
-    ; This adjusts the atan() result  so that the direction N is centered on the angle=N instead of having it as a boundary
-    ubyte angle = atan2(x1, y1, x2, y2) - 256/48
-    return 23-lsb(mkword(angle,0) / 2730)
-}
-
-sub direction_sc(byte x1, byte y1, byte x2, byte y2) -> ubyte {
-    ; From a pair of signed coordinates around the origin, calculate discrete direction between 0 and 23 into A.
-    ; shift the points into the positive quadrant
-    ubyte px1
-    ubyte py1
-    ubyte px2
-    ubyte py2
-    if x1<0 or x2<0 {
-        px1 = x1 as ubyte + 128
-        px2 = x2 as ubyte + 128
-    } else {
-        px1 = x1 as ubyte
-        px2 = x2 as ubyte
-    }
-    if y1<0 or y2<0 {
-        py1 = y1 as ubyte + 128
-        py2 = y2 as ubyte + 128
-    } else {
-        py1 = y1 as ubyte
-        py2 = y2 as ubyte
+    sub log2(ubyte value) -> ubyte {
+        ubyte result = 7
+        ubyte compare = $80
+        repeat {
+            if value&compare
+                return result
+            result--
+            if_z
+                return 0
+            compare >>= 1
+        }
     }
 
-    return direction(px1, py1, px2, py2)
-}
-
-sub direction_qd(ubyte quadrant, ubyte xdelta, ubyte ydelta) -> ubyte {
-    ; From a pair of X/Y deltas (both >=0), and quadrant 0-3, calculate discrete direction between 0 and 23.
-    when quadrant {
-        3 -> return direction(0, 0, xdelta, ydelta)
-        2 -> return direction(xdelta, 0, 0, ydelta)
-        1 -> return direction(0, ydelta, xdelta, 0)
-        else -> return direction(xdelta, ydelta, 0, 0)
+    sub log2w(uword value) -> ubyte {
+        ubyte result = 15
+        uword compare = $8000
+        repeat {
+            if value&compare
+                return result
+            result--
+            if_z
+                return 0
+            compare >>= 1
+        }
     }
-}
 
-sub atan2(ubyte x1, ubyte y1, ubyte x2, ubyte y2) -> ubyte {
-    ;; Calculate the angle, in a 256-degree circle, between two points into A.
-    ;; The points (x1, y1) and (x2, y2) have to use *unsigned coordinates only* from the positive quadrant in the carthesian plane!
-    %ir {{
-        loadm.b r65532,math.atan2.x1
-        loadm.b r65533,math.atan2.y1
-        loadm.b r65534,math.atan2.x2
-        loadm.b r65535,math.atan2.y2
-        syscall 44 (r65532.b, r65533.b, r65534.b, r65535.b): r0.b
-        returnr.b r0
-    }}
-}
+    sub direction(ubyte x1, ubyte y1, ubyte x2, ubyte y2) -> ubyte {
+        ; From a pair of positive coordinates, calculate discrete direction between 0 and 23 into A.
+        ; This adjusts the atan() result  so that the direction N is centered on the angle=N instead of having it as a boundary
+        ubyte angle = atan2(x1, y1, x2, y2) - 256/48
+        return 23-lsb(mkword(angle,0) / 2730)
+    }
+
+    sub direction_sc(byte x1, byte y1, byte x2, byte y2) -> ubyte {
+        ; From a pair of signed coordinates around the origin, calculate discrete direction between 0 and 23 into A.
+        ; shift the points into the positive quadrant
+        ubyte px1
+        ubyte py1
+        ubyte px2
+        ubyte py2
+        if x1<0 or x2<0 {
+            px1 = x1 as ubyte + 128
+            px2 = x2 as ubyte + 128
+        } else {
+            px1 = x1 as ubyte
+            px2 = x2 as ubyte
+        }
+        if y1<0 or y2<0 {
+            py1 = y1 as ubyte + 128
+            py2 = y2 as ubyte + 128
+        } else {
+            py1 = y1 as ubyte
+            py2 = y2 as ubyte
+        }
+
+        return direction(px1, py1, px2, py2)
+    }
+
+    sub direction_qd(ubyte quadrant, ubyte xdelta, ubyte ydelta) -> ubyte {
+        ; From a pair of X/Y deltas (both >=0), and quadrant 0-3, calculate discrete direction between 0 and 23.
+        when quadrant {
+            3 -> return direction(0, 0, xdelta, ydelta)
+            2 -> return direction(xdelta, 0, 0, ydelta)
+            1 -> return direction(0, ydelta, xdelta, 0)
+            else -> return direction(xdelta, ydelta, 0, 0)
+        }
+    }
+
+    sub atan2(ubyte x1, ubyte y1, ubyte x2, ubyte y2) -> ubyte {
+        ;; Calculate the angle, in a 256-degree circle, between two points into A.
+        ;; The points (x1, y1) and (x2, y2) have to use *unsigned coordinates only* from the positive quadrant in the carthesian plane!
+        %ir {{
+            loadm.b r65532,math.atan2.x1
+            loadm.b r65533,math.atan2.y1
+            loadm.b r65534,math.atan2.x2
+            loadm.b r65535,math.atan2.y2
+            syscall 44 (r65532.b, r65533.b, r65534.b, r65535.b): r0.b
+            returnr.b r0
+        }}
+    }
+
+    sub mul16_last_upper() -> uword {
+        ; This routine peeks into the internal 32 bits multiplication result buffer of the
+        ; 16*16 bits multiplication routine, to fetch the upper 16 bits of the last calculation.
+        ; Notes:
+        ;   - to avoid interference it's best to fetch and store this value immediately after the multiplication expression.
+        ;     for instance, simply printing a number may already result in new multiplication calls being performed
+        ;   - not all multiplications in the source code result in an actual multiplication call:
+        ;     some simpler multiplications will be optimized away into faster routines. These will not set the upper 16 bits at all!
+        %ir {{
+            syscall 46 (): r0.w
+            returnr.w r0
+        }}
+    }
+
+    sub diff(ubyte b1, ubyte b2) -> ubyte {
+        if b1>b2
+            return b1-b2
+        return b2-b1
+    }
+
+    sub diffw(uword w1, uword w2) -> uword {
+        if w1>w2
+            return w1-w2
+        return w2-w1
+    }
 
