version 10.5.1

what did it even do?  in any case, $2d is just a user zero page location it should no longer be associated with the monitor nowadays.

.gitignore

@@ -1,8 +1,12 @@
 .idea/workspace.xml
 .idea/discord.xml
+.idea/developer-tools.xml
+.idea/usage.statistics.xml
+.idea/shelf/
 build/
 dist/
 output/
+out/
 .*cache/
 *.directory
 *.prg
@@ -11,7 +15,6 @@ output/
 *.vm.txt
 *.vice-mon-list
 docs/build
-out/
 parser/**/*.interp
 parser/**/*.tokens
 parser/**/*.java
@@ -22,6 +25,7 @@ compiler/src/prog8/buildversion/*
 .eggs/
 /MANIFEST
 .tox/
+.kotlin/
 __pycache__/
 parser.out
 parsetab.py
@@ -30,7 +34,6 @@ parsetab.py
 compiler/lib/
 
 .gradle
-/prog8compiler.jar
 sd*.img
 *.d64
 

.idea/kotlinc.xml

@@ -4,6 +4,6 @@
     <option name="jvmTarget" value="11" />
   </component>
   <component name="KotlinJpsPluginSettings">
-    <option name="version" value="1.9.24" />
+    <option name="version" value="2.0.21" />
   </component>
 </project>

.idea/libraries/KotlinJavaRuntime.xml

@@ -1,23 +1,23 @@
 <component name="libraryTable">
   <library name="KotlinJavaRuntime" type="repository">
-    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:2.0.20" />
+    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:2.0.21" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.20/kotlin-stdlib-jdk8-2.0.20.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.20/kotlin-stdlib-2.0.20.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.21/kotlin-stdlib-jdk8-2.0.21.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.21/kotlin-stdlib-2.0.21.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.21/kotlin-stdlib-jdk7-2.0.21.jar!/" />
     </CLASSES>
     <JAVADOC>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.20/kotlin-stdlib-jdk8-2.0.20-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.20/kotlin-stdlib-2.0.20-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.21/kotlin-stdlib-jdk8-2.0.21-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.21/kotlin-stdlib-2.0.21-javadoc.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.21/kotlin-stdlib-jdk7-2.0.21-javadoc.jar!/" />
     </JAVADOC>
     <SOURCES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.20/kotlin-stdlib-jdk8-2.0.20-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.20/kotlin-stdlib-2.0.20-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.0.21/kotlin-stdlib-jdk8-2.0.21-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.0.21/kotlin-stdlib-2.0.21-sources.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.21/kotlin-stdlib-jdk7-2.0.21-sources.jar!/" />
     </SOURCES>
   </library>
 </component>

.idea/libraries/eclipse_lsp4j.xml

@@ -0,0 +1,13 @@
+<component name="libraryTable">
+  <library name="eclipse.lsp4j" type="repository">
+    <properties maven-id="org.eclipse.lsp4j:org.eclipse.lsp4j:0.23.1" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/org/eclipse/lsp4j/org.eclipse.lsp4j/0.23.1/org.eclipse.lsp4j-0.23.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/eclipse/lsp4j/org.eclipse.lsp4j.jsonrpc/0.23.1/org.eclipse.lsp4j.jsonrpc-0.23.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/google/code/gson/gson/2.11.0/gson-2.11.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/com/google/errorprone/error_prone_annotations/2.27.0/error_prone_annotations-2.27.0.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/io_kotest_framework_datatest.xml

@@ -0,0 +1,23 @@
+<component name="libraryTable">
+  <library name="io.kotest.framework.datatest" type="repository">
+    <properties maven-id="io.kotest:kotest-framework-datatest:5.9.1" />
+    <CLASSES>
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-datatest/5.9.1/kotest-framework-datatest-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-datatest-jvm/5.9.1/kotest-framework-datatest-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.23/kotlin-stdlib-1.9.23.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-core-jvm/1.8.0/kotlinx-coroutines-core-jvm-1.8.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.9.1/kotest-common-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-reflect/1.9.23/kotlin-reflect-1.9.23.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-api-jvm/5.9.1/kotest-framework-api-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.9.1/kotest-assertions-shared-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.9.1/kotest-assertions-api-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-jdk8/1.8.0/kotlinx-coroutines-jdk8-1.8.0.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/1.3.0/opentest4j-1.3.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-test-jvm/1.8.0/kotlinx-coroutines-test-jvm-1.8.0.jar!/" />
+    </CLASSES>
+    <JAVADOC />
+    <SOURCES />
+  </library>
+</component>

.idea/libraries/michael_bull_kotlin_result_jvm.xml

@@ -6,7 +6,15 @@
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.22/kotlin-stdlib-1.9.22.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />
     </CLASSES>
-    <JAVADOC />
-    <SOURCES />
+    <JAVADOC>
+      <root url="jar://$MAVEN_REPOSITORY$/com/michael-bull/kotlin-result/kotlin-result-jvm/2.0.0/kotlin-result-jvm-2.0.0-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.22/kotlin-stdlib-1.9.22-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-javadoc.jar!/" />
+    </JAVADOC>
+    <SOURCES>
+      <root url="jar://$MAVEN_REPOSITORY$/com/michael-bull/kotlin-result/kotlin-result-jvm/2.0.0/kotlin-result-jvm-2.0.0-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.22/kotlin-stdlib-1.9.22-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-sources.jar!/" />
+    </SOURCES>
   </library>
 </component>

.idea/modules.xml

@@ -2,6 +2,8 @@
 <project version="4">
   <component name="ProjectModuleManager">
     <modules>
+      <module fileurl="file://$PROJECT_DIR$/beanshell/beanshell.iml" filepath="$PROJECT_DIR$/beanshell/beanshell.iml" />
+      <module fileurl="file://$PROJECT_DIR$/benchmark-program/benchmark-program.iml" filepath="$PROJECT_DIR$/benchmark-program/benchmark-program.iml" />
       <module fileurl="file://$PROJECT_DIR$/codeCore/codeCore.iml" filepath="$PROJECT_DIR$/codeCore/codeCore.iml" />
       <module fileurl="file://$PROJECT_DIR$/codeGenCpu6502/codeGenCpu6502.iml" filepath="$PROJECT_DIR$/codeGenCpu6502/codeGenCpu6502.iml" />
       <module fileurl="file://$PROJECT_DIR$/codeGenExperimental/codeGenExperimental.iml" filepath="$PROJECT_DIR$/codeGenExperimental/codeGenExperimental.iml" />
@@ -14,6 +16,7 @@
       <module fileurl="file://$PROJECT_DIR$/examples/examples.iml" filepath="$PROJECT_DIR$/examples/examples.iml" />
       <module fileurl="file://$PROJECT_DIR$/intermediate/intermediate.iml" filepath="$PROJECT_DIR$/intermediate/intermediate.iml" />
       <module fileurl="file://$PROJECT_DIR$/parser/parser.iml" filepath="$PROJECT_DIR$/parser/parser.iml" />
+      <module fileurl="file://$PROJECT_DIR$/.idea/modules/prog8.iml" filepath="$PROJECT_DIR$/.idea/modules/prog8.iml" />
       <module fileurl="file://$PROJECT_DIR$/virtualmachine/virtualmachine.iml" filepath="$PROJECT_DIR$/virtualmachine/virtualmachine.iml" />
     </modules>
   </component>

Makefile

@@ -4,9 +4,9 @@
 
 all:
 	gradle installdist installshadowdist
-	@echo "compiler launch script can be found here: compiler/build/install/compiler-shadow/bin/p8compile"
+	@echo "compiler launch script can be found here: compiler/build/install/compiler-shadow/bin/prog8c"
 
 test:
 	gradle build
-	@echo "compiler launch script can be found here: compiler/build/install/compiler-shadow/bin/p8compile"
+	@echo "compiler launch script can be found here: compiler/build/install/compiler-shadow/bin/prog8c"
 

README.md

@@ -29,6 +29,8 @@ How to get it/build it
 - Or, if you want/need a bleeding edge development version, you can:
   - download a build artifact zipfile from a recent [github action build](https://github.com/irmen/prog8/actions).
   - you can also compile it yourself from source. [Instructions here](https://prog8.readthedocs.io/en/latest/compiling.html).
+    Note that if you are not using *gradle* to build it, you might have to perform some manual
+    tasks once to make it compile fully. These are explained in the linked instructions. 
 - Alternatively, you can also install the compiler as a package on some linux distros:
   - Arch (via AUR): [`prog8`](https://aur.archlinux.org/packages/prog8)
 
@@ -52,27 +54,26 @@ What does Prog8 provide?
 ------------------------
 
 - all advantages of a higher level language over having to write assembly code manually
-- programs run very fast because compilation to native machine code
+- programs run very fast because it's compiled to native machine code
 - code often is smaller and faster than equivalent C code compiled with CC65 or even LLVM-MOS
-- modularity, symbol scoping, subroutines
+- modularity, symbol scoping, subroutines. No need for forward declarations.
 - various data types other than just bytes (16-bit words, floats, strings)
-- floating point math is supported if the target system provides floating point library routines (C64 and Cx16 both do)
+- floating point math is supported on certain targets
 - strings can contain escaped characters but also many symbols directly if they have a petscii equivalent, such as "♠♥♣♦π▚●○╳". Characters like ^, _, \, {, } and | are also accepted and converted to the closest petscii equivalents.
 - automatic static variable allocations, automatic string and array variables and string sharing
-- subroutines with input parameters and result values
 - high-level program optimizations
-- no need for forward declarations
-- small program boilerplate/compilersupport overhead
 - programs can be run multiple times without reloading because of automatic variable (re)initializations.
-- conditional branches
+- conditional branches that map 1:1 to cpu status flags
 - ``when`` statement to provide a concise jump table alternative to if/elseif chains
 - ``in`` expression for concise and efficient multi-value/containment check 
+- ``defer`` statement to help write concise and robust subroutine cleanup logic
 - several specialized built-in functions such as ``lsb``, ``msb``, ``min``, ``max``, ``rol``, ``ror``
 - various powerful built-in libraries to do I/O, number conversions, graphics and more  
-- convenience abstractions for low level aspects such as ZeroPage handling, program startup, explicit memory addresses
 - inline assembly allows you to have full control when every cycle or byte matters
-- supports the sixteen 'virtual' 16-bit registers R0 - R15 from the Commander X16, and provides them also on the C64.
-- encode strings and characters into petscii or screencodes or even other encodings, as desired (C64/Cx16)
+- supports the sixteen 'virtual' 16-bit registers R0 - R15 from the Commander X16 (also available on other targets)
+- encode strings and characters into petscii or screencodes or even other encodings
+- Automatic ROM/RAM bank switching on certain compiler targets when calling routines in other banks
+- 50 Kb of available program RAM size on the C64 by default; because Basic ROM is banked out altogether 
 
 *Rapid edit-compile-run-debug cycle:*
 

beanshell/beanshell.iml

@@ -0,0 +1,24 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="JAVA_MODULE" version="4">
+  <component name="NewModuleRootManager" inherit-compiler-output="true">
+    <exclude-output />
+    <content url="file://$MODULE_DIR$">
+      <sourceFolder url="file://$MODULE_DIR$/src" isTestSource="false" />
+      <excludeFolder url="file://$MODULE_DIR$/build" />
+    </content>
+    <orderEntry type="inheritedJdk" />
+    <orderEntry type="sourceFolder" forTests="false" />
+    <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
+    <orderEntry type="library" name="jetbrains.kotlinx.cli.jvm" level="project" />
+    <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
+    <orderEntry type="module-library">
+      <library>
+        <CLASSES>
+          <root url="jar://$MODULE_DIR$/lib/bsh-3.0.0-SNAPSHOT.jar!/" />
+        </CLASSES>
+        <JAVADOC />
+        <SOURCES />
+      </library>
+    </orderEntry>
+  </component>
+</module>

beanshell/build.gradle.kts

@@ -0,0 +1,91 @@
+import org.jetbrains.kotlin.gradle.dsl.JvmTarget
+
+plugins {
+    kotlin("jvm")
+    id("application")
+}
+
+val serverMainClassName = "prog8lsp.MainKt"
+val applicationName = "prog8-beanshell"
+val javaVersion: String by project
+
+application {
+    mainClass.set(serverMainClassName)
+    description = "Code completions, diagnostics and more for Prog8"
+    // applicationDefaultJvmArgs = listOf("-DkotlinLanguageServer.version=$version")
+    applicationDistribution.into("bin") {
+        filePermissions {
+            user {
+                read=true
+                execute=true
+                write=true
+            }
+            other.execute = true
+            group.execute = true
+        }
+    }
+}
+
+repositories {
+    mavenCentral()
+}
+
+dependencies {
+    implementation(files("lib/bsh-3.0.0-SNAPSHOT.jar"))
+}
+
+configurations.forEach { config ->
+    config.resolutionStrategy {
+        preferProjectModules()
+    }
+}
+
+sourceSets.main {
+    java.srcDir("src")
+    resources.srcDir("resources")
+}
+
+sourceSets.test {
+    java.srcDir("src")
+    resources.srcDir("resources")
+}
+
+tasks.startScripts {
+    applicationName = "prog8-beanshell"
+}
+
+tasks.register<Exec>("fixFilePermissions") {
+    // When running on macOS or Linux the start script
+    // needs executable permissions to run.
+
+    onlyIf { !System.getProperty("os.name").lowercase().contains("windows") }
+    commandLine("chmod", "+x", "${tasks.installDist.get().destinationDir}/bin/prog8-beanshell")
+}
+
+tasks.withType<Test>() {
+    testLogging {
+        events("failed")
+        exceptionFormat = org.gradle.api.tasks.testing.logging.TestExceptionFormat.FULL
+    }
+}
+
+tasks.installDist {
+    finalizedBy("fixFilePermissions")
+}
+
+tasks.build {
+    finalizedBy("installDist")
+}
+
+java {
+    targetCompatibility = JavaVersion.VERSION_11
+    sourceCompatibility = JavaVersion.VERSION_11
+}
+
+kotlin {
+    compilerOptions.jvmTarget = JvmTarget.JVM_11
+//    jvmToolchain {
+//        languageVersion = JavaLanguageVersion.of(javaVersion.toInt())
+//    }
+}
+

beanshell/src/prog8beanshell/CommandLineReader.kt

@@ -0,0 +1,48 @@
+package prog8beanshell
+
+import java.io.FilterReader
+import java.io.Reader
+
+
+class CommandLineReader(val input: Reader): FilterReader(input)  {
+    private val normal = 0
+    private val lastCharNL = 1
+    private val sentSemi = 2
+    private var state = lastCharNL
+
+    override fun read(): Int {
+        if (state == sentSemi) {
+            this.state = lastCharNL
+            return 10
+        } else {
+            var b = input.read()
+            while(b==13) b = input.read()
+
+            if (b == 10) {
+                if (this.state == lastCharNL) {
+                    b = 59
+                    this.state = sentSemi
+                } else {
+                    this.state = lastCharNL
+                }
+            } else {
+                this.state = normal
+            }
+
+            return b
+        }
+
+    }
+
+    override fun read(buff: CharArray, off: Int, len: Int): Int {
+        val b = read()
+        if (b == -1) {
+            return -1
+        } else {
+            buff[off] = b.toChar()
+            return 1
+        }
+
+    }
+}
+

beanshell/src/prog8beanshell/Main.kt

@@ -0,0 +1,23 @@
+package prog8beanshell
+
+import bsh.FileReader
+import bsh.Interpreter
+
+
+class BeanshellInterpreter {
+
+    fun run(symbols: Map<String, Any>) {
+        val interpreter = Interpreter(CommandLineReader(FileReader(System.`in`)), System.out, System.err, true)
+        interpreter.setExitOnEOF(false)
+        symbols.forEach { (name, value) -> interpreter.set(name, value) }
+        interpreter.run()
+    }
+}
+
+fun main(args: Array<String>) {
+    val i = BeanshellInterpreter()
+    i.run(mapOf(
+        "env" to System.getenv(),
+        "args" to args
+    ))
+}

benchmark-program/Makefile

@@ -0,0 +1,10 @@
+.PHONY:  clean run
+
+run:
+	prog8c -target cx16 benchmark.p8
+	x16emu -run -prg benchmark.prg -warp
+
+clean:
+	rm -f *.prg *.PRG *.asm *.vice-* *.BIN *.PAL *.zip *.7z
+
+