 }

compiler/res/prog8lib/virtual/monogfx.p8

@@ -0,0 +1,471 @@
+; Monochrome Bitmap pixel graphics routines for the Virtual Machine
+; Using the full-screen 640x480 and 320x240 screen modes, but just black/white.
+;
+; NOTE: For sake of speed, NO BOUNDS CHECKING is performed in most routines!
+;       You'll have to make sure yourself that you're not writing outside of bitmap boundaries!
+
+
+monogfx {
+
+    %option no_symbol_prefixing
+
+    ; read-only control variables:
+    uword width = 0
+    uword height = 0
+    bool dont_stipple_flag = true            ; set to false to enable stippling mode
+
+    sub lores() {
+        ; enable 320*240 bitmap mode
+        sys.gfx_enable(0)
+        width = 320
+        height = 240
+        clear_screen(0)
+    }
+
+    sub hires() {
+        ; enable 640*480 bitmap mode
+        sys.gfx_enable(1)
+        width = 640
+        height = 480
+        clear_screen(0)
+    }
+
+    sub textmode() {
+        ; back to normal text mode
+    }
+
+    sub clear_screen(ubyte color) {
+        stipple(false)
+        if color
+            color=255
+        sys.gfx_clear(color)
+    }
+
+    sub stipple(bool enable) {
+        dont_stipple_flag = not enable
+    }
+
+    sub rect(uword xx, uword yy, uword rwidth, uword rheight, bool draw) {
+        if rwidth==0 or rheight==0
+            return
+        horizontal_line(xx, yy, rwidth, draw)
+        if rheight==1
+            return
+        horizontal_line(xx, yy+rheight-1, rwidth, draw)
+        vertical_line(xx, yy+1, rheight-2, draw)
+        if rwidth==1
+            return
+        vertical_line(xx+rwidth-1, yy+1, rheight-2, draw)
+    }
+
+    sub fillrect(uword xx, uword yy, uword rwidth, uword rheight, bool draw) {
+        ; Draw a filled rectangle of the given size and color.
+        ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
+        if rwidth==0
+            return
+        repeat rheight {
+            horizontal_line(xx, yy, rwidth, draw)
+            yy++
+        }
+    }
+
+    sub horizontal_line(uword xx, uword yy, uword length, bool draw) {
+        ubyte color = 0
+        if draw
+            color = 255
+        uword xpos
+        for xpos in xx to xx+length-1
+            plot(xpos, yy, color)
+    }
+
+    sub safe_horizontal_line(uword xx, uword yy, uword length, bool draw) {
+        ; does bounds checking and clipping
+        if msb(yy)&$80!=0 or yy>=height
+            return
+        if msb(xx)&$80!=0 {
+            length += xx
+            xx = 0
+        }
+        if xx>=width
+            return
+        if xx+length>width
+            length = width-xx
+        if length>width
+            return
+
+        horizontal_line(xx, yy, length, draw)
+    }
+
+    sub vertical_line(uword xx, uword yy, uword lheight, bool draw) {
+        ubyte color = 0
+        if draw
+            color = 255
+        uword ypos
+        for ypos in yy to yy+lheight-1
+            plot(xx, ypos, color)
+    }
+
+    sub line(uword @zp x1, uword @zp y1, uword @zp x2, uword @zp y2, bool draw) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        if y1>y2 {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            cx16.r0 = x1
+            x1 = x2
+            x2 = cx16.r0
+            cx16.r0 = y1
+            y1 = y2
+            y2 = cx16.r0
+        }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as uword +1, draw)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1, draw)
+            return
+        }
+
+        word @zp d = 0
+        cx16.r1L = true      ; 'positive_ix'
+        if dx < 0 {
+            dx = -dx
+            cx16.r1L = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+        cx16.r14 = x1       ; internal plot X
+
+        if dx >= dy {
+            if cx16.r1L {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14++
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14--
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if cx16.r1L {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, draw)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+    }
+
+    sub circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, bool draw) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+
+            yy++
+            if decisionOver2>=0 {
+                xx--
+                decisionOver2 -= xx*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+
+        sub plotq() {
+            ; cx16.r14 = x, cx16.r15 = y, draw=draw
+            plot(cx16.r14, cx16.r15, draw)
+        }
+    }
+
+    sub safe_circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, bool draw) {
+        ; Does bounds checking and clipping.
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plotq()
+            cx16.r14 = xcenter - xx
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plotq()
+            cx16.r14 = xcenter - yy
+            plotq()
+
+            yy++
+            if decisionOver2>=0 {
+                xx--
+                decisionOver2 -= xx*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+
+        sub plotq() {
+            ; cx16.r14 = x, cx16.r15 = y, draw=draw
+            safe_plot(cx16.r14, cx16.r15, draw)
+        }
+    }
+
+    sub disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+    }
+
+    sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
+        ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
+            safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+    }
+
+    sub plot(uword @zp xx, uword @zp yy, bool @zp draw) {
+        if draw {
+            if dont_stipple_flag
+                sys.gfx_plot(xx, yy, 255)
+            else {
+                if (xx ^ yy)&1
+                    sys.gfx_plot(xx, yy, 255)
+                else
+                    sys.gfx_plot(xx, yy, 0)
+            }
+        }
+        else
+            sys.gfx_plot(xx, yy, 0)
+    }
+
+    sub safe_plot(uword xx, uword yy, bool draw) {
+        ; A plot that does bounds checks to see if the pixel is inside the screen.
+        if msb(xx)&$80!=0 or msb(yy)&$80!=0
+            return
+        if xx >= width or yy >= height
+            return
+        plot(xx, yy, draw)
+    }
+
+    sub pget(uword @zp xx, uword yy) -> ubyte {
+        return sys.gfx_getpixel(xx, yy)
+    }
+
+    sub fill(uword x, uword y, bool draw) {
+        ; Non-recursive scanline flood fill.
+        ; based loosely on code found here https://www.codeproject.com/Articles/6017/QuickFill-An-efficient-flood-fill-algorithm
+        ; with the fixes applied to the seedfill_4 routine as mentioned in the comments.
+        const ubyte MAXDEPTH = 64
+        word @zp xx = x as word
+        word @zp yy = y as word
+        word[MAXDEPTH] @split @shared stack_xl
+        word[MAXDEPTH] @split @shared stack_xr
+        word[MAXDEPTH] @split @shared stack_y
+        byte[MAXDEPTH] @shared stack_dy
+        cx16.r12L = 0       ; stack pointer
+        word x1
+        word x2
+        byte dy
+        cx16.r10L = draw as ubyte
+        sub push_stack(word sxl, word sxr, word sy, byte sdy) {
+            if cx16.r12L==MAXDEPTH
+                return
+            cx16.r0s = sy+sdy
+            if cx16.r0s>=0 and cx16.r0s<=height-1 {
+                stack_xl[cx16.r12L] = sxl
+                stack_xr[cx16.r12L] = sxr
+                stack_y[cx16.r12L] = sy
+                stack_dy[cx16.r12L] = sdy
+                cx16.r12L++
+            }
+        }
+        sub pop_stack() {
+            cx16.r12L--
+            x1 = stack_xl[cx16.r12L]
+            x2 = stack_xr[cx16.r12L]
+            yy = stack_y[cx16.r12L]
+            dy = stack_dy[cx16.r12L]
+            yy+=dy
+        }
+        cx16.r11L = pget(xx as uword, yy as uword)        ; old_color
+        if cx16.r11L == cx16.r10L
+            return
+        if xx<0 or xx>width-1 or yy<0 or yy>height-1
+            return
+        push_stack(xx, xx, yy, 1)
+        push_stack(xx, xx, yy + 1, -1)
+        word left = 0
+        while cx16.r12L {
+            pop_stack()
+            xx = x1
+            while xx >= 0 {
+                if pget(xx as uword, yy as uword) != cx16.r11L
+                    break
+                xx--
+            }
+            if x1!=xx
+                horizontal_line(xx as uword+1, yy as uword, x1-xx as uword, cx16.r10L)
+            else
+                goto skip
+
+            left = xx + 1
+            if left < x1
+                push_stack(left, x1 - 1, yy, -dy)
+            xx = x1 + 1
+
+            do {
+                cx16.r9 = xx
+                while xx <= width-1 {
+                    if pget(xx as uword, yy as uword) != cx16.r11L
+                        break
+                    xx++
+                }
+                if cx16.r9!=xx
+                    horizontal_line(cx16.r9, yy as uword, (xx as uword)-cx16.r9, cx16.r10L)
+
+                push_stack(left, xx - 1, yy, dy)
+                if xx > x2 + 1
+                    push_stack(x2 + 1, xx - 1, yy, -dy)
+skip:
+                xx++
+                while xx <= x2 {
+                    if pget(xx as uword, yy as uword) == cx16.r11L
+                        break
+                    xx++
+                }
+                left = xx
+            } until xx>x2
+        }
+    }
+
+    sub text_charset(ubyte charset) {
+        ; -- select the text charset to use with the text() routine
+        ;    the charset number is the same as for the cx16.screen_set_charset() ROM function.
+        ;    1 = ISO charset, 2 = PETSCII uppercase+graphs, 3= PETSCII uppercase+lowercase.
+        ; TODO vm has no bitmap charset
+    }
+
+    sub text(uword @zp xx, uword yy, bool draw, 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.
+        ; TODO vm has no bitmap charset
+    }
+}