benchmark-program/b_3d.p8

@@ -0,0 +1,109 @@
+%import textio
+%import math
+
+rotate3d {
+    const ubyte WIDTH = 40
+    const ubyte HEIGHT = 30
+
+    sub benchmark(uword max_time) -> uword {
+
+        uword anglex
+        uword angley
+        uword anglez
+        uword frames
+
+        txt.nl()
+        cbm.SETTIM(0,0,0)
+
+        while cbm.RDTIM16()<max_time {
+            matrix_math.rotate_vertices(msb(anglex), msb(angley), msb(anglez))
+            draw_edges()   ; doesn't really draw anything in the benchmark, but does do the screen calculations
+            anglex+=500
+            angley+=215
+            anglez+=453
+            frames++
+        }
+
+        return frames
+    }
+
+    sub draw_edges() {
+
+        ; plot the points of the 3d cube
+        ; first the points on the back, then the points on the front (painter algorithm)
+
+        ubyte @zp i
+        word @zp rz
+        word @zp persp
+        byte @shared sx
+        byte @shared sy
+
+        for i in 0 to len(matrix_math.xcoor)-1 {
+            rz = matrix_math.rotatedz[i]
+            if rz >= 10 {
+                persp = 600 + rz/64
+                sx = matrix_math.rotatedx[i] / persp as byte + WIDTH/2
+                sy = matrix_math.rotatedy[i] / persp as byte + HEIGHT/2
+                ;; txt.setcc(sx as ubyte, sy as ubyte, 46, 7)
+            }
+        }
+
+        for i in 0 to len(matrix_math.xcoor)-1 {
+            rz = matrix_math.rotatedz[i]
+            if rz < 10 {
+                persp = 600 + rz/64
+                sx = matrix_math.rotatedx[i] / persp as byte + WIDTH/2
+                sy = matrix_math.rotatedy[i] / persp as byte + HEIGHT/2
+                ;; txt.setcc(sx as ubyte, sy as ubyte, 81, 7)
+            }
+        }
+
+        txt.chrout('.')
+    }
+}
+
+matrix_math {
+    ; vertices
+    word[] @split xcoor = [ -40, -40, -40, -40,  40,  40,  40, 40 ]
+    word[] @split ycoor = [ -40, -40,  40,  40, -40, -40,  40, 40 ]
+    word[] @split zcoor = [ -40,  40, -40,  40, -40,  40, -40, 40 ]
+
+    ; storage for rotated coordinates
+    word[len(xcoor)] @split rotatedx
+    word[len(ycoor)] @split rotatedy
+    word[len(zcoor)] @split rotatedz
+
+    sub rotate_vertices(ubyte ax, ubyte ay, ubyte az) {
+        ; rotate around origin (0,0,0)
+
+        ; set up the 3d rotation matrix values
+        word wcosa = math.cos8(ax)
+        word wsina = math.sin8(ax)
+        word wcosb = math.cos8(ay)
+        word wsinb = math.sin8(ay)
+        word wcosc = math.cos8(az)
+        word wsinc = math.sin8(az)
+
+        word wcosa_sinb = wcosa*wsinb / 128
+        word wsina_sinb = wsina*wsinb / 128
+
+        word Axx = wcosa*wcosb / 128
+        word Axy = (wcosa_sinb*wsinc - wsina*wcosc) / 128
+        word Axz = (wcosa_sinb*wcosc + wsina*wsinc) / 128
+        word Ayx = wsina*wcosb / 128
+        word Ayy = (wsina_sinb*wsinc + wcosa*wcosc) / 128
+        word Ayz = (wsina_sinb*wcosc - wcosa*wsinc) / 128
+        word Azx = -wsinb
+        word Azy = wcosb*wsinc / 128
+        word Azz = wcosb*wcosc / 128
+
+        ubyte @zp i
+        for i in 0 to len(xcoor)-1 {
+            ; don't normalize by dividing by 128, instead keep some precision for perspective calc later
+            rotatedx[i] = Axx*xcoor[i] + Axy*ycoor[i] + Axz*zcoor[i]
+            rotatedy[i] = Ayx*xcoor[i] + Ayy*ycoor[i] + Ayz*zcoor[i]
+            rotatedz[i] = Azx*xcoor[i] + Azy*ycoor[i] + Azz*zcoor[i]
+        }
+    }
+
+}

benchmark-program/b_adpcm.p8

@@ -0,0 +1,89 @@
+adpcm {
+
+    sub decode_benchmark(uword max_time) -> uword {
+        uword num_blocks
+        txt.nl()
+        cbm.SETTIM(0,0,0)
+
+        while cbm.RDTIM16()<max_time {
+            adpcm.init(0,0)
+            uword @requirezp nibbles_ptr = $a000 ; for benchmark purposes, the exact nibbles don't really matter, so we just take the basic ROM as input
+            repeat 252/2 {
+                unroll 2 {
+                    ubyte @zp nibble = @(nibbles_ptr)
+                    adpcm.decode_nibble(nibble & 15)     ; first word  (note: upper nibble needs to be zero!)
+                    adpcm.decode_nibble(nibble>>4)       ; second word  (note: upper nibble is zero, after the shifts.)
+                    nibbles_ptr++
+                }
+            }
+            num_blocks++
+            txt.chrout('.')
+        }
+
+        return num_blocks
+    }
+
+    ; IMA ADPCM decoder.  Supports mono and stereo streams.
+
+    ubyte[] t_index = [ -1, -1, -1, -1, 2, 4, 6, 8, -1, -1, -1, -1, 2, 4, 6, 8]
+    uword[] @split t_step = [
+            7, 8, 9, 10, 11, 12, 13, 14,
+            16, 17, 19, 21, 23, 25, 28, 31,
+            34, 37, 41, 45, 50, 55, 60, 66,
+            73, 80, 88, 97, 107, 118, 130, 143,
+            157, 173, 190, 209, 230, 253, 279, 307,
+            337, 371, 408, 449, 494, 544, 598, 658,
+            724, 796, 876, 963, 1060, 1166, 1282, 1411,
+            1552, 1707, 1878, 2066, 2272, 2499, 2749, 3024,
+            3327, 3660, 4026, 4428, 4871, 5358, 5894, 6484,
+            7132, 7845, 8630, 9493, 10442, 11487, 12635, 13899,
+            15289, 16818, 18500, 20350, 22385, 24623, 27086, 29794,
+            32767]
+
+    uword @requirezp predict       ; decoded 16 bit pcm sample for first channel.
+    ubyte @requirezp index
+    uword @requirezp pstep
+
+    sub init(uword startPredict, ubyte startIndex) {
+        ; initialize first decoding channel.
+        predict = startPredict
+        index = startIndex
+        pstep = t_step[index]
+    }
+
+    sub decode_nibble(ubyte @zp nibble) {
+        ; Decoder for a single nibble for the first channel. (value of 'nibble' needs to be strictly 0-15 !)
+        ; This is the hotspot of the decoder algorithm!
+        ; Note that the generated assembly from this is pretty efficient,
+        ; rewriting it by hand in asm seems to improve it only ~10%.
+        cx16.r0s = 0                ; difference
+        if nibble & %0100 !=0
+            cx16.r0s += pstep
+        pstep >>= 1
+        if nibble & %0010 !=0
+            cx16.r0s += pstep
+        pstep >>= 1
+        if nibble & %0001 !=0
+            cx16.r0s += pstep
+        pstep >>= 1
+        cx16.r0s += pstep
+        if nibble & %1000 !=0
+            predict -= cx16.r0
+        else
+            predict += cx16.r0
+
+        ; NOTE: the original C/Python code uses a 32 bits prediction value and clips it to a 16 bit word
+        ;       but for speed reasons we only work with 16 bit words here all the time (with possible clipping error)
+        ; if predicted > 32767:
+        ;    predicted = 32767
+        ; elif predicted < -32767:
+        ;    predicted = - 32767
+
+        index += t_index[nibble]
+        if_neg
+            index = 0
+        else if index >= len(t_step)-1
+            index = len(t_step)-1
+        pstep = t_step[index]
+    }
+}

benchmark-program/b_circles.p8

@@ -0,0 +1,111 @@
+%import gfx_lores
+%import math
+
+circles {
+    const ubyte MAX_NUM_CIRCLES = 80
+    const ubyte GROWTH_RATE = 4
+    uword[MAX_NUM_CIRCLES] @split circle_x
+    uword[MAX_NUM_CIRCLES] @split circle_y
+    ubyte[MAX_NUM_CIRCLES] circle_radius
+    ubyte color
+    uword total_num_circles
+
+    sub draw(bool use_kernal, uword max_time) -> uword {
+        if use_kernal
+            void cx16.set_screen_mode(128)
+        else
+            gfx_lores.graphics_mode()
+
+        math.rndseed(12345,6789)
+        cbm.SETTIM(0,0,0)
+
+        total_num_circles = 0
+        color = 16
+
+        while cbm.RDTIM16()<max_time {
+            if use_kernal {
+                cx16.GRAPH_set_colors(0,0,0)
+                cx16.GRAPH_clear()
+            }
+            else
+                gfx_lores.clear_screen(0)
+            total_num_circles += draw_circles(use_kernal, max_time)
+        }
+
+        if use_kernal
+            void cx16.set_screen_mode(3)
+        else {
+            gfx_lores.text_mode()
+        }
+
+        return total_num_circles
+    }
+
+    sub draw_circles(bool use_kernal, uword max_time) -> uword {
+        uword @zp x
+        uword @zp y
+        ubyte @zp radius
+
+        ubyte num_circles
+
+        while num_circles<MAX_NUM_CIRCLES and cbm.RDTIM16()<max_time {
+            x = math.rndw() % 320
+            y = math.rndw() % 240
+            radius = GROWTH_RATE
+            if not_colliding() {
+                while not_edge() and not_colliding() {
+                    radius += GROWTH_RATE
+                }
+                radius -= GROWTH_RATE
+                if radius>0 {
+                    color++
+                    if color==0
+                        color=16
+                    if use_kernal {
+                        cx16.GRAPH_set_colors(color, 255-color, 0)
+                        cx16.GRAPH_draw_oval(x-radius, y-radius, radius*2, radius*2, true)
+                    }
+                    else
+                        gfx_lores.disc(x, y as ubyte, radius, color)
+                    circle_x[num_circles] = x
+                    circle_y[num_circles] = y
+                    circle_radius[num_circles] = radius
+                    num_circles++
+                }
+            }
+        }
+
+        return num_circles
+
+        sub not_colliding() -> bool {
+            if num_circles==0
+                return true
+            ubyte @zp c
+            for c in 0 to num_circles-1 {
+                if distance(c) < (radius as uword) + circle_radius[c]
+                    return false
+            }
+            return true
+        }
+
+        sub distance(ubyte cix) -> uword {
+            word dx = x as word - circle_x[cix]
+            word dy = y as word - circle_y[cix]
+            uword sqx = dx*dx as uword
+            uword sqy = dy*dy as uword
+            return sqrt(sqx + sqy)
+        }
+
+        sub not_edge() -> bool {
+            if x as word - radius < 0
+                return false
+            if x + radius >= 320
+                return false
+            if y as word - radius < 0
+                return false
+            if y + radius >= 240
+                return false
+            return true
+        }
+    }
+}

benchmark-program/b_life.p8

@@ -0,0 +1,123 @@
+; conway's game of life.
+
+%import math
+%import textio
+
+life {
+    const ubyte WIDTH = 40
+    const ubyte HEIGHT = 30
+    const uword STRIDE = $0002+WIDTH
+    uword world1 = memory("world1", (WIDTH+2)*(HEIGHT+2), 0)
+    uword world2 = memory("world2", (WIDTH+2)*(HEIGHT+2), 0)
+    uword @requirezp active_world = world1
+
+   sub benchmark(uword max_time) -> uword {
+        txt.clear_screen()
+        sys.memset(world1, (WIDTH+2)*(HEIGHT+2), 0)
+        sys.memset(world2, (WIDTH+2)*(HEIGHT+2), 0)
+
+        set_start_gen()
+
+        uword gen
+        cbm.SETTIM(0,0,0)
+
+        while cbm.RDTIM16()<max_time {
+            next_gen()
+            gen++
+        }
+
+        return gen
+    }
+
+    sub set_start_gen() {
+
+; some way to set a custom start generation:
+;        str start_gen = "                " +
+;                        "                " +
+;                        "                " +
+;                        "          **    " +
+;                        "        *    *  " +
+;                        "       *        " +
+;                        "       *     *  " +
+;                        "       ******   " +
+;                        "                " +
+;                        "                " +
+;                        "                " +
+;                        "                " +
+;                        "                " +
+;                        "                " +
+;                        "                " +
+;                        "               "
+;
+;        for y in 0 to 15 {
+;            for x in 0 to 15 {
+;                if start_gen[y*16 + x]=='*'
+;                    active_world[offset + x] = 1
+;            }
+;            offset += STRIDE
+;        }
+
+        ; randomize whole world
+        math.rndseed(12345,9999)
+        uword offset = STRIDE+1
+        ubyte x
+        ubyte y
+        for y in 0 to HEIGHT-1 {
+            for x in 0 to WIDTH-1 {
+                active_world[offset+x] = math.rnd() & 1
+            }
+            offset += STRIDE
+        }
+    }
+
+    sub next_gen() {
+        const ubyte DXOFFSET = 0
+        const ubyte DYOFFSET = 0
+        ubyte[2] cell_chars = [sc:' ', sc:'●']
+
+        uword @requirezp new_world = world1
+        if active_world == world1
+            new_world = world2
+
+        ; To avoid re-calculating word index lookups into the new- and active world arrays,
+        ; we calculate the required pointer values upfront.
+        ; Inside the loop we can use ptr+x just fine (results in efficient LDA (ptr),Y instruction because x is a byte type),
+        ; and for each row we simply add the stride to the pointer.
+        ; It's more readable to use active_world[offset] etc, but offset is a word value, and this produces
+        ; inefficient assembly code because we can't use a register indexed mode in this case. Costly inside a loop.
+
+        uword @requirezp new_world_ptr = new_world + STRIDE+1-DXOFFSET
+        uword @requirezp active_world_ptr = active_world + STRIDE+1-DXOFFSET
+
+        ubyte x
+        ubyte y
+        for y in DYOFFSET to HEIGHT+DYOFFSET-1 {
+
+            cx16.vaddr_autoincr(1, $b000 + 256*y, 0, 2)     ;  allows us to use simple Vera data byte assigns later instead of setchr() calls
+
+            for x in DXOFFSET to WIDTH+DXOFFSET-1 {
+                ; count the living neighbors
+                ubyte cell = @(active_world_ptr + x)
+                uword @requirezp ptr = active_world_ptr + x - STRIDE - 1
+                ubyte neighbors = @(ptr) + @(ptr+1) + @(ptr+2) +
+                                  @(ptr+STRIDE) + cell + @(ptr+STRIDE+2) +
+                                  @(ptr+STRIDE*2) + @(ptr+STRIDE*2+1) + @(ptr+STRIDE*2+2)
+
+                ; apply game of life rules
+                if neighbors==3
+                    cell=1
+                else if neighbors!=4
+                    cell=0
+                @(new_world_ptr + x) = cell
+
+                ; draw new cell
+                ; txt.setchr(x,y,cell_chars[cell])
+                cx16.VERA_DATA0 = cell_chars[cell]
+            }
+            active_world_ptr += STRIDE
+            new_world_ptr += STRIDE
+        }
+
+        active_world = new_world
+    }
+}

benchmark-program/b_mandelbrot.p8

@@ -0,0 +1,54 @@
+%import textio
+%import floats
+
+mandelbrot {
+    const ubyte width = 39
+    const ubyte height = 29
+    const ubyte max_iter = 15
+
+    sub calc(uword max_time) -> uword  {
+        uword num_pixels
+        ubyte pixelx
+        ubyte pixely
+
+        txt.home()
+        cbm.SETTIM(0,0,0)
+
+        while cbm.RDTIM16() < max_time {
+            for pixely in 0 to height-1 {
+                float yy = (pixely as float)/0.40/height - 1.3
+
+                for pixelx in 0 to width-1 {
+                    float xx = (pixelx as float)/0.32/width - 2.2
+
+                    float xsquared = 0.0
+                    float ysquared = 0.0
+                    float x = 0.0
+                    float y = 0.0
+                    ubyte iter = 0
+
+                    while iter<max_iter and xsquared+ysquared<4.0 {
+                        y = x*y*2.0 + yy
+                        x = xsquared - ysquared + xx
+                        xsquared = x*x
+                        ysquared = y*y
+                        iter++
+                    }
+                    txt.color2(1, max_iter-iter)
+                    txt.spc()
+                    num_pixels++
+
+                    if cbm.RDTIM16()>=max_time
+                        goto finished
+                }
+                txt.nl()
+            }
+
+            txt.clear_screen()
+        }
+
+finished:
+        txt.color2(1, 6)
+        return num_pixels
+    }
+}

benchmark-program/b_maze.p8

@@ -0,0 +1,343 @@
+%import textio
+%import math
+
+; Even though prog8 only has support for extremely limited recursion,
+; you can write recursive algorithms with a bit of extra work by building your own explicit stack structure.
+; This program shows a depth-first maze generation algorithm (1 possible path from start to finish),
+; and a depth-first maze solver algorithm, both using a stack to store the path taken.
+
+; Note: this program can be compiled for multiple target systems.
+
+maze {
+    uword score
+
+    sub bench(uword max_time) -> uword {
+        txt.nl()
+        score=0
+        math.rndseed(2345,44332)
+        cbm.SETTIM(0,0,0)
+        while cbm.RDTIM16()<max_time {
+            maze.initialize()
+            maze.drawStartFinish()
+            if maze.generate(max_time) {
+                maze.openpassages()
+                maze.drawStartFinish()
+                if maze.solve(max_time) {
+                    maze.drawStartFinish()
+                } else break
+            } else break
+        }
+
+        return score
+    }
+
+    const uword screenwidth = 40
+    const uword screenheight = 30
+
+    const ubyte numCellsHoriz = (screenwidth-1) / 2
+    const ubyte numCellsVert = (screenheight-1) / 2
+
+    ; maze start and finish cells
+    const ubyte startCx = 0
+    const ubyte startCy = 0
+    const ubyte finishCx = numCellsHoriz-1
+    const ubyte finishCy = numCellsVert-1
+
+    ; cell properties
+    const ubyte STONE = 128
+    const ubyte WALKED = 64
+    const ubyte BACKTRACKED = 32
+    const ubyte UP = 1
+    const ubyte RIGHT = 2
+    const ubyte DOWN = 4
+    const ubyte LEFT = 8
+    const ubyte WALLCOLOR = 12
+    const ubyte EMPTYCOLOR = 0
+
+    ; unfortunately on larger screens (cx16), the number of cells exceeds 256 and doesn't fit in a regular array anymore.
+    uword cells = memory("cells", numCellsHoriz*numCellsVert, 0)
+
+    ubyte[256] cx_stack
+    ubyte[256] cy_stack
+    ubyte stackptr
+
+    ubyte[4] directionflags = [LEFT,RIGHT,UP,DOWN]
+
+    sub generate(uword max_time) -> bool {
+        ubyte cx = startCx
+        ubyte cy = startCy
+
+        stackptr = 0
+        @(celladdr(cx,cy)) &= ~STONE
+        drawCell(cx, cy)
+        uword cells_to_carve = numCellsHoriz * numCellsVert - 1
+
+        while cbm.RDTIM16()<max_time {
+carve_restart_after_repath:
+            ubyte direction = choose_uncarved_direction()
+            if direction==0 {
+                ;backtrack
+                stackptr--
+                if stackptr==255 {
+                    ; stack empty.
+                    ; repath if we are not done yet. (this is a workaround for the prog8 256 array lenght limit)
+                    if cells_to_carve!=0 {
+                        if repath()
+                            goto carve_restart_after_repath
+                    }
+                    return true
+                }
+                cx = cx_stack[stackptr]
+                cy = cy_stack[stackptr]
+            } else {
+                cx_stack[stackptr] = cx
+                cy_stack[stackptr] = cy
+                stackptr++
+                if stackptr==0 {
+                    ; stack overflow, we can't track our path any longer.
+                    ; repath if we are not done yet. (this is a workaround for the prog8 256 array lenght limit)
+                    if cells_to_carve!=0 {
+                        if repath()
+                            goto carve_restart_after_repath
+                    }
+                    return true
+                }
+                @(celladdr(cx,cy)) |= direction
+                when direction {
+                    UP -> {
+                        cy--
+                        @(celladdr(cx,cy)) |= DOWN
+                    }
+                    RIGHT -> {
+                        cx++
+                        @(celladdr(cx,cy)) |= LEFT
+
+                        score++
+                    }
+                    DOWN -> {
+                        cy++
+                        @(celladdr(cx,cy)) |= UP
+                    }
+                    LEFT -> {
+                        cx--
+                        @(celladdr(cx,cy)) |= RIGHT
+                    }
+                }
+                @(celladdr(cx,cy)) &= ~STONE
+                cells_to_carve--
+                drawCell(cx, cy)
+            }
+        }
+        return false
+
+        sub repath() -> bool {
+            ; repath: try to find a new start cell with possible directions.
+            ; we limit our number of searches so that the algorith doesn't get stuck
+            ; for too long on bad rng... just accept a few unused cells in that case.
+            repeat 255 {
+                do {
+                    cx = math.rnd() % numCellsHoriz
+                    cy = math.rnd() % numCellsVert
+                } until @(celladdr(cx, cy)) & STONE ==0
+                if available_uncarved()!=0
+                    return true
+            }
+            return false
+        }
+
+        sub available_uncarved() -> ubyte {
+            ubyte candidates = 0
+            if cx>0 and @(celladdr(cx-1, cy)) & STONE !=0
+                candidates |= LEFT
+            if cx<numCellsHoriz-1 and @(celladdr(cx+1, cy)) & STONE !=0
+                candidates |= RIGHT
+            if cy>0 and @(celladdr(cx, cy-1)) & STONE !=0
+                candidates |= UP
+            if cy<numCellsVert-1 and @(celladdr(cx, cy+1)) & STONE !=0
+                candidates |= DOWN
+            return candidates
+        }
+
+        sub choose_uncarved_direction() -> ubyte {
+            ubyte candidates =  available_uncarved()
+            if candidates==0
+                return 0
+
+            repeat {
+                ubyte choice = candidates & directionflags[math.rnd() & 3]
+                if choice!=0
+                    return choice
+            }
+        }
+    }
+
+    sub openpassages() {
+        ; open just a few extra passages, so that multiple routes are possible in theory.
+        ubyte numpassages
+        ubyte cx
+        ubyte cy
+        do {
+            do {
+                cx = math.rnd() % (numCellsHoriz-2) + 1
+                cy = math.rnd() % (numCellsVert-2) + 1
+            } until @(celladdr(cx, cy)) & STONE ==0
+            ubyte direction = directionflags[math.rnd() & 3]
+            if @(celladdr(cx, cy)) & direction == 0 {
+                when direction {
+                    LEFT -> {
+                        if @(celladdr(cx-1,cy)) & STONE == 0 {
+                            @(celladdr(cx,cy)) |= LEFT
+                            drawCell(cx,cy)
+                            numpassages++
+                        }
+                    }
+                    RIGHT -> {
+                        if @(celladdr(cx+1,cy)) & STONE == 0 {
+                            @(celladdr(cx,cy)) |= RIGHT
+                            drawCell(cx,cy)
+                            numpassages++
+                        }
+                    }
+                    UP -> {
+                        if @(celladdr(cx,cy-1)) & STONE == 0 {
+                            @(celladdr(cx,cy)) |= UP
+                            drawCell(cx,cy)
+                            numpassages++
+                        }
+                    }
+                    DOWN -> {
+                        if @(celladdr(cx,cy+1)) & STONE == 0 {
+                            @(celladdr(cx,cy)) |= DOWN
+                            drawCell(cx,cy)
+                            numpassages++
+                        }
+                    }
+                }
+            }
+        } until numpassages==10
+    }
+
+    sub solve(uword max_time) -> bool {
+        ubyte cx = startCx
+        ubyte cy = startCy
+        const uword max_path_length = 1024
+
+        ; the path through the maze can be longer than 256 so doesn't fit in a regular array.... :(
+        uword pathstack = memory("pathstack", max_path_length, 0)
+        uword pathstackptr = 0
+
+        @(celladdr(cx,cy)) |= WALKED
+        ; txt.setcc(cx*2+1, cy*2+1, 81, 1)
+
+        while cbm.RDTIM16()<max_time {
+solve_loop:
+            if cx==finishCx and cy==finishCy {
+                ;txt.home()
+                txt.print("found! path length: ")
+                txt.print_uw(pathstackptr)
+                txt.nl()
+                return true
+            }
+
+            ubyte cell = @(celladdr(cx,cy))
+            if cell & UP!=0 and @(celladdr(cx,cy-1)) & (WALKED|BACKTRACKED) ==0 {
+                @(pathstack + pathstackptr) = UP
+                ;txt.setcc(cx*2+1, cy*2, 81, 3)
+                cy--
+            }
+            else if cell & DOWN !=0 and @(celladdr(cx,cy+1)) & (WALKED|BACKTRACKED) ==0 {
+                @(pathstack + pathstackptr) = DOWN
+                ;txt.setcc(cx*2+1, cy*2+2, 81, 3)
+                cy++
+            }
+            else if cell & LEFT !=0 and @(celladdr(cx-1,cy)) & (WALKED|BACKTRACKED) ==0 {
+                @(pathstack + pathstackptr) = LEFT
+                ;txt.setcc(cx*2, cy*2+1, 81, 3)
+                cx--
+            }
+            else if cell & RIGHT !=0 and @(celladdr(cx+1,cy)) & (WALKED|BACKTRACKED) ==0 {
+                @(pathstack + pathstackptr) = RIGHT
+                ;txt.setcc(cx*2+2, cy*2+1, 81, 3)
+                cx++
+            }
+            else {
+                ; dead end, pop stack
+                pathstackptr--
+                if pathstackptr==65535 {
+                    txt.print("no solution?!\n")
+                    return true
+                }
+                @(celladdr(cx,cy)) |= BACKTRACKED
+                ;txt.setcc(cx*2+1, cy*2+1, 81, 2)
+                when @(pathstack + pathstackptr) {
+                    UP -> {
+                        ;txt.setcc(cx*2+1, cy*2+2, 81, 9)
+                        cy++
+                    }
+                    DOWN -> {
+                        ;txt.setcc(cx*2+1, cy*2, 81, 9)
+                        cy--
+                    }
+                    LEFT -> {
+                        ;txt.setcc(cx*2+2, cy*2+1, 81, 9)
+                        cx++
+                    }
+                    RIGHT -> {
+                        ;txt.setcc(cx*2, cy*2+1, 81, 9)
+                        cx--
+
+                        score++
+                    }
+                }
+                goto solve_loop
+            }
+            pathstackptr++
+            if pathstackptr==max_path_length {
+                txt.print("stack overflow, path too long\n")
+                return true
+            }
+            @(celladdr(cx,cy)) |= WALKED
+            ;txt.setcc(cx*2+1, cy*2+1, 81, 1)
+        }
+        return false
+    }
+
+    sub celladdr(ubyte cx, ubyte cy) -> uword {
+        return cells+(numCellsHoriz as uword)*cy+cx
+    }
+
+    sub drawCell(ubyte cx, ubyte cy) {
+        return
+;        ubyte x = cx * 2 + 1
+;        ubyte y = cy * 2 + 1
+;        ubyte doors = @(celladdr(cx,cy))
+;        if doors & UP !=0
+;            txt.setcc(x, y-1, ' ', EMPTYCOLOR)
+;        if doors & RIGHT !=0
+;            txt.setcc(x+1, y, ' ', EMPTYCOLOR)
+;        if doors & DOWN !=0
+;            txt.setcc(x, y+1, ' ', EMPTYCOLOR)
+;        if doors & LEFT !=0
+;            txt.setcc(x-1, y, ' ', EMPTYCOLOR)
+;        if doors & STONE !=0
+;            txt.setcc(x, y, 160, WALLCOLOR)
+;        else
+;            txt.setcc(x, y, 32, EMPTYCOLOR)
+;
+;        if doors & WALKED !=0
+;            txt.setcc(x, y, 81, 1)
+;        if doors & BACKTRACKED !=0
+;            txt.setcc(x, y, 81, 2)
+    }
+
+    sub initialize() {
+        sys.memset(cells, numCellsHoriz*numCellsVert, STONE)
+        ; txt.fill_screen(160, WALLCOLOR)
+        drawStartFinish()
+    }
+
+    sub drawStartFinish() {
+        ; txt.setcc(startCx*2+1,startCy*2+1,sc:'s',5)
+        ; txt.setcc(finishCx*2+1, finishCy*2+1, sc:'f', 13)
+    }
+}

benchmark-program/b_queens.p8

@@ -0,0 +1,63 @@
+%import textio
+
+; Recursive N-Queens solver.
+; The problem is: find all possible ways to place 8 Queen chess pieces on a chess board, so that none of them attacks any other.
+; (this program prints all solutions without taking mirroring and flipping the chess board into account)
+; Note: this program can be compiled for multiple target systems.
+
+queens {
+    const ubyte NUMQUEENS=8
+    ubyte[NUMQUEENS] board
+
+    sub could_place(ubyte row, ubyte col) -> bool {
+        if row==0
+            return true
+        ubyte i
+        for i in 0 to row-1 {
+            if board[i]==col or board[i]-i==col-row or board[i]+i==col+row
+                return false
+        }
+        return true
+    }
+
+    uword solution_count
+    uword maximum_duration
+
+    sub place_queen(ubyte row) -> bool {
+        if row == NUMQUEENS {
+            solution_count++
+            txt.chrout('.')
+            return cbm.RDTIM16()<maximum_duration
+        }
+        bool continue_running=true
+        ubyte col
+        for col in 0 to NUMQUEENS-1 {
+            if could_place(row, col) {
+                board[row] = col
+                ; we need to save the local variables row and col.
+                sys.push(row)
+                sys.push(col)
+                continue_running = place_queen(row + 1)
+                ; restore the local variables after the recursive call.
+                col = sys.pop()
+                row = sys.pop()
+                board[row] = 0
+
+                if not continue_running
+                    break
+            }
+        }
+        return continue_running
+    }
+
+    sub bench(uword max_time) -> uword {
+        solution_count = 0
+        maximum_duration = max_time
+        txt.nl()
+        cbm.SETTIM(0,0,0)
+        while cbm.RDTIM16() < maximum_duration {
+            void place_queen(0)
+        }
+        return solution_count
+    }
+}