compiler/res/prog8lib/virtual/string.p8

@@ -78,6 +78,13 @@ string {
         }
     }
 
+    sub append(str target, str suffix) -> ubyte {
+        ; Append the suffix string to the target. (make sure the buffer is large enough!)
+        ; Returns the length of the resulting string.
+        cx16.r0L = length(target)
+        return copy(suffix, target+cx16.r0L) + cx16.r0L
+    }
+
     sub compare(str st1, str st2) -> byte {
         ; Compares two strings for sorting.
         ; Returns -1 (255), 0 or 1 depending on wether string1 sorts before, equal or after string2.

compiler/res/prog8lib/virtual/syslib.p8

@@ -79,6 +79,13 @@ sys {
         }}
     }
 
+    sub disable_caseswitch() {
+        ; no-op
+    }
+
+    sub enable_caseswitch() {
+        ; no-op
+    }
 
     sub gfx_enable(ubyte mode) {
         %ir {{

compiler/src/prog8/CompilerMain.kt

@@ -42,13 +42,15 @@ private fun compileMain(args: Array<String>): Boolean {
     val startEmulator2 by cli.option(ArgType.Boolean, fullName = "emu2", description = "auto-start alternative emulator after successful compilation")
     val experimentalCodegen by cli.option(ArgType.Boolean, fullName = "expericodegen", description = "use experimental/alternative codegen")
     val dontWriteAssembly by cli.option(ArgType.Boolean, fullName = "noasm", description="don't create assembly code")
-    val dontOptimize by cli.option(ArgType.Boolean, fullName = "noopt", description = "don't perform any optimizations")
+    val dontOptimize by cli.option(ArgType.Boolean, fullName = "noopt", description = "don't perform code optimizations")
     val outputDir by cli.option(ArgType.String, fullName = "out", description = "directory for output files instead of current directory").default(".")
     val quietAssembler by cli.option(ArgType.Boolean, fullName = "quietasm", description = "don't print assembler output results")
+    val warnSymbolShadowing by cli.option(ArgType.Boolean, fullName = "warnshadow", description="show assembler warnings about symbol shadowing")
     val sourceDirs by cli.option(ArgType.String, fullName="srcdirs", description = "list of extra paths, separated with ${File.pathSeparator}, to search in for imported modules").multiple().delimiter(File.pathSeparator)
     val includeSourcelines by cli.option(ArgType.Boolean, fullName = "sourcelines", description = "include original Prog8 source lines in generated asm code")
     val splitWordArrays by cli.option(ArgType.Boolean, fullName = "splitarrays", description = "treat all word arrays as tagged with @split to make them lsb/msb split in memory")
-    val compilationTarget by cli.option(ArgType.String, fullName = "target", description = "target output of the compiler (one of '${C64Target.NAME}', '${C128Target.NAME}', '${Cx16Target.NAME}', '${AtariTarget.NAME}', '${VMTarget.NAME}') (required)")
+    val breakpointCpuInstruction by cli.option(ArgType.Boolean, fullName = "breakinstr", description = "also use a CPU instruction for %breakpoint")
+    val compilationTarget by cli.option(ArgType.String, fullName = "target", description = "target output of the compiler (one of '${C64Target.NAME}', '${C128Target.NAME}', '${Cx16Target.NAME}', '${AtariTarget.NAME}', '${PETTarget.NAME}', '${VMTarget.NAME}') (required)")
     val startVm by cli.option(ArgType.Boolean, fullName = "vm", description = "load and run a .p8ir IR source file in the VM")
     val watchMode by cli.option(ArgType.Boolean, fullName = "watch", description = "continuous compilation mode (watch for file changes)")
     val varsHighBank by cli.option(ArgType.Int, fullName = "varshigh", description = "put uninitialized variables in high memory area instead of at the end of the program. On the cx16 target the value specifies the HiRAM bank to use, on other systems this value is ignored.")
@@ -83,7 +85,7 @@ private fun compileMain(args: Array<String>): Boolean {
             return false
         }
 
-        if (compilationTarget !in setOf(C64Target.NAME, C128Target.NAME, Cx16Target.NAME, AtariTarget.NAME, VMTarget.NAME)) {
+        if (compilationTarget !in setOf(C64Target.NAME, C128Target.NAME, Cx16Target.NAME, AtariTarget.NAME, PETTarget.NAME, VMTarget.NAME)) {
             System.err.println("Invalid compilation target: $compilationTarget")
             return false
         }
@@ -113,6 +115,7 @@ private fun compileMain(args: Array<String>): Boolean {
                     filepath,
                     dontOptimize != true,
                     dontWriteAssembly != true,
+                    warnSymbolShadowing == true,
                     quietAssembler == true,
                     asmListfile == true,
                     includeSourcelines == true,
@@ -120,6 +123,7 @@ private fun compileMain(args: Array<String>): Boolean {
                     varsHighBank,
                     compilationTarget!!,
                     splitWordArrays == true,
+                    breakpointCpuInstruction = false,
                     processedSymbols,
                     srcdirs,
                     outputPath
@@ -180,6 +184,7 @@ private fun compileMain(args: Array<String>): Boolean {
                     filepath,
                     dontOptimize != true,
                     dontWriteAssembly != true,
+                    warnSymbolShadowing == true,
                     quietAssembler == true,
                     asmListfile == true,
                     includeSourcelines == true,
@@ -187,6 +192,7 @@ private fun compileMain(args: Array<String>): Boolean {
                     varsHighBank,
                     compilationTarget!!,
                     splitWordArrays == true,
+                    breakpointCpuInstruction == true,
                     processedSymbols,
                     srcdirs,
                     outputPath

compiler/src/prog8/buildversion/BuildVersion.kt

@@ -5,11 +5,11 @@ package prog8.buildversion
  */
 const val MAVEN_GROUP = "prog8"
 const val MAVEN_NAME = "compiler"
-const val VERSION = "9.3-SNAPSHOT"
-const val GIT_REVISION = 4012
-const val GIT_SHA = "9167ba499de99696aa7c6887d4867eae70eb5caf"
-const val GIT_DATE = "2023-07-30T15:49:35Z"
+const val VERSION = "9.6-SNAPSHOT"
+const val GIT_REVISION = 4159
+const val GIT_SHA = "335213b55f971452a05c904216cb0fd72f5518a1"
+const val GIT_DATE = "2023-10-21T00:16:58Z"
 const val GIT_BRANCH = "master"
-const val BUILD_DATE = "2023-07-30T16:41:27Z"
-const val BUILD_UNIX_TIME = 1690735287514L
+const val BUILD_DATE = "2023-10-21T20:22:14Z"
+const val BUILD_UNIX_TIME = 1697919734197L
 const val DIRTY = 1