benchmark-program/b_textelite.p8

@@ -0,0 +1,990 @@
+%import textio
+%import conv
+%import string
+%import string
+
+
+textelite {
+
+    const ubyte numforLave = 7      ;  Lave is 7th generated planet in galaxy one
+    const ubyte numforZaonce = 129
+    const ubyte numforDiso = 147
+    const ubyte numforRiedquat = 46
+    uword num_commands
+
+    sub bench(uword max_time) -> uword {
+        num_commands = 0
+        txt.lowercase()
+        cbm.SETTIM(0,0,0)
+        while cbm.RDTIM16()<max_time {
+            reinit()
+            run_commands(max_time)
+        }
+        return num_commands
+    }
+
+    sub reinit() {
+        ;txt.clear_screen()
+        ;txt.print("\n --- TextElite v1.3 ---\n")
+        txt.print("\nnew game\n")
+        elite_planet.set_seed(0, 0)
+        elite_galaxy.travel_to(1, numforLave)
+        elite_market.init(0)  ;  Lave's market is seeded with 0
+        elite_ship.init()
+        elite_planet.display(false, 0)
+        input_index = 0
+    }
+
+    sub run_commands(uword max_time) {
+        while cbm.RDTIM16() < max_time {
+            str input = "????????"
+            ;txt.print("\nCash: ")
+            ;elite_util.print_10s(elite_ship.cash)
+            ;txt.print("\nCommand (?=help): ")
+            ubyte num_chars = next_input(input)
+            ;txt.nl()
+            if num_chars!=0 {
+                when input[0] {
+                    'q' -> {
+                        bool has_error = false
+                        if elite_galaxy.number != 2 {
+                            txt.print("\nERROR: galaxy is not 2: ")
+                            txt.print_ub(elite_galaxy.number)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_planet.number != 164 {
+                            txt.print("\nERROR: planet is not 164: ")
+                            txt.print_ub(elite_planet.number)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_planet.x != 116 {
+                            txt.print("\nERROR: planet.x is not 116: ")
+                            txt.print_ub(elite_planet.x)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_planet.y != 201 {
+                            txt.print("\nERROR: planet.y is not 201: ")
+                            txt.print_ub(elite_planet.y)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if "ribeen" != elite_planet.name {
+                            txt.print("\nERROR: planet.name is not 'ribeen': ")
+                            txt.print(elite_planet.name)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_ship.cash != 1212 {
+                            txt.print("\nERROR: cash is not 1212: ")
+                            txt.print_uw(elite_ship.cash)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_ship.fuel != 50 {
+                            txt.print("\nERROR: fuel is not 50:")
+                            txt.print_ub(elite_ship.fuel)
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_ship.cargohold[0] != 3 {
+                            txt.print("\nERROR: food is not 3:")
+                            txt.print_ub(elite_ship.cargohold[0])
+                            txt.nl()
+                            has_error=true
+                        }
+                        if elite_ship.cargohold[1] != 0 {
+                            txt.print("\nERROR: textiles is not 0:")
+                            txt.print_ub(elite_ship.cargohold[1])
+                            txt.nl()
+                            has_error=true
+                        }
+                        if has_error
+                            sys.exit(1)
+                        return
+                    }
+                    'b' -> elite_trader.do_buy()
+                    's' -> elite_trader.do_sell()
+                    'f' -> elite_trader.do_fuel()
+                    'j' -> elite_trader.do_jump()
+                    't' -> elite_trader.do_teleport()
+                    'g' -> elite_trader.do_next_galaxy()
+                    'i' -> elite_trader.do_info()
+                    'm' -> {
+                        if input[1]=='a' and input[2]=='p'
+                            elite_trader.do_map()
+                        else
+                            elite_trader.do_show_market()
+                    }
+                    'l' -> elite_trader.do_local()
+                    'c' -> elite_trader.do_cash()
+                    'h' -> elite_trader.do_hold()
+                }
+                num_commands++
+            }
+        }
+    }
+
+    str[] inputs = [
+        "i",
+        "diso",
+        "i",
+        "lave",
+        "m",
+        "b",
+        "food",
+        "15",
+        "map",
+        "g",
+        "map",
+        "l",
+        "j",
+        "zao",
+        "s",
+        "food",
+        "12",
+        "tele",
+        "quti",
+        "tele",
+        "aro",
+        "i",
+        "diso",
+        "i",
+        "lave",
+        "i",
+        "zao",
+        "galhyp",
+        "fuel",
+        "20",
+        "j",
+        "rib",
+        "i",
+        "rib",
+        "i",
+        "tiri",
+        "q",
+        0
+    ]
+
+    ubyte input_index
+
+    sub next_input(str buffer) -> ubyte {
+        input_index++
+        return string.copy(inputs[input_index], buffer)
+    }
+}
+
+elite_trader {
+    str input = "??????????"
+    ubyte num_chars
+
+    sub do_jump() {
+        ;txt.print("\nJump to what system? ")
+        jump_to_system()
+    }
+
+    sub do_teleport() {
+        ;txt.print("\nCheat! Teleport to what system? ")
+        ubyte fuel = elite_ship.fuel
+        elite_ship.fuel = 255
+        jump_to_system()
+        elite_ship.fuel = fuel
+    }
+
+    sub jump_to_system() {
+        void textelite.next_input(input)
+        ubyte current_planet = elite_planet.number
+        ubyte x = elite_planet.x
+        ubyte y = elite_planet.y
+        if elite_galaxy.search_closest_planet(input) {
+            ubyte distance = elite_planet.distance(x, y)
+            if distance <= elite_ship.fuel {
+                elite_galaxy.init_market_for_planet()
+                elite_ship.fuel -= distance
+                ;txt.print("\n\nHyperspace jump! Arrived at:\n")
+                elite_planet.display(true,0 )
+                return
+            }
+            ;txt.print("\nInsufficient fuel\n")
+        } else {
+            ;txt.print(" Not found!\n")
+        }
+        elite_galaxy.travel_to(elite_galaxy.number, current_planet)
+    }
+
+    sub do_buy() {
+        ;txt.print("\nBuy what commodity? ")
+        str commodity = "???????????????"
+        void textelite.next_input(commodity)
+        ubyte ci = elite_market.match(commodity)
+        if ci & 128 !=0 {
+            txt.print("Unknown\n")
+        } else {
+            ;txt.print("\nHow much? ")
+            void textelite.next_input(input)
+            ubyte amount = conv.str2ubyte(input)
+            if elite_market.current_quantity[ci] < amount {
+                txt.print(" Insufficient supply!\n")
+            } else {
+                uword price = elite_market.current_price[ci] * amount
+                ;txt.print(" Total price: ")
+                ;elite_util.print_10s(price)
+                if price > elite_ship.cash {
+                    txt.print(" Not enough cash!\n")
+                } else {
+                    elite_ship.cash -= price
+                    elite_ship.cargohold[ci] += amount
+                    elite_market.current_quantity[ci] -= amount
+                }
+            }
+        }
+    }
+
+    sub do_sell() {
+        ;txt.print("\nSell what commodity? ")
+        str commodity = "???????????????"
+        void textelite.next_input(commodity)
+        ubyte ci = elite_market.match(commodity)
+        if ci & 128 !=0 {
+            txt.print("Unknown\n")
+        } else {
+            ;txt.print("\nHow much? ")
+            void textelite.next_input(input)
+            ubyte amount = conv.str2ubyte(input)
+            if elite_ship.cargohold[ci] < amount {
+                txt.print(" Insufficient supply!\n")
+            } else {
+                uword price = elite_market.current_price[ci] * amount
+                ;txt.print(" Total price: ")
+                ;elite_util.print_10s(price)
+                elite_ship.cash += price
+                elite_ship.cargohold[ci] -= amount
+                elite_market.current_quantity[ci] += amount
+            }
+        }
+    }
+
+    sub do_fuel() {
+        ;txt.print("\nBuy fuel. Amount? ")
+        void textelite.next_input(input)
+        ubyte buy_fuel = 10*conv.str2ubyte(input)
+        ubyte max_fuel = elite_ship.Max_fuel - elite_ship.fuel
+        if buy_fuel > max_fuel
+            buy_fuel = max_fuel
+        uword price = buy_fuel as uword * elite_ship.Fuel_cost
+        if price > elite_ship.cash {
+            txt.print("Not enough cash!\n")
+        } else {
+            elite_ship.cash -= price
+            elite_ship.fuel += buy_fuel
+        }
+    }
+
+    sub do_cash() {
+        ;txt.print("\nCheat! Set cash amount: ")
+        void textelite.next_input(input)
+        elite_ship.cash = conv.str2uword(input)
+    }
+
+    sub do_hold() {
+        ;txt.print("\nCheat! Set cargohold size: ")
+        void textelite.next_input(input)
+        elite_ship.Max_cargo = conv.str2ubyte(input)
+    }
+
+    sub do_next_galaxy() {
+        txt.print("\n>>>>> Galaxy Hyperjump!\n")
+        elite_galaxy.travel_to(elite_galaxy.number+1, elite_planet.number)
+        elite_planet.display(false, 0)
+    }
+
+    sub do_info() {
+        ;txt.print("\nSystem name (empty=current): ")
+        num_chars = textelite.next_input(input)
+        if num_chars!=0 {
+            ubyte current_planet = elite_planet.number
+            ubyte x = elite_planet.x
+            ubyte y = elite_planet.y
+            if elite_galaxy.search_closest_planet(input) {
+                ubyte distance = elite_planet.distance(x, y)
+                elite_planet.display(false, distance)
+            } else {
+                ;txt.print(" Not found!")
+            }
+            elite_galaxy.travel_to(elite_galaxy.number, current_planet)
+        } else {
+            elite_planet.display(false, 0)
+        }
+    }
+
+    sub do_local() {
+        elite_galaxy.local_area()
+    }
+
+    sub do_map() {
+        ;txt.print("\n(l)ocal or (g)alaxy starmap? ")
+        num_chars = textelite.next_input(input)
+        if num_chars!=0 {
+            elite_galaxy.starmap(input[0]=='l')
+        }
+    }
+
+    sub do_show_market() {
+        elite_market.display()
+        ;txt.print("\nFuel: ")
+        ;elite_util.print_10s(elite_ship.fuel)
+        ;txt.print("   Cargohold space: ")
+        ;txt.print_ub(elite_ship.cargo_free())
+        ;txt.print("t\n")
+    }
+}
+
+elite_ship {
+    const ubyte Max_fuel = 70
+    const ubyte Fuel_cost = 2
+    ubyte Max_cargo = 20
+
+    ubyte fuel
+    uword cash
+    ubyte[17] cargohold
+
+    sub init() {
+        sys.memset(cargohold, len(cargohold), 0)
+        fuel = Max_fuel
+        cash = 1000
+    }
+}
+
+elite_market {
+    ubyte[17] baseprices = [$13, $14, $41, $28, $53, $C4, $EB, $9A, $75, $4E, $7C, $B0, $20, $61, $AB, $2D, $35]
+    byte[17] gradients = [-$02, -$01, -$03, -$05, -$05, $08, $1D, $0E, $06, $01, $0d, -$09, -$01, -$01, -$02, -$01, $0F]
+    ubyte[17] basequants = [$06, $0A, $02, $E2, $FB, $36, $08, $38, $28, $11, $1D, $DC, $35, $42, $37, $FA, $C0]
+    ubyte[17] maskbytes = [$01, $03, $07, $1F, $0F, $03, $78, $03, $07, $1F, $07, $3F, $03, $07, $1F, $0F, $07]
+    str[17] names = ["Food", "Textiles", "Radioactives", "Slaves", "Liquor/Wines", "Luxuries", "Narcotics", "Computers",
+                     "Machinery", "Alloys", "Firearms", "Furs", "Minerals", "Gold", "Platinum", "Gem-Stones", "Alien Items"]
+
+    ubyte[17] current_quantity
+    uword[17] current_price
+
+    sub init(ubyte fluct) {
+        ; Prices and availabilities are influenced by the planet's economy type
+        ; (0-7) and a random "fluctuation" byte that was kept within the saved
+        ; commander position to keep the market prices constant over gamesaves.
+        ; Availabilities must be saved with the game since the player alters them
+        ; by buying (and selling(?))
+        ;
+        ; Almost all commands are one byte only and overflow "errors" are
+        ; extremely frequent and exploited.
+        ;
+        ; Trade Item prices are held internally in a single byte=true value/4.
+        ; The decimal point in prices is introduced only when printing them.
+        ; Internally, all prices are integers.
+        ; The player's cash is held in four bytes.
+        ubyte ci
+        for ci in 0 to len(names)-1 {
+            word product
+            byte changing
+            product = elite_planet.economy as word * gradients[ci]
+            changing = fluct & maskbytes[ci]  as byte
+            ubyte q = (basequants[ci] as word + changing - product) as ubyte
+            if q & $80 !=0
+                q = 0  ; clip to positive 8-bit
+            current_quantity[ci] = q & $3f
+            q = (baseprices[ci] + changing + product) as ubyte
+            current_price[ci] = q * $0004
+        }
+        current_quantity[16] = 0        ; force nonavailability of Alien Items
+    }
+
+    sub display() {
+        return
+;        ubyte ci
+;        txt.nl()
+;        elite_planet.print_name_uppercase()
+;        txt.print(" trade market:\n    COMMODITY / PRICE / AVAIL / IN HOLD\n")
+;        for ci in 0 to len(names)-1 {
+;            elite_util.print_right(13, names[ci])
+;            txt.print("   ")
+;            elite_util.print_10s(current_price[ci])
+;            txt.column(24)
+;            txt.print_ub(current_quantity[ci])
+;            txt.chrout(' ')
+;            when units[ci] {
+;                0 -> txt.chrout('t')
+;                1 -> txt.print("kg")
+;                2 -> txt.chrout('g')
+;            }
+;            txt.column(32)
+;            txt.print_ub(elite_ship.cargohold[ci])
+;            txt.nl()
+;        }
+    }
+
+    sub match(uword nameptr) -> ubyte {
+        ubyte ci
+        for ci in 0 to len(names)-1 {
+            if elite_util.prefix_matches(nameptr, names[ci])
+                return ci
+        }
+        return 255
+    }
+}
+
+elite_galaxy {
+    const uword GALSIZE = 256
+    const uword base0 = $5A4A       ; seeds for the first galaxy
+    const uword base1 = $0248
+    const uword base2 = $B753
+
+    str pn_pairs = "..lexegezacebisousesarmaindirea.eratenberalavetiedorquanteisrion"
+
+    ubyte number
+
+    uword[3] seed
+
+    sub init(ubyte galaxynum) {
+        number = 1
+        elite_planet.number = 255
+        seed[0] = base0
+        seed[1] = base1
+        seed[2] = base2
+        repeat galaxynum-1 {
+            nextgalaxy()
+        }
+    }
+
+    sub nextgalaxy() {
+        textelite.num_commands++
+
+        seed[0] = twist(seed[0])
+        seed[1] = twist(seed[1])
+        seed[2] = twist(seed[2])
+        number++
+        if number==9
+            number = 1
+    }
+
+    sub travel_to(ubyte galaxynum, ubyte system) {
+        init(galaxynum)
+        generate_next_planet()   ; always at least planet 0  (separate to avoid repeat ubyte overflow)
+        repeat system {
+            generate_next_planet()
+            textelite.num_commands++
+        }
+        elite_planet.name = make_current_planet_name()
+        init_market_for_planet()
+    }
+
+    sub init_market_for_planet() {
+        elite_market.init(lsb(seed[0])+msb(seed[2]))
+    }
+
+    sub search_closest_planet(uword nameptr) -> bool {
+        textelite.num_commands++
+
+        ubyte x = elite_planet.x
+        ubyte y = elite_planet.y
+        ubyte current_planet_num = elite_planet.number
+
+        init(number)
+        bool found = false
+        ubyte current_closest_pi
+        ubyte current_distance = 127
+        ubyte pi
+        for pi in 0 to 255 {
+            generate_next_planet()
+            elite_planet.name = make_current_planet_name()
+            if elite_util.prefix_matches(nameptr, elite_planet.name) {
+                ubyte distance = elite_planet.distance(x, y)
+                if distance < current_distance {
+                    current_distance = distance
+                    current_closest_pi = pi
+                    found = true
+                }
+            }
+        }
+
+        if found
+            travel_to(number, current_closest_pi)
+        else
+            travel_to(number, current_planet_num)
+
+        return found
+    }
+
+    sub local_area() {
+        ubyte current_planet = elite_planet.number
+        ubyte px = elite_planet.x
+        ubyte py = elite_planet.y
+        ubyte pn = 0
+
+        init(number)
+;        txt.print("\nGalaxy #")
+;        txt.print_ub(number)
+;        txt.print(" - systems in vicinity:\n")
+        do {
+            generate_next_planet()
+            ubyte distance = elite_planet.distance(px, py)
+            if distance <= elite_ship.Max_fuel {
+;                if distance <= elite_ship.fuel
+;                    txt.chrout('*')
+;                else
+;                    txt.chrout('-')
+;                txt.spc()
+                elite_planet.name = make_current_planet_name()
+                elite_planet.display(true, distance)
+            }
+            pn++
+        } until pn==0
+
+        travel_to(number, current_planet)
+    }
+
+    sub starmap(bool local) {
+        ubyte current_planet = elite_planet.number
+        ubyte px = elite_planet.x
+        ubyte py = elite_planet.y
+        str current_name = "        "       ; 8 max
+        ubyte pn = 0
+
+        current_name = elite_planet.name
+        init(number)
+;        txt.clear_screen()
+;        txt.print("Galaxy #")
+;        txt.print_ub(number)
+;        if local
+;            txt.print(" - local systems")
+;        else
+;            txt.print(" - galaxy")
+;        txt.print(" starmap:\n")
+        ubyte max_distance = 255
+        if local
+            max_distance = elite_ship.Max_fuel
+        ubyte home_sx
+        ubyte home_sy
+        ubyte home_distance
+
+        do {
+            generate_next_planet()
+            ubyte distance = elite_planet.distance(px, py)
+            if distance <= max_distance {
+                elite_planet.name = make_current_planet_name()
+                elite_planet.name[0] = string.upperchar(elite_planet.name[0])
+                uword tx = elite_planet.x
+                uword ty = elite_planet.y
+                if local {
+                    tx = tx + 24 - px
+                    ty = ty + 24 - py
+                }
+                ubyte sx = display_scale_x(tx)
+                ubyte sy = display_scale_y(ty)
+                ubyte char = '*'
+                if elite_planet.number==current_planet
+                    char = '%'
+                if local {
+                    print_planet_details(elite_planet.name, sx, sy, distance)
+                } else if elite_planet.number==current_planet {
+                    home_distance = distance
+                    home_sx = sx
+                    home_sy = sy
+                }
+                ; txt.setchr(2+sx, 2+sy, char)
+            }
+            pn++
+        } until pn==0
+
+        if not local
+            print_planet_details(current_name, home_sx, home_sy, home_distance)
+
+;        if local
+;            txt.plot(0, display_scale_y(64) + 4)
+;        else
+;            txt.plot(0, display_scale_y(256) + 4 as ubyte)
+        travel_to(number, current_planet)
+
+        sub print_planet_details(str name, ubyte screenx, ubyte screeny, ubyte d) {
+            return
+;            txt.plot(2+screenx-2, 2+screeny+1)
+;            txt.print(name)
+;            if d!=0 {
+;                txt.plot(2+screenx-2, 2+screeny+2)
+;                elite_util.print_10s(d)
+;                txt.print(" LY")
+;            }
+        }
+
+        sub display_scale_x(uword x) -> ubyte {
+            if local
+                return x/2 as ubyte
+            return x/8 as ubyte
+        }
+
+        sub display_scale_y(uword y) -> ubyte {
+            if local
+                return y/4 as ubyte
+            return y/16 as ubyte
+        }
+    }
+
+    ubyte pn_pair1
+    ubyte pn_pair2
+    ubyte pn_pair3
+    ubyte pn_pair4
+    bool longname
+
+    sub generate_next_planet() {
+        determine_planet_properties()
+        longname = lsb(seed[0]) & 64 !=0
+
+        ; Always four iterations of random number
+        pn_pair1 = (msb(seed[2]) & 31) * 2
+        tweakseed()
+        pn_pair2 = (msb(seed[2]) & 31) * 2
+        tweakseed()
+        pn_pair3 = (msb(seed[2]) & 31) * 2
+        tweakseed()
+        pn_pair4 = (msb(seed[2]) & 31) * 2
+        tweakseed()
+    }
+
+    sub make_current_planet_name() -> str {
+        ubyte ni = 0
+        str name = "         "    ; max 8
+
+        if pn_pairs[pn_pair1] != '.' {
+            name[ni] = pn_pairs[pn_pair1]
+            ni++
+        }
+        if pn_pairs[pn_pair1+1] != '.' {
+            name[ni] = pn_pairs[pn_pair1+1]
+            ni++
+        }
+        if pn_pairs[pn_pair2] != '.' {
+            name[ni] = pn_pairs[pn_pair2]
+            ni++
+        }
+        if pn_pairs[pn_pair2+1] != '.' {
+            name[ni] = pn_pairs[pn_pair2+1]
+            ni++
+        }
+        if pn_pairs[pn_pair3] != '.' {
+            name[ni] = pn_pairs[pn_pair3]
+            ni++
+        }
+        if pn_pairs[pn_pair3+1] != '.' {
+            name[ni] = pn_pairs[pn_pair3+1]
+            ni++
+        }
+
+        if longname {
+            if pn_pairs[pn_pair4] != '.' {
+                name[ni] = pn_pairs[pn_pair4]
+                ni++
+            }
+            if pn_pairs[pn_pair4+1] != '.' {
+                name[ni] = pn_pairs[pn_pair4+1]
+                ni++
+            }
+        }
+
+        name[ni] = 0
+        return name
+    }
+
+    sub determine_planet_properties() {
+        ; create the planet's characteristics
+        elite_planet.number++
+        elite_planet.x = msb(seed[1])
+        elite_planet.y = msb(seed[0])
+        elite_planet.govtype = lsb(seed[1]) >> 3 & 7  ; bits 3,4 &5 of w1
+        elite_planet.economy = msb(seed[0]) & 7  ; bits 8,9 &A of w0
+        if elite_planet.govtype <= 1
+            elite_planet.economy = (elite_planet.economy | 2)
+        elite_planet.techlevel = (msb(seed[1]) & 3) + (elite_planet.economy ^ 7)
+        elite_planet.techlevel += elite_planet.govtype >> 1
+        if elite_planet.govtype & 1 !=0
+            elite_planet.techlevel++
+        elite_planet.population = 4 * elite_planet.techlevel + elite_planet.economy
+        elite_planet.population += elite_planet.govtype + 1
+        elite_planet.productivity = ((elite_planet.economy ^ 7) + 3) * (elite_planet.govtype + 4)
+        elite_planet.productivity *= elite_planet.population * 8
+        ubyte seed2_msb = msb(seed[2])
+        elite_planet.radius = mkword((seed2_msb & 15) + 11, elite_planet.x)
+        elite_planet.species_is_alien = lsb(seed[2]) & 128 !=0      ; bit 7 of w2_lo
+        if elite_planet.species_is_alien {
+            elite_planet.species_size = (seed2_msb >> 2) & 7      ; bits 2-4 of w2_hi
+            elite_planet.species_color = seed2_msb >> 5           ; bits 5-7 of w2_hi
+            elite_planet.species_look = (seed2_msb ^ msb(seed[1])) & 7   ;bits 0-2 of (w0_hi EOR w1_hi)
+            elite_planet.species_kind = (elite_planet.species_look + (seed2_msb & 3)) & 7      ;Add bits 0-1 of w2_hi to A from previous step, and take bits 0-2 of the result
+        }
+
+        elite_planet.goatsoup_seed[0] = lsb(seed[1])
+        elite_planet.goatsoup_seed[1] = msb(seed[1])
+        elite_planet.goatsoup_seed[2] = lsb(seed[2])
+        elite_planet.goatsoup_seed[3] = seed2_msb
+    }
+
+    sub tweakseed() {
+        uword temp = seed[0] + seed[1] + seed[2]
+        seed[0] = seed[1]
+        seed[1] = seed[2]
+        seed[2] = temp
+    }
+
+    sub twist(uword x) -> uword {
+        ubyte xh = msb(x)
+        ubyte xl = lsb(x)
+        xh <<= 1        ; make sure carry flag is not used on first shift!
+        rol(xl)
+        return mkword(xh, xl)
+    }
+}
+
+elite_planet {
+    str[] words81 = ["fabled", "notable", "well known", "famous", "noted"]
+    str[] words82 = ["very", "mildly", "most", "reasonably", ""]
+    str[] words83 = ["ancient", "\x95", "great", "vast", "pink"]
+    str[] words84 = ["\x9E \x9D plantations", "mountains", "\x9C", "\x94 forests", "oceans"]
+    str[] words85 = ["shyness", "silliness", "mating traditions", "loathing of \x86", "love for \x86"]
+    str[] words86 = ["food blenders", "tourists", "poetry", "discos", "\x8E"]
+    str[] words87 = ["talking tree", "crab", "bat", "lobst", "\xB2"]
+    str[] words88 = ["beset", "plagued", "ravaged", "cursed", "scourged"]
+    str[] words89 = ["\x96 civil war", "\x9B \x98 \x99s", "a \x9B disease", "\x96 earthquakes", "\x96 solar activity"]
+    str[] words8A = ["its \x83 \x84", "the \xB1 \x98 \x99", "its inhabitants' \x9A \x85", "\xA1", "its \x8D \x8E"]
+    str[] words8B = ["juice", "brandy", "water", "brew", "gargle blasters"]
+    str[] words8C = ["\xB2", "\xB1 \x99", "\xB1 \xB2", "\xB1 \x9B", "\x9B \xB2"]
+    str[] words8D = ["fabulous", "exotic", "hoopy", "unusual", "exciting"]
+    str[] words8E = ["cuisine", "night life", "casinos", "sit coms", " \xA1 "]
+    str[] words8F = ["\xB0", "The planet \xB0", "The world \xB0", "This planet", "This world"]
+    str[] words90 = ["n unremarkable", " boring", " dull", " tedious", " revolting"]
+    str[] words91 = ["planet", "world", "place", "little planet", "dump"]
+    str[] words92 = ["wasp", "moth", "grub", "ant", "\xB2"]
+    str[] words93 = ["poet", "arts graduate", "yak", "snail", "slug"]
+    str[] words94 = ["tropical", "dense", "rain", "impenetrable", "exuberant"]
+    str[] words95 = ["funny", "wierd", "unusual", "strange", "peculiar"]
+    str[] words96 = ["frequent", "occasional", "unpredictable", "dreadful", "deadly"]
+    str[] words97 = ["\x82 \x81 for \x8A", "\x82 \x81 for \x8A and \x8A", "\x88 by \x89", "\x82 \x81 for \x8A but \x88 by \x89", "a\x90 \x91"]
+    str[] words98 = ["\x9B", "mountain", "edible", "tree", "spotted"]
+    str[] words99 = ["\x9F", "\xA0", "\x87oid", "\x93", "\x92"]
+    str[] words9A = ["ancient", "exceptional", "eccentric", "ingrained", "\x95"]
+    str[] words9B = ["killer", "deadly", "evil", "lethal", "vicious"]
+    str[] words9C = ["parking meters", "dust clouds", "ice bergs", "rock formations", "volcanoes"]
+    str[] words9D = ["plant", "tulip", "banana", "corn", "\xB2weed"]
+    str[] words9E = ["\xB2", "\xB1 \xB2", "\xB1 \x9B", "inhabitant", "\xB1 \xB2"]
+    str[] words9F = ["shrew", "beast", "bison", "snake", "wolf"]
+    str[] wordsA0 = ["leopard", "cat", "monkey", "goat", "fish"]
+    str[] wordsA1 = ["\x8C \x8B", "\xB1 \x9F \xA2", "its \x8D \xA0 \xA2", "\xA3 \xA4", "\x8C \x8B"]
+    str[] wordsA2 = ["meat", "cutlet", "steak", "burgers", "soup"]
+    str[] wordsA3 = ["ice", "mud", "Zero-G", "vacuum", "\xB1 ultra"]
+    str[] wordsA4 = ["hockey", "cricket", "karate", "polo", "tennis"]
+
+    uword[] @shared wordlists = [
+        words81, words82, words83, words84, words85, words86, words87, words88,
+        words89, words8A, words8B, words8C, words8D, words8E, words8F, words90,
+        words91, words92, words93, words94, words95, words96, words97, words98,
+        words99, words9A, words9B, words9C, words9D, words9E, words9F, wordsA0,
+        wordsA1, wordsA2, wordsA3, wordsA4]
+
+    str pairs0 = "abouseitiletstonlonuthnoallexegezacebisousesarmaindirea.eratenbe"
+
+    ubyte[4] goatsoup_rnd = [0, 0, 0, 0]
+    ubyte[4] goatsoup_seed = [0, 0, 0, 0]
+
+    str name = "        "       ; 8 max
+    ubyte number                ; starts at 0 in new galaxy, then increases by 1 for each generated planet
+    ubyte x
+    ubyte y
+    ubyte economy
+    ubyte govtype
+    ubyte techlevel
+    ubyte population
+    uword productivity
+    uword radius
+    bool species_is_alien      ; otherwise "Human Colonials"
+    ubyte species_size
+    ubyte species_color
+    ubyte species_look
+    ubyte species_kind
+
+    sub set_seed(uword s1, uword s2) {
+        goatsoup_seed[0] = lsb(s1)
+        goatsoup_seed[1] = msb(s1)
+        goatsoup_seed[2] = lsb(s2)
+        goatsoup_seed[3] = msb(s2)
+        reset_rnd()
+    }
+
+    sub reset_rnd() {
+        goatsoup_rnd[0] = goatsoup_seed[0]
+        goatsoup_rnd[1] = goatsoup_seed[1]
+        goatsoup_rnd[2] = goatsoup_seed[2]
+        goatsoup_rnd[3] = goatsoup_seed[3]
+    }
+
+    sub random_name() -> str {
+        ubyte ii
+        str randname = "        "       ; 8 chars max
+        ubyte nx = 0
+        for ii in 0 to goatsoup_rnd_number() & 3 {
+            ubyte xx = goatsoup_rnd_number() & $3e
+            if pairs0[xx] != '.' {
+                randname[nx] = pairs0[xx]
+                nx++
+            }
+            xx++
+            if pairs0[xx] != '.' {
+                randname[nx] = pairs0[xx]
+                nx++
+            }
+        }
+        randname[nx] = 0
+        randname[0] = string.upperchar(randname[0])
+        return randname
+    }
+
+    sub goatsoup_rnd_number() -> ubyte {
+        ubyte xx = goatsoup_rnd[0] * 2
+        uword a = xx as uword + goatsoup_rnd[2]
+        if goatsoup_rnd[0] > 127
+            a ++
+        goatsoup_rnd[0] = lsb(a)
+        goatsoup_rnd[2] = xx
+        xx = goatsoup_rnd[1]
+        ubyte ac = xx + goatsoup_rnd[3] + msb(a)
+        goatsoup_rnd[1] = ac
+        goatsoup_rnd[3] = xx
+        return ac
+    }
+
+    sub distance(ubyte px, ubyte py) -> ubyte {
+        uword ax
+        uword ay
+        if px>x
+            ax=px-x
+        else
+            ax=x-px
+        if py>y
+            ay=py-y
+        else
+            ay=y-py
+        ay /= 2
+        ubyte d = sqrt(ax*ax + ay*ay)
+        if d>63
+            return 255
+        return d*4
+    }
+
+    sub soup() -> str {
+        str planet_result = " " * 160
+        uword[6] source_stack
+        ubyte stack_ptr = 0
+        str start_source = "\x8F is \x97."
+        uword source_ptr = &start_source
+        uword result_ptr = &planet_result
+
+        reset_rnd()
+        recursive_soup()
+        return planet_result
+
+        sub recursive_soup() {
+            repeat {
+                ubyte c = @(source_ptr)
+                source_ptr++
+                if c == $00 {
+                    @(result_ptr) = 0
+                    return
+                }
+                else if c <= $80 {
+                    @(result_ptr) = c
+                    result_ptr++
+                }
+                else {
+                    if c <= $a4 {
+                        ubyte rnr = goatsoup_rnd_number()
+                        ubyte wordNr = ((rnr >= $33) as ubyte) + ((rnr >= $66) as ubyte) + ((rnr >= $99) as ubyte) + ((rnr >= $CC) as ubyte)
+                        source_stack[stack_ptr] = source_ptr
+                        stack_ptr++
+                        source_ptr = getword(c, wordNr)
+                        recursive_soup()    ; RECURSIVE CALL - ignore the warning message from the compiler; we don't use local variables or parameters so we're safe in this case
+                        stack_ptr--
+                        source_ptr = source_stack[stack_ptr]
+                    } else {
+                        if c == $b0 {
+                            @(result_ptr) = string.upperchar(name[0])
+                            result_ptr++
+                            concat_string(&name + 1)
+                        }
+                        else if c == $b1 {
+                            @(result_ptr) = string.upperchar(name[0])
+                            result_ptr++
+                            ubyte ni
+                            for ni in 1 to len(name) {
+                                ubyte cc = name[ni]
+                                if cc in ['e', 'o', 0]
+                                    break
+                                else {
+                                    @(result_ptr) = cc
+                                    result_ptr++
+                                }
+                            }
+                            @(result_ptr) = 'i'
+                            result_ptr++
+                            @(result_ptr) = 'a'
+                            result_ptr++
+                            @(result_ptr) = 'n'
+                            result_ptr++
+                        }
+                        else if c == $b2 {
+                            concat_string(random_name())
+                        }
+                        else {
+                            @(result_ptr) = c
+                            result_ptr++
+                        }
+                    }
+                }
+            }
+        }
+
+        sub concat_string(uword str_ptr) {
+            repeat {
+                ubyte c = @(str_ptr)
+                if c==0
+                    break
+                else {
+                    @(result_ptr) = c
+                    str_ptr++
+                    result_ptr++
+                }
+            }
+        }
+    }
+
+    sub display(bool compressed, ubyte distance) {
+        txt.print(soup())
+        txt.nl()
+    }
+
+    sub getword(ubyte listnum, ubyte wordidx) -> uword {
+        uword list = wordlists[listnum-$81]
+        return peekw(list + wordidx*2)
+    }
+}
+
+elite_util {
+    sub prefix_matches(uword prefixptr, uword stringptr) -> bool {
+        repeat {
+            ubyte pc = @(prefixptr)
+            ubyte sc = @(stringptr)
+            if pc == 0
+                return true
+            ; to lowercase for case insensitive compare:
+            if string.lowerchar(pc)!=string.lowerchar(sc)
+                return false
+            prefixptr++
+            stringptr++
+        }
+    }
+}

benchmark-program/benchmark.p8

@@ -0,0 +1,112 @@
+
+;  This benchmark program is meant to check for regressions in the
+;  Prog8 compiler's code-generator (performance wise).
+;
+;  As the X16 computer is a more or less fixed system, it's not very useful
+;  to benchmark the computer itself with.
+
+
+%import textio
+%import b_adpcm
+%import b_circles
+%import b_3d
+%import b_life
+%import b_mandelbrot
+%import b_queens
+%import b_textelite
+%import b_maze
+
+%zeropage basicsafe
+%option no_sysinit
+
+
+main {
+
+    str[20] benchmark_names
+    uword[20] benchmark_score
+
+
+    sub start() {
+        ubyte benchmark_number
+
+        void cx16.set_screen_mode(3)
+        txt.color2(1, 6)
+        txt.clear_screen()
+
+        txt.print("\n\n\n  prog8 compiler benchmark tests.\n")
+        sys.wait(60)
+
+        benchmark_number = 0
+
+        announce_benchmark("maze solver")
+        benchmark_score[benchmark_number]  = maze.bench(300)
+        benchmark_number++
+
+        announce_benchmark("n-queens")
+        benchmark_score[benchmark_number]  = queens.bench(300)
+        benchmark_number++
+
+        announce_benchmark("mandelbrot (floating point)")
+        benchmark_score[benchmark_number]  = mandelbrot.calc(400)
+        benchmark_number++
+
+        announce_benchmark("game of life")
+        benchmark_score[benchmark_number]  = life.benchmark(300)
+        benchmark_number++
+
+        announce_benchmark("3d model rotation")
+        benchmark_score[benchmark_number]  = rotate3d.benchmark(300)
+        benchmark_number++
+
+        announce_benchmark("adpcm audio decoding")
+        benchmark_score[benchmark_number]  = adpcm.decode_benchmark(300)
+        benchmark_number++
+
+        announce_benchmark("circles with gfx_lores")
+        benchmark_score[benchmark_number]  = circles.draw(false, 300)
+        benchmark_number++
+
+;        announce_benchmark("circles with kernal")
+;        benchmark_score[benchmark_number]  = circles.draw(true, 300)
+;        benchmark_number++
+
+        announce_benchmark("text-elite")
+        benchmark_score[benchmark_number]  = textelite.bench(120)
+        benchmark_number++
+
+        benchmark_names[benchmark_number] = 0
+        benchmark_score[benchmark_number] = 0
+
+        void cx16.set_screen_mode(3)
+        txt.uppercase()
+        txt.color2(1, 6)
+        uword final_score
+        benchmark_number = 0
+        txt.print("\nscore benchmark\n\n")
+        do {
+            txt.spc()
+            txt.print_uw(benchmark_score[benchmark_number])
+            txt.column(6)
+            txt.print(benchmark_names[benchmark_number])
+            final_score += benchmark_score[benchmark_number]
+            txt.nl()
+            benchmark_number++
+        } until benchmark_names[benchmark_number]==0
+
+        txt.print("\n\nfinal score : ")
+        txt.print_uw(final_score)
+        txt.nl()
+
+        sub announce_benchmark(str name) {
+            benchmark_names[benchmark_number] = name
+            void cx16.set_screen_mode(3)
+            txt.uppercase()
+            txt.color2(1, 6)
+            txt.clear_screen()
+            txt.plot(4, 6)
+            txt.print(benchmark_names[benchmark_number])
+            txt.nl()
+            sys.wait(60)
+        }
+    }
+}

codeCore/src/prog8/code/SymbolTable.kt

@@ -1,5 +1,6 @@
 package prog8.code
 
+import prog8.code.ast.PtAsmSub
 import prog8.code.ast.PtNode
 import prog8.code.ast.PtProgram
 import prog8.code.core.*
@@ -97,7 +98,7 @@ enum class StNodeType {
     // MODULE,     // not used with current scoping rules
     BLOCK,
     SUBROUTINE,
-    ROMSUB,
+    EXTSUB,
     LABEL,
     STATICVAR,
     MEMVAR,
@@ -178,31 +179,48 @@ open class StNode(val name: String,
 
 class StStaticVariable(name: String,
                        val dt: DataType,
-                       val onetimeInitializationNumericValue: Double?,      // regular (every-run-time) initialization is done via regular assignments
-                       val onetimeInitializationStringValue: StString?,
-                       val onetimeInitializationArrayValue: StArray?,
+                       val initializationStringValue: StString?,
+                       val initializationArrayValue: StArray?,
                        val length: Int?,            // for arrays: the number of elements, for strings: number of characters *including* the terminating 0-byte
                        val zpwish: ZeropageWish,    // used in the variable allocator
+                       val align: Int,
                        astNode: PtNode) : StNode(name, StNodeType.STATICVAR, astNode) {
 
-    val uninitialized = onetimeInitializationArrayValue==null && onetimeInitializationStringValue==null && onetimeInitializationNumericValue==null
+    var initializationNumericValue: Double? = null
+        private set
+
+    fun setOnetimeInitNumeric(number: Double) {
+        // In certain cases the init value of an existing var should be updated,
+        // so we can't ask this as a constructor parameter.
+        // This has to do with the way Prog8 does the (re)initialization of such variables: via code assignment statements.
+        // Certain codegens might want to put them back into the variable directly.
+        // For strings and arrays this doesn't occur - these are always already specced at creation time.
+        initializationNumericValue = number
+    }
+
+    val uninitialized: Boolean
+        get() = initializationArrayValue==null && initializationStringValue==null && initializationNumericValue==null
 
     init {
         if(length!=null) {
-            require(onetimeInitializationNumericValue == null)
-            if(onetimeInitializationArrayValue!=null)
-                require(onetimeInitializationArrayValue.isEmpty() ||onetimeInitializationArrayValue.size==length)
+            require(initializationNumericValue == null)
+            if(initializationArrayValue!=null)
+                require(initializationArrayValue.isEmpty() ||initializationArrayValue.size==length)
         }
-        if(onetimeInitializationNumericValue!=null) {
+        if(initializationNumericValue!=null) {
             require(dt in NumericDatatypes || dt==DataType.BOOL)
         }
-        if(onetimeInitializationArrayValue!=null) {
+        if(initializationArrayValue!=null) {
             require(dt in ArrayDatatypes)
-            require(length==onetimeInitializationArrayValue.size)
+            require(length==initializationArrayValue.size)
         }
-        if(onetimeInitializationStringValue!=null) {
+        if(initializationStringValue!=null) {
             require(dt == DataType.STR)
-            require(length == onetimeInitializationStringValue.first.length+1)
+            require(length == initializationStringValue.first.length+1)
+        }
+        if(align > 0) {
+            require(dt == DataType.STR || dt in ArrayDatatypes)
+            require(zpwish != ZeropageWish.REQUIRE_ZEROPAGE && zpwish != ZeropageWish.PREFER_ZEROPAGE)
         }
     }
 }
@@ -239,18 +257,24 @@ class StSub(name: String, val parameters: List<StSubroutineParameter>, val retur
         StNode(name, StNodeType.SUBROUTINE, astNode)
 
 
-class StRomSub(name: String,
-               val address: UInt?,      // null in case of asmsub, specified in case of romsub
-               val parameters: List<StRomSubParameter>,
-               val returns: List<StRomSubParameter>,
+class StExtSub(name: String,
+               val address: PtAsmSub.Address?,      // null in case of asmsub, specified in case of extsub.
+               val parameters: List<StExtSubParameter>,
+               val returns: List<StExtSubParameter>,
                astNode: PtNode) :
-    StNode(name, StNodeType.ROMSUB, astNode)
+    StNode(name, StNodeType.EXTSUB, astNode)
 
 
 
 class StSubroutineParameter(val name: String, val type: DataType)
-class StRomSubParameter(val register: RegisterOrStatusflag, val type: DataType)
-class StArrayElement(val number: Double?, val addressOfSymbol: String?, val boolean: Boolean?)
+class StExtSubParameter(val register: RegisterOrStatusflag, val type: DataType)
+class StArrayElement(val number: Double?, val addressOfSymbol: String?, val boolean: Boolean?) {
+    init {
+        if(number!=null) require(addressOfSymbol==null && boolean==null)
+        if(addressOfSymbol!=null) require(number==null && boolean==null)
+        if(boolean!=null) require(addressOfSymbol==null && number==null)
+    }
+}
 
 typealias StString = Pair<String, Encoding>
 typealias StArray = List<StArrayElement>

codeCore/src/prog8/code/SymbolTableMaker.kt

@@ -3,7 +3,7 @@ package prog8.code
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.code.target.VMTarget
-import java.util.*
+
 
 class SymbolTableMaker(private val program: PtProgram, private val options: CompilationOptions) {
     fun make(): SymbolTable {
@@ -13,8 +13,8 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
             st.add(StNode(it.key, StNodeType.BUILTINFUNC, PtIdentifier(it.key, it.value.returnType ?: DataType.UNDEFINED, Position.DUMMY)))
         }
 
-        val scopestack = Stack<StNode>()
-        scopestack.push(st)
+        val scopestack = ArrayDeque<StNode>()
+        scopestack.add(st)
         program.children.forEach {
             addToSt(it, scopestack)
         }
@@ -35,12 +35,12 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
         return st
     }
 