compiler/src/prog8/compiler/Compiler.kt

@@ -30,6 +30,7 @@ class CompilationResult(val compilerAst: Program,   // deprecated, use codegenAs
 class CompilerArguments(val filepath: Path,
                         val optimize: Boolean,
                         val writeAssembly: Boolean,
+                        val warnSymbolShadowing: Boolean,
                         val quietAssembler: Boolean,
                         val asmListfile: Boolean,
                         val includeSourcelines: Boolean,
@@ -37,6 +38,7 @@ class CompilerArguments(val filepath: Path,
                         val varsHighBank: Int?,
                         val compilationTarget: String,
                         val splitWordArrays: Boolean,
+                        val breakpointCpuInstruction: Boolean,
                         val symbolDefs: Map<String, String>,
                         val sourceDirs: List<String> = emptyList(),
                         val outputDir: Path = Path(""),
@@ -52,6 +54,7 @@ fun compileProgram(args: CompilerArguments): CompilationResult? {
             C64Target.NAME -> C64Target()
             C128Target.NAME -> C128Target()
             Cx16Target.NAME -> Cx16Target()
+            PETTarget.NAME -> PETTarget()
             AtariTarget.NAME -> AtariTarget()
             VMTarget.NAME -> VMTarget()
             else -> throw IllegalArgumentException("invalid compilation target")
@@ -66,11 +69,13 @@ fun compileProgram(args: CompilerArguments): CompilationResult? {
             compilationOptions = options
 
             with(compilationOptions) {
+                warnSymbolShadowing = args.warnSymbolShadowing
                 optimize = args.optimize
                 asmQuiet = args.quietAssembler
                 asmListfile = args.asmListfile
                 includeSourcelines = args.includeSourcelines
                 experimentalCodegen = args.experimentalCodegen
+                breakpointCpuInstruction = args.breakpointCpuInstruction
                 varsHighBank = args.varsHighBank
                 splitWordArrays = args.splitWordArrays
                 outputDir = args.outputDir.normalize()
@@ -102,11 +107,12 @@ fun compileProgram(args: CompilerArguments): CompilationResult? {
 
             if (args.writeAssembly) {
 
+                // re-initialize memory areas with final compilationOptions
                 compilationOptions.compTarget.machine.initializeMemoryAreas(compilationOptions)
                 program.processAstBeforeAsmGeneration(compilationOptions, args.errors)
                 args.errors.report()
 
-                val intermediateAst = IntermediateAstMaker(program).transform()
+                val intermediateAst = IntermediateAstMaker(program, args.errors).transform()
 //                println("*********** COMPILER AST RIGHT BEFORE ASM GENERATION *************")
 //                printProgram(program)
 //                println("*********** AST RIGHT BEFORE ASM GENERATION *************")
@@ -251,6 +257,10 @@ fun parseMainModule(filepath: Path,
         importer.importImplicitLibraryModule("math")
     }
     importer.importImplicitLibraryModule("prog8_lib")
+    if(program.allBlocks.any { it.options().any { option->option=="verafxmuls" } }) {
+        if(compTarget.name==Cx16Target.NAME)
+            importer.importImplicitLibraryModule("verafx")
+    }
 
     if (compilerOptions.launcher == CbmPrgLauncherType.BASIC && compilerOptions.output != OutputType.PRG)
         errors.err("BASIC launcher requires output type PRG", program.toplevelModule.position)
@@ -292,6 +302,14 @@ fun determineCompilationOptions(program: Program, compTarget: ICompilationTarget
         .map { it[0].int!!..it[1].int!! }
         .toList()
 
+    val zpAllowed = toplevelModule.statements
+        .asSequence()
+        .filter { it is Directive && it.directive == "%zpallowed" }
+        .map { (it as Directive).args }
+        .filter { it.size==2 && it[0].int!=null && it[1].int!=null }
+        .map { it[0].int!!..it[1].int!! }
+        .toList()
+
     val outputType = if (outputTypeStr == null) {
         if(compTarget is AtariTarget)
             OutputType.XEX
@@ -321,7 +339,7 @@ fun determineCompilationOptions(program: Program, compTarget: ICompilationTarget
 
     return CompilationOptions(
         outputType, launcherType,
-        zpType, zpReserved, floatsEnabled, noSysInit,
+        zpType, zpReserved, zpAllowed, floatsEnabled, noSysInit,
         compTarget, 0u
     )
 }

compiler/src/prog8/compiler/ModuleImporter.kt

@@ -127,7 +127,7 @@ class ModuleImporter(private val program: Program,
 
     private fun removeDirectivesFromImportedModule(importedModule: Module) {
         // Most global directives don't apply for imported modules, so remove them
-        val moduleLevelDirectives = listOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address")
+        val moduleLevelDirectives = listOf("%output", "%launcher", "%zeropage", "%zpreserved", "%zpallowed", "%address")
         var directives = importedModule.statements.filterIsInstance<Directive>()
         importedModule.statements.removeAll(directives.toSet())
         directives = directives.filter{ it.directive !in moduleLevelDirectives }

compiler/src/prog8/compiler/astprocessing/AstChecker.kt

@@ -7,6 +7,8 @@ import prog8.ast.expressions.*
 import prog8.ast.statements.*
 import prog8.ast.walk.IAstVisitor
 import prog8.code.core.*
+import prog8.code.target.C64Target
+import prog8.code.target.Cx16Target
 import prog8.code.target.VMTarget
 import prog8.compiler.builtinFunctionReturnType
 import java.io.CharConversionException
@@ -76,14 +78,19 @@ internal class AstChecker(private val program: Program,
     }
 
     override fun visit(unrollLoop: UnrollLoop) {
-        if(unrollLoop.iterations<0 || unrollLoop.iterations>65535)
-            errors.err("invalid number of unrolls", unrollLoop.position)
-        unrollLoop.body.statements.forEach {
-            if(it !is InlineAssembly && it !is Assignment && it !is BuiltinFunctionCallStatement && it !is FunctionCallStatement && it !is PostIncrDecr)
-                errors.err("invalid statement in unroll loop", it.position)
-        }
-        if(unrollLoop.iterations * unrollLoop.body.statements.size > 256) {
-            errors.warn("large number of unrolls, potential code size issue", unrollLoop.position)
+        val iterations = unrollLoop.iterations.constValue(program)?.number?.toInt()
+        if(iterations==null) {
+            errors.err("unroll needs constant number of iterations", unrollLoop.position)
+        } else {
+            if (iterations < 0 || iterations > 65535)
+                errors.err("invalid number of unrolls", unrollLoop.position)
+            unrollLoop.body.statements.forEach {
+                if (it !is InlineAssembly && it !is Assignment && it !is BuiltinFunctionCallStatement && it !is FunctionCallStatement && it !is PostIncrDecr)
+                    errors.err("invalid statement in unroll loop", it.position)
+            }
+            if (iterations * unrollLoop.body.statements.size > 256) {
+                errors.warn("large number of unrolls, potential code size issue", unrollLoop.position)
+            }
         }
         super.visit(unrollLoop)
     }
@@ -220,7 +227,7 @@ internal class AstChecker(private val program: Program,
     override fun visit(jump: Jump) {
         val ident = jump.identifier
         if(ident!=null) {
-            val targetStatement = checkFunctionOrLabelExists(ident, jump)
+            val targetStatement = ident.checkFunctionOrLabelExists(program, jump, errors)
             if(targetStatement!=null) {
                 if(targetStatement is BuiltinFunctionPlaceholder)
                     errors.err("can't jump to a builtin function", jump.position)
@@ -306,7 +313,7 @@ internal class AstChecker(private val program: Program,
         if(subroutine.name in BuiltinFunctions)
             err("cannot redefine a built-in function")
 