-    private fun addToSt(node: PtNode, scope: Stack<StNode>) {
+    private fun addToSt(node: PtNode, scope: ArrayDeque<StNode>) {
         val stNode = when(node) {
             is PtAsmSub -> {
-                val parameters = node.parameters.map { StRomSubParameter(it.first, it.second.type) }
-                val returns = node.returns.map { StRomSubParameter(it.first, it.second) }
-                StRomSub(node.name, node.address, parameters, returns, node)
+                val parameters = node.parameters.map { StExtSubParameter(it.first, it.second.type) }
+                val returns = node.returns.map { StExtSubParameter(it.first, it.second) }
+                StExtSub(node.name, node.address, parameters, returns, node)
             }
             is PtBlock -> {
                 StNode(node.name, StNodeType.BLOCK, node)
@@ -65,23 +65,26 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                 val numElements: Int?
                 val value = node.value
                 if(value!=null) {
-                    val number = (value as? PtNumber)?.number
-                    initialNumeric = if(number==0.0) null else number       // 0 as init value -> just uninitialized
                     when (value) {
                         is PtString -> {
                             initialString = StString(value.value, value.encoding)
                             initialArray = null
+                            initialNumeric = null
                             numElements = value.value.length + 1   // include the terminating 0-byte
                         }
                         is PtArray -> {
                             initialArray = makeInitialArray(value)
                             initialString = null
+                            initialNumeric = null
                             numElements = initialArray.size
                             require(node.arraySize?.toInt()==numElements)
                         }
                         else -> {
+                            require(value is PtNumber)
                             initialString = null
                             initialArray = null
+                            val number = value.number
+                            initialNumeric = number
                             numElements = node.arraySize?.toInt()
                         }
                     }
@@ -94,7 +97,10 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
 //                if(node.type in SplitWordArrayTypes) {
 //                    ... split array also add _lsb and _msb to symboltable?
 //                }
-                StStaticVariable(node.name, node.type, initialNumeric, initialString, initialArray, numElements, node.zeropage, node)
+                val stVar = StStaticVariable(node.name, node.type, initialString, initialArray, numElements, node.zeropage, node.align.toInt(), node)
+                if(initialNumeric!=null)
+                    stVar.setOnetimeInitNumeric(initialNumeric)
+                stVar
             }
             is PtBuiltinFunctionCall -> {
                 if(node.name=="memory") {
@@ -104,7 +110,7 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                     val size = (node.args[1] as PtNumber).number.toUInt()
                     val align = (node.args[2] as PtNumber).number.toUInt()
                     // don't add memory slabs in nested scope, just put them in the top level of the ST
-                    scope.firstElement().add(StMemorySlab("prog8_memoryslab_$slabname", size, align, node))
+                    scope.first().add(StMemorySlab("prog8_memoryslab_$slabname", size, align, node))
                 }
                 null
             }
@@ -112,14 +118,14 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
         }
 
         if(stNode!=null) {
-            scope.peek().add(stNode)
-            scope.push(stNode)
+            scope.last().add(stNode)
+            scope.add(stNode)
         }
         node.children.forEach {
             addToSt(it, scope)
         }
         if(stNode!=null)
-            scope.pop()
+            scope.removeLast()
     }
 
     private fun makeInitialArray(value: PtArray): List<StArrayElement> {
@@ -130,7 +136,6 @@ class SymbolTableMaker(private val program: PtProgram, private val options: Comp
                         TODO("address-of array element $it in initial array value")
                     StArrayElement(null, it.identifier.name, null)
                 }
-                is PtIdentifier -> StArrayElement(null, it.name, null)
                 is PtNumber -> StArrayElement(it.number, null, null)
                 is PtBool -> StArrayElement(null, null, it.value)
                 else -> throw AssemblyError("invalid array element $it")

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

@@ -80,18 +80,11 @@ class PtBlock(name: String,
               val options: Options,
               position: Position
 ) : PtNamedNode(name, position), IPtStatementContainer {
-    enum class BlockAlignment {
-        NONE,
-        WORD,
-        PAGE
-    }
-
     class Options(val address: UInt? = null,
                   val forceOutput: Boolean = false,
                   val noSymbolPrefixing: Boolean = false,
                   val veraFxMuls: Boolean = false,
-                  val ignoreUnused: Boolean = false,
-                  val alignment: BlockAlignment = BlockAlignment.NONE)
+                  val ignoreUnused: Boolean = false)
 }
 
 
@@ -103,12 +96,20 @@ class PtInlineAssembly(val assembly: String, val isIR: Boolean, position: Positi
 }
 
 
-class PtLabel(name: String, position: Position) : PtNamedNode(name, position)
+class PtLabel(name: String, position: Position) : PtNamedNode(name, position) {
+    companion object {
+        // all automatically generated labels everywhere need to have the same label name prefix:
+        const val GeneratedLabelPrefix = "p8_label_gen_"
+    }
+}
 
 
 class PtBreakpoint(position: Position): PtNode(position)
 
 
+class PtAlign(val align: UInt, position: Position): PtNode(position)
+
+
 class PtIncludeBinary(val file: Path, val offset: UInt?, val length: UInt?, position: Position) : PtNode(position)
 
 

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

@@ -1,7 +1,7 @@
 package prog8.code.ast
 
 import prog8.code.core.*
-import java.util.*
+import java.util.Objects
 import kotlin.math.abs
 import kotlin.math.truncate
 
@@ -42,7 +42,18 @@ sealed class PtExpression(val type: DataType, position: Position) : PtNode(posit
                 else
                     other.left isSameAs left && other.right isSameAs right
             }
-            is PtContainmentCheck -> other is PtContainmentCheck && other.type==type && other.element isSameAs element && other.iterable isSameAs iterable
+            is PtContainmentCheck -> {
+                if(other !is PtContainmentCheck || other.type != type || !(other.needle isSameAs needle))
+                    false
+                else {
+                    if(haystackHeapVar!=null)
+                        other.haystackHeapVar!=null && other.haystackHeapVar!! isSameAs haystackHeapVar!!
+                    else if(haystackValues!=null)
+                        other.haystackValues!=null && other.haystackValues!! isSameAs haystackValues!!
+                    else
+                        false
+                }
+            }
             is PtIdentifier -> other is PtIdentifier && other.type==type && other.name==name
             is PtIrRegister -> other is PtIrRegister && other.type==type && other.register==register
             is PtMemoryByte -> other is PtMemoryByte && other.address isSameAs address
@@ -71,10 +82,11 @@ sealed class PtExpression(val type: DataType, position: Position) : PtNode(posit
     }
 
     fun asConstInteger(): Int? = (this as? PtNumber)?.number?.toInt() ?: (this as? PtBool)?.asInt()
+    fun asConstValue(): Double? = (this as? PtNumber)?.number ?: (this as? PtBool)?.asInt()?.toDouble()
 
     fun isSimple(): Boolean {
         return when(this) {
-            is PtAddressOf -> true
+            is PtAddressOf -> this.arrayIndexExpr==null || this.arrayIndexExpr?.isSimple()==true
             is PtArray -> true
             is PtArrayIndexer -> index is PtNumber || index is PtIdentifier
             is PtBinaryExpression -> false
@@ -95,6 +107,7 @@ sealed class PtExpression(val type: DataType, position: Position) : PtNode(posit
             is PtRange -> true
             is PtString -> true
             is PtTypeCast -> value.isSimple()
+            is PtIfExpression -> condition.isSimple() && truevalue.isSimple() && falsevalue.isSimple()
         }
     }
 
@@ -152,6 +165,7 @@ class PtArrayIndexer(elementType: DataType, position: Position): PtExpression(el
 
 
 class PtArray(type: DataType, position: Position): PtExpression(type, position) {
+    // children are always one of 3 types: PtBool, PtNumber or PtAddressOf.
     override fun hashCode(): Int = Objects.hash(children, type)
     override fun equals(other: Any?): Boolean {
         if(other==null || other !is PtArray)
@@ -194,11 +208,28 @@ class PtBinaryExpression(val operator: String, type: DataType, position: Positio
 }
 
 
-class PtContainmentCheck(position: Position): PtExpression(DataType.BOOL, position) {
-    val element: PtExpression
+class PtIfExpression(type: DataType, position: Position): PtExpression(type, position) {
+    val condition: PtExpression
         get() = children[0] as PtExpression
-    val iterable: PtIdentifier
-        get() = children[1] as PtIdentifier
+    val truevalue: PtExpression
+        get() = children[1] as PtExpression
+    val falsevalue: PtExpression
+        get() = children[2] as PtExpression
+}
+
+
+class PtContainmentCheck(position: Position): PtExpression(DataType.BOOL, position) {
+    val needle: PtExpression
+        get() = children[0] as PtExpression
+    val haystackHeapVar: PtIdentifier?
+        get() = children[1] as? PtIdentifier
+    val haystackValues: PtArray?
+        get() = children[1] as? PtArray
+
+    companion object {
+        val MAX_SIZE_FOR_INLINE_CHECKS_BYTE = 5
+        val MAX_SIZE_FOR_INLINE_CHECKS_WORD = 4
+    }
 }
 
 
@@ -208,6 +239,13 @@ class PtFunctionCall(val name: String,
                      position: Position) : PtExpression(type, position) {
     val args: List<PtExpression>
         get() = children.map { it as PtExpression }
+
+    init {
+        if(void) require(type==DataType.UNDEFINED) {
+            "void fcall should have undefined datatype"
+        }
+        // note: non-void calls can have UNDEFINED type: is if they return more than 1 value
+    }
 }
 
 
@@ -227,12 +265,6 @@ class PtMemoryByte(position: Position) : PtExpression(DataType.UBYTE, position)
 
 
 class PtBool(val value: Boolean, position: Position) : PtExpression(DataType.BOOL, position) {
-
-    companion object {
-        fun fromNumber(number: Number, position: Position): PtBool =
-            PtBool(number != 0.0, position)
-    }
-
     override fun hashCode(): Int = Objects.hash(type, value)
 
     override fun equals(other: Any?): Boolean {
@@ -302,6 +334,7 @@ class PtRange(type: DataType, position: Position) : PtExpression(type, position)
     fun toConstantIntegerRange(): IntProgression? {
         fun makeRange(fromVal: Int, toVal: Int, stepVal: Int): IntProgression {
             return when {
+                fromVal == toVal -> fromVal .. toVal
                 fromVal <= toVal -> when {
                     stepVal <= 0 -> IntRange.EMPTY
                     stepVal == 1 -> fromVal..toVal
@@ -317,12 +350,11 @@ class PtRange(type: DataType, position: Position) : PtExpression(type, position)
 
         val fromLv = from as? PtNumber
         val toLv = to as? PtNumber
-        val stepLv = step as? PtNumber
-        if(fromLv==null || toLv==null || stepLv==null)
+        if(fromLv==null || toLv==null)
             return null
         val fromVal = fromLv.number.toInt()
         val toVal = toLv.number.toInt()
-        val stepVal = stepLv.number.toInt()
+        val stepVal = step.number.toInt()
         return makeRange(fromVal, toVal, stepVal)
     }
 }
@@ -346,7 +378,3 @@ class PtTypeCast(type: DataType, position: Position) : PtExpression(type, positi
 
 // special node that isn't created from compiling user code, but used internally in the Intermediate Code
 class PtIrRegister(val register: Int, type: DataType, position: Position) : PtExpression(type, position)
-
-
-fun constValue(expr: PtExpression): Double? = if(expr is PtNumber) expr.number else if(expr is PtBool) expr.asInt().toDouble() else null
-fun constIntValue(expr: PtExpression): Int? = if(expr is PtNumber) expr.number.toInt() else if(expr is PtBool) expr.asInt() else null

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

@@ -10,6 +10,7 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
     fun type(dt: DataType) = "!${dt.name.lowercase()}!"
     fun txt(node: PtNode): String {
         return when(node) {
+            is PtAlign -> "%align ${node.align}"
             is PtAssignTarget -> if(node.void) "<void>" else "<target>"
             is PtAssignment -> "<assign>"
             is PtAugmentedAssign -> "<inplace-assign> ${node.operator}"
@@ -21,7 +22,17 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
                 else
                     "&"
             }
-            is PtArray -> "array len=${node.children.size} ${type(node.type)}"
+            is PtArray -> {
+                val valuelist = node.children.map {
+                    when (it) {
+                        is PtBool -> it.toString()
+                        is PtNumber -> it.number.toString()
+                        is PtIdentifier -> it.name
+                        else -> "?"
+                    }
+                }.joinToString(", ")
+                "array len=${node.children.size} ${type(node.type)} [ $valuelist ]"
+            }
             is PtArrayIndexer -> "<arrayindexer> ${type(node.type)} ${if(node.splitWords) "[splitwords]" else ""}"
             is PtBinaryExpression -> "<expr> ${node.operator} ${type(node.type)}"
             is PtBuiltinFunctionCall -> {
@@ -78,16 +89,18 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
                     }"
                 }
                 val str = if (node.inline) "inline " else ""
-                if(node.address==null) {
+                if(node.address == null) {
                     str + "asmsub ${node.name}($params) $clobbers $returns"
                 } else {
-                    str + "romsub ${node.address.toHex()} = ${node.name}($params) $clobbers $returns"
+                    val bank = if(node.address.constbank!=null) "@bank ${node.address.constbank}"
+                        else if(node.address.varbank!=null) "@bank ${node.address.varbank?.name}"
+                        else ""
+                    str + "extsub $bank ${node.address.address.toHex()} = ${node.name}($params) $clobbers $returns"
                 }
             }
             is PtBlock -> {
                 val addr = if(node.options.address==null) "" else "@${node.options.address.toHex()}"
-                val align = if(node.options.alignment==PtBlock.BlockAlignment.NONE) "" else "align=${node.options.alignment}"
-                "\nblock '${node.name}' $addr $align"
+                "\nblock '${node.name}' $addr"
             }
             is PtConstant -> {
                 val value = when(node.type) {
@@ -116,13 +129,20 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
             }
             is PtVariable -> {
                 val split = if(node.type in SplitWordArrayTypes) "@split" else ""
+                val align = when(node.align) {
+                    0u -> ""
+                    2u -> "@alignword"
+                    64u -> "@align64"
+                    256u -> "@alignpage"
+                    else -> throw IllegalArgumentException("invalid alignment size")
+                }
                 val str = if(node.arraySize!=null) {
                     val eltType = ArrayToElementTypes.getValue(node.type)
-                    "${eltType.name.lowercase()}[${node.arraySize}] $split ${node.name}"
+                    "${eltType.name.lowercase()}[${node.arraySize}] $split $align ${node.name}"
                 }
                 else if(node.type in ArrayDatatypes) {
                     val eltType = ArrayToElementTypes.getValue(node.type)
-                    "${eltType.name.lowercase()}[] $split ${node.name}"
+                    "${eltType.name.lowercase()}[] $split $align ${node.name}"
                 }
                 else
                     "${node.type.name.lowercase()} ${node.name}"
@@ -144,6 +164,8 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
                 else
                     "->"
             }
+            is PtDefer -> "<defer>"
+            is PtIfExpression -> "<ifexpr>"
         }
     }
 

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

@@ -5,17 +5,21 @@ import prog8.code.core.*
 
 sealed interface IPtSubroutine {
     val name: String
+    val scopedName: String
 }
 
 class PtAsmSub(
     name: String,
-    val address: UInt?,
+    val address: Address?,
     val clobbers: Set<CpuRegister>,
     val parameters: List<Pair<RegisterOrStatusflag, PtSubroutineParameter>>,
     val returns: List<Pair<RegisterOrStatusflag, DataType>>,
     val inline: Boolean,
     position: Position
-) : PtNamedNode(name, position), IPtSubroutine
+) : PtNamedNode(name, position), IPtSubroutine {
+
+    class Address(val constbank: UByte?, var varbank: PtIdentifier?, val address: UInt)
+}
 
 
 class PtSub(
@@ -130,7 +134,8 @@ class PtRepeatLoop(position: Position) : PtNode(position) {
 
 
 class PtReturn(position: Position) : PtNode(position) {
-    val hasValue = children.any()
+    val hasValue: Boolean
+        get() = children.any()
     val value: PtExpression?
         get() {
             return if(children.any())
@@ -147,7 +152,15 @@ sealed interface IPtVariable {
 }
 
 
-class PtVariable(name: String, override val type: DataType, val zeropage: ZeropageWish, val value: PtExpression?, val arraySize: UInt?, position: Position) : PtNamedNode(name, position), IPtVariable {
+class PtVariable(
+    name: String,
+    override val type: DataType,
+    val zeropage: ZeropageWish,
+    val align: UInt,
+    val value: PtExpression?,
+    val arraySize: UInt?,
+    position: Position
+) : PtNamedNode(name, position), IPtVariable {
     init {
         value?.let {it.parent=this}
     }
@@ -178,3 +191,6 @@ class PtWhenChoice(val isElse: Boolean, position: Position) : PtNode(position) {
     val statements: PtNodeGroup
         get() = children[1] as PtNodeGroup
 }
+
+
+class PtDefer(position: Position): PtNode(position), IPtStatementContainer

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

@@ -0,0 +1,16 @@
+package prog8.code.ast
+
+import prog8.code.SymbolTable
+import prog8.code.core.*
+
+fun verifyFinalAstBeforeAsmGen(program: PtProgram, options: CompilationOptions, st: SymbolTable, errors: IErrorReporter) {
+/*
+    walkAst(program) { node, _ ->
+        if(node is PtVariable) {
+            if(node.value!=null) {
+                // require(node.type in ArrayDatatypes || node.type==DataType.STR) { "final check before asmgen: only string and array variables can still have an init value ${node.name} ${node.type} ${node.position}"}
+            }
+        }
+    }
+*/
+}