-        if(subroutine.parameters.size>6 && !subroutine.isAsmSubroutine)
+        if(subroutine.parameters.size>6 && !subroutine.isAsmSubroutine && !subroutine.definingBlock.isInLibrary)
             errors.warn("subroutine has a large number of parameters, this slows down code execution a lot", subroutine.position)
 
         val uniqueNames = subroutine.parameters.asSequence().map { it.name }.toSet()
@@ -766,7 +773,7 @@ internal class AstChecker(private val program: Program,
                         directive.args[0].name != "full")
                     err("invalid zp type, expected basicsafe, floatsafe, kernalsafe, dontuse, or full")
             }
-            "%zpreserved" -> {
+            "%zpreserved", "%zpallowed" -> {
                 if(directive.parent !is Module)
                     err("this directive may only occur at module level")
                 if(directive.args.size!=2 || directive.args[0].int==null || directive.args[1].int==null)
@@ -817,8 +824,16 @@ internal class AstChecker(private val program: Program,
                     err("this directive may only occur in a block or at module level")
                 if(directive.args.isEmpty())
                     err("missing option directive argument(s)")
-                else if(directive.args.map{it.name in arrayOf("enable_floats", "force_output", "no_sysinit", "align_word", "align_page", "merge", "splitarrays", "no_symbol_prefixing")}.any { !it })
+                else if(directive.args.map{it.name in arrayOf("enable_floats", "force_output", "no_sysinit", "align_word", "align_page", "merge", "splitarrays", "no_symbol_prefixing", "verafxmuls")}.any { !it })
                     err("invalid option directive argument(s)")
+                if(directive.args.any {it.name=="align_word"} && directive.args.any { it.name=="align_page"})
+                    err("conflicting alignment options")
+                if(directive.parent is Block) {
+                    if(directive.args.any {it.name !in arrayOf("align_word", "align_page", "no_symbol_prefixing", "force_output", "merge", "splitarrays", "verafxmuls")})
+                        err("using an option that is not valid for blocks")
+                }
+                if(directive.args.any { it.name=="verafxmuls" } && compilerOptions.compTarget.name != Cx16Target.NAME)
+                    err("verafx option is only valid on cx16 target")
             }
             else -> throw SyntaxError("invalid directive ${directive.directive}", directive.position)
         }
@@ -1068,7 +1083,7 @@ internal class AstChecker(private val program: Program,
         val stmtOfExpression = findParentNode<Statement>(functionCallExpr)
                 ?: throw FatalAstException("cannot determine statement scope of function call expression at ${functionCallExpr.position}")
 
-        val targetStatement = checkFunctionOrLabelExists(functionCallExpr.target, stmtOfExpression)
+        val targetStatement = functionCallExpr.target.checkFunctionOrLabelExists(program, stmtOfExpression, errors)
         if(targetStatement!=null)
             checkFunctionCall(targetStatement, functionCallExpr.args, functionCallExpr.position)
 
@@ -1120,7 +1135,7 @@ internal class AstChecker(private val program: Program,
     }
 
     override fun visit(functionCallStatement: FunctionCallStatement) {
-        val targetStatement = checkFunctionOrLabelExists(functionCallStatement.target, functionCallStatement)
+        val targetStatement = functionCallStatement.target.checkFunctionOrLabelExists(program, functionCallStatement, errors)
         if(targetStatement!=null) {
             checkFunctionCall(targetStatement, functionCallStatement.args, functionCallStatement.position)
             checkUnusedReturnValues(functionCallStatement, targetStatement, errors)
@@ -1178,8 +1193,13 @@ internal class AstChecker(private val program: Program,
 
             if(funcName[0] in InplaceModifyingBuiltinFunctions) {
                 // in-place modification, can't be done on literals
-                if(functionCallStatement.args.any { it !is IdentifierReference && it !is ArrayIndexedExpression && it !is DirectMemoryRead }) {
-                    errors.err("invalid argument to a in-place modifying function", functionCallStatement.args.first().position)
+                if(funcName[0]=="setlsb" || funcName[0]=="setmsb") {
+                    val firstArg = functionCallStatement.args[0]
+                    if(firstArg !is IdentifierReference && firstArg !is ArrayIndexedExpression)
+                        errors.err("invalid argument to a in-place modifying function", functionCallStatement.args.first().position)
+                } else {
+                    if(functionCallStatement.args.any { it !is IdentifierReference && it !is ArrayIndexedExpression && it !is DirectMemoryRead })
+                        errors.err("invalid argument to a in-place modifying function", functionCallStatement.args.first().position)
                 }
             }
 
@@ -1328,6 +1348,9 @@ internal class AstChecker(private val program: Program,
         if(whenStmt.choices.isEmpty())
             errors.err("empty when statement", whenStmt.position)
 
+        if(whenStmt.condition.constValue(program)!=null)
+            errors.warn("when-value is a constant and will always result in the same choice", whenStmt.condition.position)
+
         super.visit(whenStmt)
     }
 
@@ -1421,25 +1444,6 @@ internal class AstChecker(private val program: Program,
             errors.err("%asm containing IR code cannot be translated to 6502 assembly", inlineAssembly.position)
     }
 
-    private fun checkFunctionOrLabelExists(target: IdentifierReference, statement: Statement): Statement? {
-        when (val targetStatement = target.targetStatement(program)) {
-            is Label, is Subroutine, is BuiltinFunctionPlaceholder -> return targetStatement
-            is VarDecl -> {
-                if(statement is Jump) {
-                    if (targetStatement.datatype == DataType.UWORD)
-                        return targetStatement
-                    else
-                        errors.err("wrong address variable datatype, expected uword", target.position)
-                }
-                else
-                    errors.err("cannot call that: ${target.nameInSource.joinToString(".")}", target.position)
-            }
-            null -> errors.undefined(target.nameInSource, target.position)
-            else -> errors.err("cannot call that: ${target.nameInSource.joinToString(".")}", target.position)
-        }
-        return null
-    }
-
     private fun checkValueTypeAndRangeString(targetDt: DataType, value: StringLiteral) : Boolean {
         return if (targetDt == DataType.STR) {
             when {

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

@@ -5,13 +5,12 @@ import prog8.ast.Program
 import prog8.ast.expressions.CharLiteral
 import prog8.ast.expressions.IdentifierReference
 import prog8.ast.expressions.NumericLiteral
-import prog8.ast.statements.Directive
-import prog8.ast.statements.InlineAssembly
-import prog8.ast.statements.Subroutine
-import prog8.ast.statements.VarDeclOrigin
+import prog8.ast.expressions.StringLiteral
+import prog8.ast.statements.*
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
 import prog8.code.core.*
+import java.io.CharConversionException
 
 
 internal fun Program.checkValid(errors: IErrorReporter, compilerOptions: CompilationOptions) {
@@ -66,13 +65,35 @@ internal fun Program.charLiteralsToUByteLiterals(target: ICompilationTarget, err
                 errors.err("compilation target doesn't support this text encoding", char.position)
                 return noModifications
             }
-            return listOf(IAstModification.ReplaceNode(
-                char,
-                NumericLiteral(DataType.UBYTE, target.encodeString(char.value.toString(), char.encoding)[0].toDouble(), char.position),
-                parent
-            ))
+            return try {
+                val encoded = target.encodeString(char.value.toString(), char.encoding)
+                listOf(IAstModification.ReplaceNode(
+                    char,
+                    NumericLiteral(DataType.UBYTE, encoded[0].toDouble(), char.position),
+                    parent
+                ))
+            } catch (x: CharConversionException) {
+                errors.err(x.message ?: "can't encode character", char.position)
+                noModifications
+            }
+        }
+
+        override fun after(string: StringLiteral, parent: Node): Iterable<IAstModification> {
+            // this only *checks* for errors for string encoding. The actual encoding is done much later
+            require(string.encoding != Encoding.DEFAULT)
+            if(string.encoding != Encoding.DEFAULT && string.encoding !in target.supportedEncodings) {
+                errors.err("compilation target doesn't support this text encoding", string.position)
+                return noModifications
+            }
+            try {
+                target.encodeString(string.value, string.encoding)
+            } catch (x: CharConversionException) {
+                errors.err(x.message ?: "can't encode string", string.position)
+            }
+            return noModifications
         }
     }
+
     walker.visit(this)
     walker.applyModifications()
 }
@@ -151,4 +172,25 @@ internal fun Subroutine.hasRtsInAsm(): Boolean {
         .asSequence()
         .filterIsInstance<InlineAssembly>()
         .any { it.hasReturnOrRts() }
+}
+
+internal fun IdentifierReference.checkFunctionOrLabelExists(program: Program, statement: Statement, errors: IErrorReporter): Statement? {
+    when (val targetStatement = this.targetStatement(program)) {
+        is Label, is Subroutine, is BuiltinFunctionPlaceholder -> return targetStatement
+        is VarDecl -> {
+            if(statement is Jump) {
+                if (targetStatement.datatype == DataType.UWORD)
+                    return targetStatement
+                else
+                    errors.err("wrong address variable datatype, expected uword", this.position)
+            }
+            else
+                errors.err("cannot call that: ${this.nameInSource.joinToString(".")}", this.position)
+        }
+        null -> {
+            errors.undefined(this.nameInSource, this.position)
+        }
+        else -> errors.err("cannot call that: ${this.nameInSource.joinToString(".")}", this.position)
+    }
+    return null
 }

compiler/src/prog8/compiler/astprocessing/AstPreprocessor.kt

@@ -1,9 +1,12 @@
 package prog8.compiler.astprocessing
 
+import prog8.ast.IFunctionCall
 import prog8.ast.Node
 import prog8.ast.Program
+import prog8.ast.base.FatalAstException
 import prog8.ast.base.SyntaxError
 import prog8.ast.expressions.*
+import prog8.ast.findParentNode
 import prog8.ast.statements.*
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
@@ -201,4 +204,42 @@ class AstPreprocessor(val program: Program,
 
         return noModifications
     }
+
+    override fun after(bfc: BuiltinFunctionCall, parent: Node): Iterable<IAstModification> {
+        val stmtOfExpression = findParentNode<Statement>(bfc)
+            ?: throw FatalAstException("cannot determine statement scope of function call expression at ${bfc.position}")
+        checkStringParam(bfc as IFunctionCall, stmtOfExpression)
+        return noModifications
+    }
+
+    override fun after(bfcs: BuiltinFunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        checkStringParam(bfcs as IFunctionCall, bfcs)
+        return noModifications
+    }
+
+    override fun after(functionCallExpr: FunctionCallExpression, parent: Node): Iterable<IAstModification> {
+        val stmtOfExpression = findParentNode<Statement>(functionCallExpr)
+            ?: throw FatalAstException("cannot determine statement scope of function call expression at ${functionCallExpr.position}")
+
+        checkStringParam(functionCallExpr as IFunctionCall, stmtOfExpression)
+        return noModifications
+    }
+
+    override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        checkStringParam(functionCallStatement as IFunctionCall, functionCallStatement)
+        return noModifications
+    }
+
+    private fun checkStringParam(call: IFunctionCall, stmt: Statement) {
+        val targetStatement = call.target.checkFunctionOrLabelExists(program, stmt, errors)
+        if(targetStatement!=null) {
+            if(targetStatement is Subroutine) {
+                for(arg in call.args.zip(targetStatement.parameters)) {
+                    if(arg.first.inferType(program).isBytes && arg.second.type==DataType.STR) {
+                        errors.err("cannot use byte value for string parameter", arg.first.position)
+                    }
+                }
+            }
+        }
+    }
 }

compiler/src/prog8/compiler/astprocessing/BeforeAsmAstChanger.kt

@@ -4,8 +4,10 @@ import prog8.ast.IStatementContainer
 import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.FatalAstException
-import prog8.ast.expressions.*
-import prog8.ast.getTempVar
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.ContainmentCheck
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.NumericLiteral
 import prog8.ast.statements.*
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
@@ -39,62 +41,6 @@ internal class BeforeAsmAstChanger(val program: Program, private val options: Co
         return noModifications
     }
 
-    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
-        // Try to replace A = B <operator> Something  by A= B, A = A <operator> Something
-        // this triggers the more efficent augmented assignment code generation more often.
-        // But it can only be done if the target variable IS NOT OCCURRING AS AN OPERAND ITSELF.
-
-        if(options.compTarget.name==VMTarget.NAME)   // don't apply this optimization for Vm target
-            return noModifications
-
-        if(!assignment.isAugmentable
-            && assignment.target.identifier != null
-            && !assignment.target.isIOAddress(options.compTarget.machine)) {
-            val binExpr = assignment.value as? BinaryExpression
-
-            if (binExpr != null && binExpr.operator !in ComparisonOperators) {
-                if (binExpr.left !is BinaryExpression) {
-                    if (binExpr.right.referencesIdentifier(assignment.target.identifier!!.nameInSource)) {
-                        // the right part of the expression contains the target variable itself.
-                        // we can't 'split' it trivially because the variable will be changed halfway through.
-                        if(binExpr.operator in AssociativeOperators) {
-                            // A = <something-without-A>  <associativeoperator>  <otherthing-with-A>
-                            // use the other part of the expression to split.
-                            val sourceDt = binExpr.right.inferType(program).getOrElse { throw AssemblyError("unknown dt") }
-                            val (_, right) = binExpr.right.typecastTo(assignment.target.inferType(program).getOrElse { throw AssemblyError(
-                                "unknown dt"
-                            )
-                            }, sourceDt, implicit=true)
-                            val assignRight = Assignment(assignment.target, right, AssignmentOrigin.ASMGEN, assignment.position)
-                            return listOf(
-                                IAstModification.InsertBefore(assignment, assignRight, parent as IStatementContainer),
-                                IAstModification.ReplaceNode(binExpr.right, binExpr.left, binExpr),
-                                IAstModification.ReplaceNode(binExpr.left, assignment.target.toExpression(), binExpr)
-                            )
-                        }
-                    } else {
-                        if(binExpr.left isSameAs assignment.target)
-                            return noModifications
-                        val typeCast = binExpr.left as? TypecastExpression
-                        if(typeCast!=null && typeCast.expression isSameAs assignment.target)
-                            return noModifications
-                        val sourceDt = binExpr.left.inferType(program).getOrElse { throw AssemblyError("unknown dt") }
-                        val (_, left) = binExpr.left.typecastTo(assignment.target.inferType(program).getOrElse { throw AssemblyError(
-                            "unknown dt"
-                        )
-                        }, sourceDt, implicit=true)
-                        val assignLeft = Assignment(assignment.target, left, AssignmentOrigin.ASMGEN, assignment.position)
-                        return listOf(
-                            IAstModification.InsertBefore(assignment, assignLeft, parent as IStatementContainer),
-                            IAstModification.ReplaceNode(binExpr.left, assignment.target.toExpression(), binExpr)
-                        )
-                    }
-                }
-            }
-        }
-        return noModifications
-    }
-
     override fun after(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
         if(scope.statements.any { it is VarDecl || it is IStatementContainer })
             throw FatalAstException("anonymousscope may no longer contain any vardecls or subscopes")
@@ -211,35 +157,4 @@ internal class BeforeAsmAstChanger(val program: Program, private val options: Co
 