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

@@ -49,11 +49,14 @@ class FSignature(val pure: Boolean,      // does it have side effects?
         return when {
             actualParamTypes.isEmpty() -> CallConvention(emptyList(), returns)
             actualParamTypes.size==1 -> {
-                // one parameter goes via register/registerpair
+                // One parameter goes via register/registerpair.
+                // this avoids repeated code for every caller to store the value in the subroutine's argument variable.
+                // (that store is still done, but only coded once at the start at the subroutine itself rather than at every call site).
+                // TODO can we start using the X register as well for a third byte or word+byte combo
                 val paramConv = when(val paramType = actualParamTypes[0]) {
                     DataType.UBYTE, DataType.BYTE -> ParamConvention(paramType, RegisterOrPair.A, false)
                     DataType.UWORD, DataType.WORD -> ParamConvention(paramType, RegisterOrPair.AY, false)
-                    DataType.FLOAT -> ParamConvention(paramType, RegisterOrPair.AY, false)
+                    DataType.FLOAT -> ParamConvention(paramType, RegisterOrPair.AY, false)      // TODO is this correct? shouldn't it be FAC1?
                     in PassByReferenceDatatypes -> ParamConvention(paramType, RegisterOrPair.AY, false)
                     else -> ParamConvention(paramType, null, false)
                 }
@@ -79,7 +82,6 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     // cmp returns a status in the carry flag, but not a proper return value
     "cmp"       to FSignature(false, listOf(FParam("value1", IntegerDatatypes), FParam("value2", NumericDatatypes)), null),
     "prog8_lib_stringcompare"     to FSignature(true, listOf(FParam("str1", arrayOf(DataType.STR)), FParam("str2", arrayOf(DataType.STR))), DataType.BYTE),
-    "prog8_lib_arraycopy"         to FSignature(false, listOf(FParam("source", ArrayDatatypes), FParam("target", ArrayDatatypes)), null),
     "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),
     "prog8_ifelse_bittest_set"    to FSignature(true, listOf(FParam("variable", ByteDatatypes), FParam("bitnumber", arrayOf(DataType.UBYTE))), DataType.BOOL),
@@ -128,6 +130,7 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "rrestore"  to FSignature(false, emptyList(), null),
     "memory"    to FSignature(true, listOf(FParam("name", arrayOf(DataType.STR)), FParam("size", arrayOf(DataType.UWORD)), FParam("alignment", arrayOf(DataType.UWORD))), DataType.UWORD),
     "callfar"   to FSignature(false, listOf(FParam("bank", arrayOf(DataType.UBYTE)), FParam("address", arrayOf(DataType.UWORD)), FParam("arg", arrayOf(DataType.UWORD))), DataType.UWORD),
+    "callfar2"  to FSignature(false, listOf(FParam("bank", arrayOf(DataType.UBYTE)), FParam("address", arrayOf(DataType.UWORD)), FParam("argA", arrayOf(DataType.UBYTE)), FParam("argX", arrayOf(DataType.UBYTE)), FParam("argY", arrayOf(DataType.UBYTE)), FParam("argC", arrayOf(DataType.BOOL))), DataType.UWORD),
     "call"      to FSignature(false, listOf(FParam("address", arrayOf(DataType.UWORD))), DataType.UWORD),
 )
 

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

@@ -14,6 +14,7 @@ class CompilationOptions(val output: OutputType,
                          val compTarget: ICompilationTarget,
                          // these are set later, based on command line arguments or options in the source code:
                          var loadAddress: UInt,
+                         var memtopAddress: UInt,
                          var warnSymbolShadowing: Boolean = false,
                          var optimize: Boolean = false,
                          var asmQuiet: Boolean = false,
@@ -27,7 +28,7 @@ class CompilationOptions(val output: OutputType,
                          var slabsHighBank: Int? = null,
                          var slabsGolden: Boolean = false,
                          var splitWordArrays: Boolean = false,
-                         var strictBool: Boolean = true,
+                         var addMissingRts: Boolean = false,        // deprecated, will likely go way in future version
                          var breakpointCpuInstruction: String? = null,
                          var outputDir: Path = Path(""),
                          var symbolDefs: Map<String, String> = emptyMap()

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

@@ -13,7 +13,9 @@ interface IMachineDefinition {
     val FLOAT_MAX_NEGATIVE: Double
     val FLOAT_MAX_POSITIVE: Double
     val FLOAT_MEM_SIZE: Int
+    val STARTUP_CODE_RESERVED_SIZE: UInt        // this is here, so that certain compiler targets are able to tune this
     val PROGRAM_LOAD_ADDRESS : UInt
+    val PROGRAM_MEMTOP_ADDRESS: UInt
     val BSSHIGHRAM_START: UInt
     val BSSHIGHRAM_END: UInt
     val BSSGOLDENRAM_START: UInt

codeCore/src/prog8/code/optimize/Optimizer.kt

@@ -1,6 +1,6 @@
 package prog8.code.optimize
 
-import prog8.code.StRomSub
+import prog8.code.StExtSub
 import prog8.code.SymbolTable
 import prog8.code.ast.*
 import prog8.code.core.*
@@ -10,9 +10,8 @@ fun optimizeIntermediateAst(program: PtProgram, options: CompilationOptions, st:
     if (!options.optimize)
         return
     while (errors.noErrors() &&
-        (optimizeCommonSubExpressions(program, errors)
-                + optimizeBitTest(program, options)
-                + optimizeAssignTargets(program, st, errors)) > 0
+            (optimizeBitTest(program, options)
+            + optimizeAssignTargets(program, st, errors)) > 0
     ) {
         // keep rolling
     }
@@ -28,97 +27,6 @@ private fun walkAst(root: PtNode, act: (node: PtNode, depth: Int) -> Boolean) {
 }
 
 
-private var tempVarCounter = 0
-
-private fun optimizeCommonSubExpressions(program: PtProgram, errors: IErrorReporter): Int {
-
-    fun extractableSubExpr(expr: PtExpression): Boolean {
-        if(expr is PtArrayIndexer && expr.index.isSimple())
-            return false
-        if (expr is PtMemoryByte && expr.address.isSimple())
-            return false
-
-        val result = if(expr is PtBinaryExpression)
-            expr.type !in ByteDatatypes ||
-            !(expr.left.isSimple() && expr.right.isSimple()) ||
-            (expr.operator !in LogicalOperators && expr.operator !in BitwiseOperators)
-        else if (expr is PtArrayIndexer && expr.type !in ByteDatatypes)
-            true
-        else
-            !expr.isSimple()
-        return result
-    }
-
-    // for each Binaryexpression, recurse to find a common subexpression pair therein.
-    val commons = mutableMapOf<PtBinaryExpression, Pair<PtExpression, PtExpression>>()
-    walkAst(program) { node: PtNode, depth: Int ->
-        if(node is PtBinaryExpression) {
-            val subExpressions = mutableListOf<PtExpression>()
-            walkAst(node.left) { subNode: PtNode, subDepth: Int ->
-                if (subNode is PtExpression) {
-                    if(extractableSubExpr(subNode)) subExpressions.add(subNode)
-                    true
-                } else false
-            }
-            walkAst(node.right) { subNode: PtNode, subDepth: Int ->
-                if (subNode is PtExpression) {
-                    if(extractableSubExpr(subNode)) subExpressions.add(subNode)
-                    true
-                } else false
-            }
-
-            outer@for (first in subExpressions) {
-                for (second in subExpressions) {
-                    if (first!==second && first isSameAs second) {
-                        commons[node] = first to second
-                        break@outer     // do only 1 replacement at a time per binaryexpression
-                    }
-                }
-            }
-            false
-        } else true
-    }
-
-    // replace common subexpressions by a temp variable that is assigned only once.
-    // TODO: check for commonalities across multiple separate expressions in the current scope, not only inside a single line
-    commons.forEach { binexpr, (occurrence1, occurrence2) ->
-        val (stmtContainer, stmt) = findContainingStatements(binexpr)
-        val occurrence1idx = occurrence1.parent.children.indexOf(occurrence1)
-        val occurrence2idx = occurrence2.parent.children.indexOf(occurrence2)
-        val containerScopedName = findScopeName(stmtContainer)
-        tempVarCounter++
-        val tempvarName = "prog8_subexprvar_$tempVarCounter"
-        // TODO: some tempvars could be reused, if they are from different lines
-
-        val datatype = occurrence1.type
-        val singleReplacement1 = PtIdentifier("$containerScopedName.$tempvarName", datatype, occurrence1.position)
-        val singleReplacement2 = PtIdentifier("$containerScopedName.$tempvarName", datatype, occurrence2.position)
-        occurrence1.parent.children[occurrence1idx] = singleReplacement1
-        singleReplacement1.parent = occurrence1.parent
-        occurrence2.parent.children[occurrence2idx] = singleReplacement2
-        singleReplacement2.parent = occurrence2.parent
-
-        val tempassign = PtAssignment(binexpr.position).also { assign ->
-            assign.add(PtAssignTarget(false, binexpr.position).also { tgt->
-                tgt.add(PtIdentifier("$containerScopedName.$tempvarName", datatype, binexpr.position))
-            })
-            assign.add(occurrence1)
-            occurrence1.parent = assign
-        }
-        stmtContainer.children.add(stmtContainer.children.indexOf(stmt), tempassign)
-        tempassign.parent = stmtContainer
-
-        val tempvar = PtVariable(tempvarName, datatype, ZeropageWish.NOT_IN_ZEROPAGE, null, null, binexpr.position)
-        stmtContainer.add(0, tempvar)
-        tempvar.parent = stmtContainer
-
-        // errors.info("common subexpressions replaced by a tempvar, maybe simplify the expression manually", binexpr.position)
-    }
-
-    return commons.size
-}
-
-
 private fun optimizeAssignTargets(program: PtProgram, st: SymbolTable, errors: IErrorReporter): Int {
     var changes = 0
     walkAst(program) { node: PtNode, depth: Int ->
@@ -131,7 +39,7 @@ private fun optimizeAssignTargets(program: PtProgram, st: SymbolTable, errors: I
             }
             if(functionName!=null) {
                 val stNode = st.lookup(functionName)
-                if (stNode is StRomSub) {
+                if (stNode is StExtSub) {
                     require(node.children.size==stNode.returns.size+1) {
                         "number of targets must match return values"
                     }
@@ -152,7 +60,7 @@ private fun optimizeAssignTargets(program: PtProgram, st: SymbolTable, errors: I
                 if(node.children.dropLast(1).all { (it as PtAssignTarget).void }) {
                     // all targets are now void, the whole assignment can be discarded and replaced by just a (void) call to the subroutine
                     val index = node.parent.children.indexOf(node)
-                    val voidCall = PtFunctionCall(functionName, true, value.type, value.position)
+                    val voidCall = PtFunctionCall(functionName, true, DataType.UNDEFINED, value.position)
                     value.children.forEach { voidCall.add(it) }
                     node.parent.children[index] = voidCall
                     voidCall.parent = node.parent
@@ -170,38 +78,73 @@ private fun optimizeBitTest(program: PtProgram, options: CompilationOptions): In
     if(options.compTarget.machine.cpu == CpuType.VIRTUAL)
         return 0        // the special bittest optimization is not yet valid for the IR
 
+    fun makeBittestCall(condition: PtBinaryExpression, and: PtBinaryExpression, variable: PtIdentifier, bitmask: Int): PtBuiltinFunctionCall {
+        require(bitmask==128 || bitmask==64)
+        val setOrNot = if(condition.operator=="!=") "set" else "notset"
+        val bittestCall = PtBuiltinFunctionCall("prog8_ifelse_bittest_$setOrNot", false, true, DataType.BOOL, condition.position)
+        bittestCall.add(variable)
+        if(bitmask==128)
+            bittestCall.add(PtNumber(DataType.UBYTE, 7.0, and.right.position))
+        else
+            bittestCall.add(PtNumber(DataType.UBYTE, 6.0, and.right.position))
+        return bittestCall
+    }
+
+    fun isAndByteCondition(condition: PtBinaryExpression?): Triple<PtBinaryExpression, PtIdentifier, Int>? {
+        if(condition!=null && (condition.operator=="==" || condition.operator=="!=")) {
+            if (condition.right.asConstInteger() == 0) {
+                val and = condition.left as? PtBinaryExpression
+                if (and != null && and.operator == "&" && and.type == DataType.UBYTE) {
+                    val bitmask = and.right.asConstInteger()
+                    if(bitmask==128 || bitmask==64) {
+                        val variable = and.left as? PtIdentifier
+                        if (variable != null && variable.type in ByteDatatypes) {
+                            return Triple(and, variable, bitmask)
+                        }
+                        val typecast = and.left as? PtTypeCast
+                        if (typecast != null && typecast.type == DataType.UBYTE) {
+                            val castedVariable = typecast.value as? PtIdentifier
+                            if(castedVariable!=null && castedVariable.type in ByteDatatypes)
+                                return Triple(and, castedVariable, bitmask)
+                        }
+                    }
+                }
+            }
+        }
+        return null
+    }
+
     var changes = 0
     var recurse = true
     walkAst(program) { node: PtNode, depth: Int ->
         if(node is PtIfElse) {
             val condition = node.condition as? PtBinaryExpression
-            if(condition!=null && (condition.operator=="==" || condition.operator=="!=")) {
-                if(condition.right.asConstInteger()==0) {
-                    val and = condition.left as? PtBinaryExpression
-                    if(and != null && and.operator=="&" && and.type == DataType.UBYTE) {
-                        val variable = and.left as? PtIdentifier
-                        val bitmask = and.right.asConstInteger()
-                        if(variable!=null && variable.type in ByteDatatypes && (bitmask==128 || bitmask==64)) {
-                            val setOrNot = if(condition.operator=="!=") "set" else "notset"
-                            val index = node.parent.children.indexOf(node)
-                            val bittestCall = PtBuiltinFunctionCall("prog8_ifelse_bittest_$setOrNot", false, true, DataType.BOOL, node.condition.position)
-                            bittestCall.add(variable)
-                            if(bitmask==128)
-                                bittestCall.add(PtNumber(DataType.UBYTE, 7.0, and.right.position))
-                            else
-                                bittestCall.add(PtNumber(DataType.UBYTE, 6.0, and.right.position))
-                            val ifElse = PtIfElse(node.position)
-                            ifElse.add(bittestCall)
-                            ifElse.add(node.ifScope)
-                            if(node.hasElse())
-                                ifElse.add(node.elseScope)
-                            node.parent.children[index] = ifElse
-                            ifElse.parent = node.parent
-                            changes++
-                            recurse = false
-                        }
-                    }
-                }
+            val check = isAndByteCondition(condition)
+            if(check!=null) {
+                val (and, variable, bitmask) = check
+                val bittestCall = makeBittestCall(condition!!, and, variable, bitmask)
+                val ifElse = PtIfElse(node.position)
+                ifElse.add(bittestCall)
+                ifElse.add(node.ifScope)
+                if (node.hasElse())
+                    ifElse.add(node.elseScope)
+                val index = node.parent.children.indexOf(node)
+                node.parent.children[index] = ifElse
+                ifElse.parent = node.parent
+                changes++
+                recurse = false
+            }
+        }
+        if (node is PtIfExpression) {
+            val condition = node.condition as? PtBinaryExpression
+            val check = isAndByteCondition(condition)
+            if(check!=null) {
+                val (and, variable, bitmask) = check
+                val bittestCall = makeBittestCall(condition!!, and, variable, bitmask)
+                node.children[0] = bittestCall
+                bittestCall.parent = node
+                changes++
+                recurse = false
             }
         }
         recurse
@@ -235,24 +178,3 @@ internal fun isSame(identifier: PtIdentifier, type: DataType, returnedRegister:
     }
     return false   // there are no identifiers directly corresponding to cpu registers
 }
-
-
-internal fun findScopeName(node: PtNode): String {
-    var parent=node
-    while(parent !is PtNamedNode)
-        parent = parent.parent
-    return parent.scopedName
-}
-
-
-internal fun findContainingStatements(node: PtNode): Pair<PtNode, PtNode> {      // returns (parentstatementcontainer, childstatement)
-    var parent = node.parent
-    var child = node
-    while(true) {
-        if(parent is IPtStatementContainer) {
-            return parent to child
-        }
-        child=parent
-        parent=parent.parent
-    }
-}

codeCore/src/prog8/code/target/C64Target.kt

@@ -27,6 +27,7 @@ val CompilationTargets = listOf(
     Cx16Target.NAME,
     PETTarget.NAME,
     AtariTarget.NAME,
+    Neo6502Target.NAME,
     VMTarget.NAME
 )
 
@@ -37,5 +38,6 @@ fun getCompilationTargetByName(name: String) = when(name.lowercase()) {
     PETTarget.NAME -> PETTarget()
     AtariTarget.NAME -> AtariTarget()
     VMTarget.NAME -> VMTarget()
+    Neo6502Target.NAME -> Neo6502Target()
     else -> throw IllegalArgumentException("invalid compilation target")
-}
+}

codeCore/src/prog8/code/target/Neo6502Target.kt

@@ -0,0 +1,30 @@
+package prog8.code.target
+
+import prog8.code.core.*
+import prog8.code.target.neo6502.Neo6502MachineDefinition
+
+
+class Neo6502Target: ICompilationTarget, IStringEncoding by Encoder, IMemSizer {
+    override val name = NAME
+    override val machine = Neo6502MachineDefinition()
+    override val defaultEncoding = Encoding.ISO
+
+    companion object {
+        const val NAME = "neo"
+    }
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypesWithBoolean -> 1
+            in WordDatatypes, in PassByReferenceDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            else -> throw IllegalArgumentException("invalid datatype")
+        }
+    }
+
+    override fun memorySize(arrayDt: DataType, numElements: Int) =
+        if(arrayDt== DataType.UWORD)
+            numElements    // pointer to bytes.
+        else
+            memorySize(ArrayToElementTypes.getValue(arrayDt)) * numElements
+}

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

@@ -11,7 +11,9 @@ class AtariMachineDefinition: IMachineDefinition {
     override val FLOAT_MAX_POSITIVE = 9.999999999e97
     override val FLOAT_MAX_NEGATIVE = -9.999999999e97
     override val FLOAT_MEM_SIZE = 6
+    override val STARTUP_CODE_RESERVED_SIZE = 20u
     override val PROGRAM_LOAD_ADDRESS = 0x2000u
+    override val PROGRAM_MEMTOP_ADDRESS = 0xffffu  // TODO what's memtop?
 