         return noModifications
     }
-
-    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
-
-        if(options.compTarget.name!=VMTarget.NAME) {    // don't apply this optimization/check for Vm target
-            val index = arrayIndexedExpression.indexer.indexExpr
-            if (index !is NumericLiteral && index !is IdentifierReference) {
-                // replace complex indexing expression with a temp variable to hold the computed index first
-                return getAutoIndexerVarFor(arrayIndexedExpression)
-            }
-        }
-
-        return noModifications
-    }
-
-    private fun getAutoIndexerVarFor(expr: ArrayIndexedExpression): MutableList<IAstModification> {
-        val modifications = mutableListOf<IAstModification>()
-        val statement = expr.containingStatement
-        val dt = expr.indexer.indexExpr.inferType(program)
-        val (tempVarName, _) = program.getTempVar(dt.getOrElse { throw FatalAstException("invalid dt") })
-        val target = AssignTarget(IdentifierReference(tempVarName, expr.indexer.position), null, null, expr.indexer.position)
-        val assign = Assignment(target, expr.indexer.indexExpr, AssignmentOrigin.ASMGEN, expr.indexer.position)
-        modifications.add(IAstModification.InsertBefore(statement, assign, statement.parent as IStatementContainer))
-        modifications.add(
-            IAstModification.ReplaceNode(
-                expr.indexer.indexExpr,
-                target.identifier!!.copy(),
-                expr.indexer
-            )
-        )
-        return modifications
-    }
 }

compiler/src/prog8/compiler/astprocessing/BeforeAsmTypecastCleaner.kt

@@ -56,7 +56,7 @@ internal class BeforeAsmTypecastCleaner(val program: Program,
                         listOf(
                             IAstModification.ReplaceNode(
                                 typecast,
-                                AddressOf(identifier, typecast.position),
+                                AddressOf(identifier, null, typecast.position),
                                 parent
                             )
                         )

compiler/src/prog8/compiler/astprocessing/CodeDesugarer.kt

@@ -24,6 +24,7 @@ internal class CodeDesugarer(val program: Program, private val errors: IErrorRep
     // - replace peek() and poke() by direct memory accesses.
     // - repeat-forever loops replaced by label+jump.
     // - pointer[word] replaced by @(pointer+word)
+    // - @(&var) and @(&var+1) replaced by lsb(var) and msb(var) if var is a word
 
 
     override fun before(breakStmt: Break, parent: Node): Iterable<IAstModification> {
@@ -200,6 +201,41 @@ _after:
             }
         }
 
+        if(expr.operator=="*" && expr.inferType(program).isInteger && expr.left isSameAs expr.right) {
+            // replace squaring with call to builtin function to do this in a more optimized way
+            val function = if(expr.left.inferType(program).isBytes) "prog8_lib_square_byte" else "prog8_lib_square_word"
+            val squareCall = BuiltinFunctionCall(
+                IdentifierReference(listOf(function), expr.position),
+                mutableListOf(expr.left.copy()), expr.position)
+            return listOf(IAstModification.ReplaceNode(expr, squareCall, parent))
+        }
+
         return noModifications
     }
+
+    override fun after(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
+        // for word variables:
+        // @(&var) --> lsb(var)
+        // @(&var+1) --> msb(var)
+
+        val addrOf = memread.addressExpression as? AddressOf
+        if(addrOf?.arrayIndex!=null)
+            return noModifications
+        if(addrOf!=null && addrOf.identifier.inferType(program).isWords) {
+            val lsb = FunctionCallExpression(IdentifierReference(listOf("lsb"), memread.position), mutableListOf(addrOf.identifier), memread.position)
+            return listOf(IAstModification.ReplaceNode(memread, lsb, parent))
+        }
+        val expr = memread.addressExpression as? BinaryExpression
+        if(expr!=null && expr.operator=="+") {
+            val addressOf = expr.left as? AddressOf
+            val offset = (expr.right as? NumericLiteral)?.number?.toInt()
+            if(addressOf!=null && offset==1) {
+                val msb = FunctionCallExpression(IdentifierReference(listOf("msb"), memread.position), mutableListOf(addressOf.identifier), memread.position)
+                return listOf(IAstModification.ReplaceNode(memread, msb, parent))
+            }
+        }
+
+        return noModifications
+    }
+
 }

compiler/src/prog8/compiler/astprocessing/IntermediateAstMaker.kt

@@ -19,7 +19,7 @@ import kotlin.io.path.isRegularFile
 /**
  *  Convert 'old' compiler-AST into the 'new' simplified AST with baked types.
  */
-class IntermediateAstMaker(private val program: Program) {
+class IntermediateAstMaker(private val program: Program, private val errors: IErrorReporter) {
     fun transform(): PtProgram {
         val ptProgram = PtProgram(
             program.name,
@@ -161,6 +161,7 @@ class IntermediateAstMaker(private val program: Program) {
     private fun transform(srcBlock: Block): PtBlock {
         var alignment = PtBlock.BlockAlignment.NONE
         var forceOutput = false
+        var veraFxMuls = false
         var noSymbolPrefixing = false
         val directives = srcBlock.statements.filterIsInstance<Directive>()
         for (directive in directives.filter { it.directive == "%option" }) {
@@ -169,15 +170,16 @@ class IntermediateAstMaker(private val program: Program) {
                     "align_word" -> alignment = PtBlock.BlockAlignment.WORD
                     "align_page" -> alignment = PtBlock.BlockAlignment.PAGE
                     "no_symbol_prefixing" -> noSymbolPrefixing = true
-                    "force_output" -> forceOutput=true
+                    "force_output" -> forceOutput = true
                     "merge", "splitarrays"  -> { /* ignore this one */ }
+                    "verafxmuls" -> veraFxMuls = true
                     else -> throw FatalAstException("weird directive option: ${arg.name}")
                 }
             }
         }
         val (vardecls, statements) = srcBlock.statements.partition { it is VarDecl }
         val src = srcBlock.definingModule.source
-        val block = PtBlock(srcBlock.name, srcBlock.address, srcBlock.isInLibrary, forceOutput, noSymbolPrefixing, alignment, src, srcBlock.position)
+        val block = PtBlock(srcBlock.name, srcBlock.address, srcBlock.isInLibrary, forceOutput, noSymbolPrefixing, veraFxMuls, alignment, src, srcBlock.position)
         makeScopeVarsDecls(vardecls).forEach { block.add(it) }
         for (stmt in statements)
             block.add(transformStatement(stmt))
@@ -326,7 +328,7 @@ class IntermediateAstMaker(private val program: Program) {
 
     private fun transform(srcUnroll: UnrollLoop): PtNodeGroup {
         val result = PtNodeGroup()
-        repeat(srcUnroll.iterations) {
+        repeat(srcUnroll.iterations.constValue(program)!!.number.toInt()) {
             srcUnroll.body.statements.forEach {
                 result.add(transformStatement(it))
             }
@@ -398,13 +400,26 @@ class IntermediateAstMaker(private val program: Program) {
     }
 
     private fun transform(srcVar: VarDecl): PtNode {
-        return when(srcVar.type) {
+        when(srcVar.type) {
             VarDeclType.VAR -> {
                 val value = if(srcVar.value!=null) transformExpression(srcVar.value!!) else null
-                PtVariable(srcVar.name, srcVar.datatype, srcVar.zeropage, value, srcVar.arraysize?.constIndex()?.toUInt(), srcVar.position)
+                if(srcVar.datatype in ArrayDatatypes) {
+                    if(value==null) {
+                        val blockOptions = srcVar.definingBlock.options()
+                        if("align_page" in blockOptions || "align_word" in blockOptions) {
+                            errors.warn("converting uninitialized array to explicit zeros because of block alignment option", srcVar.position)
+                            val zeros = PtArray(srcVar.datatype, srcVar.position)
+                            repeat(srcVar.arraysize!!.constIndex()!!) {
+                                zeros.children.add(PtNumber(ArrayToElementTypes.getValue(srcVar.datatype), 0.0, srcVar.position))
+                            }
+                            return PtVariable(srcVar.name, srcVar.datatype, srcVar.zeropage, zeros, srcVar.arraysize?.constIndex()?.toUInt(), srcVar.position)
+                        }
+                    }
+                }
+                return PtVariable(srcVar.name, srcVar.datatype, srcVar.zeropage, value, srcVar.arraysize?.constIndex()?.toUInt(), srcVar.position)
             }
-            VarDeclType.CONST -> PtConstant(srcVar.name, srcVar.datatype, (srcVar.value as NumericLiteral).number, srcVar.position)
-            VarDeclType.MEMORY -> PtMemMapped(srcVar.name, srcVar.datatype, (srcVar.value as NumericLiteral).number.toUInt(), srcVar.arraysize?.constIndex()?.toUInt(), srcVar.position)
+            VarDeclType.CONST -> return PtConstant(srcVar.name, srcVar.datatype, (srcVar.value as NumericLiteral).number, srcVar.position)
+            VarDeclType.MEMORY -> return PtMemMapped(srcVar.name, srcVar.datatype, (srcVar.value as NumericLiteral).number.toUInt(), srcVar.arraysize?.constIndex()?.toUInt(), srcVar.position)
         }
     }
 
@@ -440,6 +455,8 @@ class IntermediateAstMaker(private val program: Program) {
             addr.add(PtIdentifier(name+"_lsb", dt, src.identifier.position))        // NOTE: assumes _lsb is first in memory! (immediately followed by _msb)
         else
             addr.add(transform(src.identifier))
+        if(src.arrayIndex!=null)
+            addr.add(transformExpression(src.arrayIndex!!.indexExpr))
         return addr
     }
 

compiler/src/prog8/compiler/astprocessing/NotExpressionAndIfComparisonExprChanger.kt

@@ -1,16 +1,17 @@
 package prog8.compiler.astprocessing
 
-import prog8.ast.*
+import prog8.ast.Node
+import prog8.ast.Program
 import prog8.ast.base.FatalAstException
-import prog8.ast.expressions.*
-import prog8.ast.statements.AssignTarget
-import prog8.ast.statements.Assignment
-import prog8.ast.statements.AssignmentOrigin
-import prog8.ast.statements.IfElse
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.NumericLiteral
+import prog8.ast.expressions.PrefixExpression
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
-import prog8.code.core.*
-import prog8.code.target.VMTarget
+import prog8.code.core.DataType
+import prog8.code.core.ICompilationTarget
+import prog8.code.core.IErrorReporter
+import prog8.code.core.IntegerDatatypes
 
 internal class NotExpressionAndIfComparisonExprChanger(val program: Program, val errors: IErrorReporter, val compTarget: ICompilationTarget) : AstWalker() {
 
@@ -99,110 +100,4 @@ internal class NotExpressionAndIfComparisonExprChanger(val program: Program, val
         }
         return noModifications
     }
-
-
-    override fun before(ifElse: IfElse, parent: Node): Iterable<IAstModification> {
-        if(compTarget.name == VMTarget.NAME)  // don't apply this optimization for Vm target
-            return noModifications
-
-        val binExpr = ifElse.condition as? BinaryExpression
-        if(binExpr==null || binExpr.operator !in ComparisonOperators)
-            return noModifications
-
-        // Simplify the conditional expression, introduce simple assignments if required.
-        // NOTE: sometimes this increases code size because additional stores/loads are generated for the
-        //       intermediate variables. We assume these are optimized away from the resulting assembly code later.
-        val simplify = simplifyConditionalExpression(binExpr)
-        val modifications = mutableListOf<IAstModification>()
-        if(simplify.rightVarAssignment!=null) {
-            modifications += IAstModification.ReplaceNode(binExpr.right, simplify.rightOperandReplacement!!, binExpr)
-            modifications += IAstModification.InsertBefore(
-                ifElse,
-                simplify.rightVarAssignment,
-                parent as IStatementContainer
-            )
-        }
-        if(simplify.leftVarAssignment!=null) {
-            modifications += IAstModification.ReplaceNode(binExpr.left, simplify.leftOperandReplacement!!, binExpr)
-            modifications += IAstModification.InsertBefore(
-                ifElse,
-                simplify.leftVarAssignment,
-                parent as IStatementContainer
-            )
-        }
-
-        return modifications
-    }
-
-    private class CondExprSimplificationResult(
-        val leftVarAssignment: Assignment?,
-        val leftOperandReplacement: Expression?,
-        val rightVarAssignment: Assignment?,
-        val rightOperandReplacement: Expression?
-    )
-
-    private fun simplifyConditionalExpression(expr: BinaryExpression): CondExprSimplificationResult {
-
-        // TODO: somehow figure out if the expr will result in stack-evaluation STILL after being split off,
-        //       in that case: do *not* split it off but just keep it as it is (otherwise code size increases)
-        // NOTE: do NOT move this to an earler ast transform phase (such as StatementReorderer or StatementOptimizer) - it WILL result in larger code.
-
-        if(compTarget.name == VMTarget.NAME)  // don't apply this optimization for Vm target
-            return CondExprSimplificationResult(null, null, null, null)
-
-        val leftDt = expr.left.inferType(program)
-        val rightDt = expr.right.inferType(program)
-
-        if(!leftDt.isInteger || !rightDt.isInteger) {
-            // we can't reasonably simplify non-integer expressions
-            return CondExprSimplificationResult(null, null, null, null)
-        }
-
-        var leftAssignment: Assignment? = null
-        var leftOperandReplacement: Expression? = null
-        var rightAssignment: Assignment? = null
-        var rightOperandReplacement: Expression? = null
-        val separateLeftExpr = !expr.left.isSimple
-                && expr.left !is IFunctionCall
-                && expr.left !is ContainmentCheck
-        val separateRightExpr = !expr.right.isSimple
-                && expr.right !is IFunctionCall
-                && expr.right !is ContainmentCheck
-
-        if(separateLeftExpr) {
-            val name = getTempRegisterName(leftDt)
-            leftOperandReplacement = IdentifierReference(name, expr.position)
-            leftAssignment = Assignment(
-                AssignTarget(IdentifierReference(name, expr.position), null, null, expr.position),
-                expr.left.copy(),
-                AssignmentOrigin.ASMGEN, expr.position
-            )
-        }
-        if(separateRightExpr) {
-            val (tempVarName, _) = program.getTempVar(rightDt.getOrElse { throw FatalAstException("invalid dt") }, true)
-            rightOperandReplacement = IdentifierReference(tempVarName, expr.position)
-            rightAssignment = Assignment(
-                AssignTarget(IdentifierReference(tempVarName, expr.position), null, null, expr.position),
-                expr.right.copy(),
-                AssignmentOrigin.ASMGEN, expr.position
-            )
-        }
-        return CondExprSimplificationResult(
-            leftAssignment, leftOperandReplacement,
-            rightAssignment, rightOperandReplacement
-        )
-    }
-    
-    fun getTempRegisterName(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.BOOL -> 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")
-        }
-    }
 }