     override val BSSHIGHRAM_START = 0u    // TODO
     override val BSSHIGHRAM_END = 0u      // TODO

codeCore/src/prog8/code/target/c128/C128MachineDefinition.kt

@@ -13,7 +13,9 @@ class C128MachineDefinition: IMachineDefinition {
     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 STARTUP_CODE_RESERVED_SIZE = 20u
     override val PROGRAM_LOAD_ADDRESS = 0x1c01u
+    override val PROGRAM_MEMTOP_ADDRESS = 0xff00u
 
     override val BSSHIGHRAM_START = 0u    // TODO
     override val BSSHIGHRAM_END = 0u      // TODO

codeCore/src/prog8/code/target/c64/C64MachineDefinition.kt

@@ -14,7 +14,10 @@ class C64MachineDefinition: IMachineDefinition {
     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 STARTUP_CODE_RESERVED_SIZE = 20u
     override val PROGRAM_LOAD_ADDRESS = 0x0801u
+    override val PROGRAM_MEMTOP_ADDRESS = 0xcfe0u      // $a000  if floats are used
+            // note that at $cfe0-$cfff are the 16 'virtual registers' R0-R15
 
     override val BSSHIGHRAM_START = 0xc000u
     override val BSSHIGHRAM_END = 0xcfffu

codeCore/src/prog8/code/target/cx16/CX16MachineDefinition.kt

@@ -13,7 +13,9 @@ class CX16MachineDefinition: IMachineDefinition {
     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 STARTUP_CODE_RESERVED_SIZE = 20u
     override val PROGRAM_LOAD_ADDRESS = 0x0801u
+    override val PROGRAM_MEMTOP_ADDRESS = 0x9f00u
 
     override val BSSHIGHRAM_START = 0xa000u     // hiram bank 1, 8Kb, assumed to be active
     override val BSSHIGHRAM_END = 0xbfffu       // Rom starts at $c000

codeCore/src/prog8/code/target/neo6502/Neo6502MachineDefinition.kt

@@ -0,0 +1,50 @@
+package prog8.code.target.neo6502
+
+import prog8.code.core.*
+import java.nio.file.Path
+
+
+class Neo6502MachineDefinition: IMachineDefinition {
+
+    override val cpu = CpuType.CPU65c02
+
+    override val FLOAT_MAX_POSITIVE = 9.999999999e97
+    override val FLOAT_MAX_NEGATIVE = -9.999999999e97
+    override val FLOAT_MEM_SIZE = 6
+    override val STARTUP_CODE_RESERVED_SIZE = 20u
+    override val PROGRAM_LOAD_ADDRESS = 0x0800u
+    override val PROGRAM_MEMTOP_ADDRESS = 0xfc00u       // kernal starts here
+
+    override val BSSHIGHRAM_START = 0u    // TODO
+    override val BSSHIGHRAM_END = 0u      // TODO
+    override val BSSGOLDENRAM_START = 0u  // TODO
+    override val BSSGOLDENRAM_END = 0u    // TODO
+
+    override lateinit var zeropage: Zeropage
+    override lateinit var golden: GoldenRam
+
+    override fun getFloatAsmBytes(num: Number) = TODO("atari float asm bytes from number")
+    override fun convertFloatToBytes(num: Double): List<UByte> = TODO("atari float to bytes")
+    override fun convertBytesToFloat(bytes: List<UByte>): Double = TODO("atari bytes to float")
+
+    override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
+        if(selectedEmulator!=1) {
+            System.err.println("The neo target only supports the main emulator (neo).")
+            return
+        }
+
+        val cmdline = listOf("neo", "${programNameWithPath}.bin@800", "cold")
+
+        println("\nStarting Neo6502 emulator...")
+        val processb = ProcessBuilder(cmdline).inheritIO()
+        val process: Process = processb.start()
+        process.waitFor()
+    }
+
+    override fun isIOAddress(address: UInt): Boolean = address in 0xff00u..0xff0fu
+
+    override fun initializeMemoryAreas(compilerOptions: CompilationOptions) {
+        zeropage = Neo6502Zeropage(compilerOptions)
+        golden = GoldenRam(compilerOptions, UIntRange.EMPTY)
+    }
+}

codeCore/src/prog8/code/target/neo6502/Neo6502Zeropage.kt

@@ -0,0 +1,48 @@
+package prog8.code.target.neo6502
+
+import prog8.code.core.*
+
+class Neo6502Zeropage(options: CompilationOptions) : Zeropage(options) {
+
+    override val SCRATCH_B1 = 0xfau      // temp storage for a single byte
+    override val SCRATCH_REG = 0xfbu     // temp storage for a register, must be B1+1
+    override val SCRATCH_W1 = 0xfcu      // temp storage 1 for a word  $fc+$fd
+    override val SCRATCH_W2 = 0xfeu      // temp storage 2 for a word  $fe+$ff
+
+    init {
+        if (options.floats) {
+            throw InternalCompilerException("Neo6502 target doesn't support floating point routines")
+        }
+
+        when (options.zeropage) {
+            ZeropageType.DONTUSE -> {
+                free.clear()  // don't use zeropage at all
+            }
+            else -> {
+                free.addAll(0x22u..0xffu)
+            }
+        }
+
+        val distinctFree = free.distinct()
+        free.clear()
+        free.addAll(distinctFree)
+
+        removeReservedFromFreePool()
+        allocateCx16VirtualRegisters()
+        retainAllowed()
+    }
+
+    override fun allocateCx16VirtualRegisters() {
+        // Note: the 16 virtual registers R0-R15 are not regular allocated variables, they're *memory mapped* elsewhere to fixed addresses.
+        // However, to be able for the compiler to "see" them as zero page variables, we have to register them here as well.
+        // This is important because the compiler sometimes treats ZP variables more efficiently (for example if it's a pointer)
+        for(reg in 0..15) {
+            allocatedVariables["cx16.r${reg}"]   = VarAllocation((2+reg*2).toUInt(), DataType.UWORD, 2)       // cx16.r0 .. cx16.r15
+            allocatedVariables["cx16.r${reg}s"]  = VarAllocation((2+reg*2).toUInt(), DataType.WORD, 2)        // cx16.r0s .. cx16.r15s
+            allocatedVariables["cx16.r${reg}L"]  = VarAllocation((2+reg*2).toUInt(), DataType.UBYTE, 1)       // cx16.r0L .. cx16.r15L
+            allocatedVariables["cx16.r${reg}H"]  = VarAllocation((3+reg*2).toUInt(), DataType.UBYTE, 1)       // cx16.r0H .. cx16.r15H
+            allocatedVariables["cx16.r${reg}sL"] = VarAllocation((2+reg*2).toUInt(), DataType.BYTE, 1)        // cx16.r0sL .. cx16.r15sL
+            allocatedVariables["cx16.r${reg}sH"] = VarAllocation((3+reg*2).toUInt(), DataType.BYTE, 1)        // cx16.r0sH .. cx16.r15sH
+        }
+    }
+}

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

@@ -13,7 +13,9 @@ class PETMachineDefinition: IMachineDefinition {
     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 STARTUP_CODE_RESERVED_SIZE = 20u
     override val PROGRAM_LOAD_ADDRESS = 0x0401u
+    override val PROGRAM_MEMTOP_ADDRESS = 0x8000u
 
     override val BSSHIGHRAM_START = 0u
     override val BSSHIGHRAM_END = 0u

codeCore/src/prog8/code/target/virtual/VirtualMachineDefinition.kt

@@ -13,7 +13,9 @@ class VirtualMachineDefinition: IMachineDefinition {
     override val FLOAT_MAX_POSITIVE = Float.MAX_VALUE.toDouble()
     override val FLOAT_MAX_NEGATIVE = -Float.MAX_VALUE.toDouble()
     override val FLOAT_MEM_SIZE = 8             // 64-bits double
+    override val STARTUP_CODE_RESERVED_SIZE = 0u  // not actually used
     override val PROGRAM_LOAD_ADDRESS = 0u      // not actually used
+    override val PROGRAM_MEMTOP_ADDRESS = 0xffffu  // not actually used
 
     override val BSSHIGHRAM_START = 0u          // not actually used
     override val BSSHIGHRAM_END = 0u            // not actually used

codeGenCpu6502/build.gradle

@@ -28,6 +28,7 @@ dependencies {
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:2.0.0"
 
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.1'
+    testImplementation 'io.kotest:kotest-framework-datatest:5.9.1'
     testImplementation 'org.junit.jupiter:junit-jupiter:5.9.2'
     testRuntimeOnly 'org.junit.platform:junit-platform-launcher'
 }

codeGenCpu6502/codeGenCpu6502.iml

@@ -14,5 +14,6 @@
     <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
     <orderEntry type="library" name="io.kotest.assertions.core.jvm" level="project" />
     <orderEntry type="library" name="io.kotest.runner.junit5.jvm" level="project" />
+    <orderEntry type="library" name="io.kotest.framework.datatest" level="project" />
   </component>
 </module>

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

@@ -9,7 +9,6 @@ import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.code.target.Cx16Target
 import prog8.codegen.cpu6502.assignment.*
-import java.util.*
 import kotlin.io.path.Path
 import kotlin.io.path.writeLines
 
@@ -17,15 +16,15 @@ import kotlin.io.path.writeLines
 internal const val subroutineFloatEvalResultVar1 = "prog8_float_eval_result1"
 internal const val subroutineFloatEvalResultVar2 = "prog8_float_eval_result2"
 
-class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
+class AsmGen6502(val prefixSymbols: Boolean, private val lastGeneratedLabelSequenceNr: Int): ICodeGeneratorBackend {
     override fun generate(
         program: PtProgram,
         symbolTable: SymbolTable,
         options: CompilationOptions,
-        errors: IErrorReporter
+        errors: IErrorReporter,
     ): IAssemblyProgram? {
         val st = if(prefixSymbols) prefixSymbols(program, options, symbolTable) else symbolTable
-        val asmgen = AsmGen6502Internal(program, st, options, errors)
+        val asmgen = AsmGen6502Internal(program, st, options, errors, lastGeneratedLabelSequenceNr)
         return asmgen.compileToAssembly()
     }
 
@@ -37,10 +36,9 @@ class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
             when(node) {
                 is PtAsmSub, is PtSub -> node.name = "p8s_${node.name}"
                 is PtBlock -> node.name = "p8b_${node.name}"
-                is PtLabel -> node.name = "p8l_${node.name}"
+                is PtLabel -> if(!node.name.startsWith(PtLabel.GeneratedLabelPrefix)) node.name = "p8l_${node.name}"   // don't prefix autogenerated labels
                 is PtConstant -> node.name = "p8c_${node.name}"
                 is PtVariable, is PtMemMapped, is PtSubroutineParameter -> node.name = "p8v_${node.name}"
-                else -> node.name = "p8_${node.name}"
             }
         }
 
@@ -49,6 +47,9 @@ class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
                 is PtAsmSub -> {
                     prefixNamedNode(node)
                     node.parameters.forEach { (_, param) -> prefixNamedNode(param) }
+                    if(node.address?.varbank!=null) {
+                        node.address!!.varbank = node.address!!.varbank!!.prefix(node, st)
+                    }
                 }
                 is PtSub -> {
                     prefixNamedNode(node)
@@ -104,7 +105,7 @@ class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
                 is PtIdentifier -> parent.children[index] = node.prefix(parent, st)
                 is PtFunctionCall ->  throw AssemblyError("PtFunctionCall should be processed in their own list, last")
                 is PtJump -> parent.children[index] = node.prefix(parent, st)
-                is PtVariable -> parent.children[index] = node.prefix(st)
+                is PtVariable -> parent.children[index] = node.prefix(parent, st)
                 else -> throw AssemblyError("weird node to prefix $node")
             }
         }
@@ -124,18 +125,22 @@ class AsmGen6502(val prefixSymbols: Boolean): ICodeGeneratorBackend {
 }
 
 private fun prefixScopedName(name: String, type: Char): String {
-    if('.' !in name)
+    if('.' !in name) {
+        if(name.startsWith(PtLabel.GeneratedLabelPrefix))
+            return name
         return "p8${type}_$name"
+    }
     val parts = name.split('.')
     val firstPrefixed = "p8b_${parts[0]}"
-    val lastPrefixed = "p8${type}_${parts.last()}"
+    val lastPart = parts.last()
+    val lastPrefixed = if(lastPart.startsWith(PtLabel.GeneratedLabelPrefix)) lastPart else  "p8${type}_$lastPart"
     // the parts in between are assumed to be subroutine scopes.
     val inbetweenPrefixed = parts.drop(1).dropLast(1).map{ "p8s_$it" }
     val prefixed = listOf(firstPrefixed) + inbetweenPrefixed + listOf(lastPrefixed)
     return prefixed.joinToString(".")
 }
 
-private fun PtVariable.prefix(st: SymbolTable): PtVariable {
+private fun PtVariable.prefix(parent: PtNode, st: SymbolTable): PtVariable {
     name = prefixScopedName(name, 'v')
     if(value==null)
         return this
@@ -145,7 +150,6 @@ private fun PtVariable.prefix(st: SymbolTable): PtVariable {
         val newValue = PtArray(arrayValue.type, arrayValue.position)
         arrayValue.children.forEach { elt ->
             when(elt) {
-                is PtIdentifier -> newValue.add(elt.prefix(arrayValue, st))
                 is PtBool -> newValue.add(elt)
                 is PtNumber -> newValue.add(elt)
                 is PtAddressOf -> {
@@ -163,7 +167,9 @@ private fun PtVariable.prefix(st: SymbolTable): PtVariable {
                 else -> throw AssemblyError("weird array value element $elt")
             }
         }
-        PtVariable(name, type, zeropage, newValue, arraySize, position)
+        val result = PtVariable(name, type, zeropage, align, newValue, arraySize, position)
+        result.parent = parent
+        result
     }
     else this
 }
@@ -199,7 +205,7 @@ private fun PtIdentifier.prefix(parent: PtNode, st: SymbolTable): PtIdentifier {
 
     val prefixType = when(target!!.type) {
         StNodeType.BLOCK -> 'b'
-        StNodeType.SUBROUTINE, StNodeType.ROMSUB -> 's'
+        StNodeType.SUBROUTINE, StNodeType.EXTSUB -> 's'
         StNodeType.LABEL -> 'l'
         StNodeType.STATICVAR, StNodeType.MEMVAR -> 'v'
         StNodeType.CONSTANT -> 'c'
@@ -218,7 +224,8 @@ class AsmGen6502Internal (
     val program: PtProgram,
     internal val symbolTable: SymbolTable,
     internal val options: CompilationOptions,
-    internal val errors: IErrorReporter
+    internal val errors: IErrorReporter,
+    private var generatedLabelSequenceNumber: Int
 ) {
 
     internal val optimizedByteMultiplications = setOf(3,5,6,7,9,10,11,12,13,14,15,20,25,40,50,80,100)
@@ -232,7 +239,7 @@ class AsmGen6502Internal (
     private val functioncallAsmGen = FunctionCallAsmGen(program, this)
     private val programGen = ProgramAndVarsGen(program, options, errors, symbolTable, functioncallAsmGen, this, allocator, zeropage)
     private val anyExprGen = AnyExprAsmGen(this)
-    private val assignmentAsmGen = AssignmentAsmGen(program, this, options, anyExprGen, allocator)
+    private val assignmentAsmGen = AssignmentAsmGen(program, this, anyExprGen, allocator)
     private val builtinFunctionsAsmGen = BuiltinFunctionsAsmGen(program, this, assignmentAsmGen)
     private val ifElseAsmgen = IfElseAsmGen(program, symbolTable, this, allocator, assignmentAsmGen, errors)
 
@@ -374,7 +381,7 @@ class AsmGen6502Internal (
         val name = asmSymbolName(identifier.name)
 
         // see if we're inside a subroutine, if so, remove the whole prefix and just make the variable name locally scoped (64tass scopes it to the proper .proc block)
-        val subName = identifier.definingSub()?.scopedName
+        val subName = identifier.definingISub()?.scopedName
         return if (subName != null && name.length>subName.length && name.startsWith(subName) && name[subName.length] == '.')
             name.drop(subName.length + 1)
         else
@@ -385,14 +392,14 @@ class AsmGen6502Internal (
         val name = asmVariableName(identifier.name)
 
         // see if we're inside a subroutine, if so, remove the whole prefix and just make the variable name locally scoped (64tass scopes it to the proper .proc block)
-        val subName = identifier.definingSub()?.scopedName
+        val subName = identifier.definingISub()?.scopedName
         return if (subName != null && name.length>subName.length && name.startsWith(subName) && name[subName.length] == '.')
             name.drop(subName.length+1)
         else
             name
     }
 
-    fun asmVariableName(st: StNode, scope: PtSub?): String {
+    fun asmVariableName(st: StNode, scope: IPtSubroutine?): String {
         val name = asmVariableName(st.scopedName)
         if(scope==null)
             return name
@@ -580,6 +587,7 @@ class AsmGen6502Internal (
     internal fun translate(stmt: PtNode) {
         outputSourceLine(stmt)
         when(stmt) {
+            is PtAlign -> if(stmt.align > 1u) out("  .align  ${stmt.align.toHex()}")
             is PtReturn -> translate(stmt)
             is PtSub -> programGen.translateSubroutine(stmt)
             is PtAsmSub -> programGen.translateAsmSubroutine(stmt)
@@ -602,9 +610,10 @@ class AsmGen6502Internal (
             is PtRepeatLoop -> translate(stmt)
             is PtWhen -> translate(stmt)
             is PtIncludeBinary -> translate(stmt)
-            is PtBreakpoint -> translate(stmt)
+            is PtBreakpoint -> translateBrk()
             is PtVariable, is PtConstant, is PtMemMapped -> { /* do nothing; variables are handled elsewhere */ }
             is PtBlock -> throw AssemblyError("block should have been handled elsewhere")
+            is PtDefer -> throw AssemblyError("defer should have been transformed")
             is PtNodeGroup -> stmt.children.forEach { translate(it) }
             is PtNop -> {}
             else -> throw AssemblyError("missing asm translation for $stmt")
@@ -772,7 +781,7 @@ class AsmGen6502Internal (
 
     private fun translate(stmt: PtRepeatLoop) {
         val endLabel = makeLabel("repeatend")
-        loopEndLabels.push(endLabel)
+        loopEndLabels.add(endLabel)
 
         when (stmt.count) {
             is PtNumber -> {
@@ -780,7 +789,7 @@ class AsmGen6502Internal (
                 when {
                     iterations == 0 -> {}
                     iterations == 1 -> translate(stmt.statements)
-                    iterations<0 || iterations>65535 -> throw AssemblyError("invalid number of iterations")
+                    iterations<0 || iterations>65536 -> throw AssemblyError("invalid number of iterations")
                     iterations <= 256 -> repeatByteCount(iterations, stmt)
                     else -> repeatWordCount(iterations, stmt)
                 }
@@ -816,14 +825,14 @@ class AsmGen6502Internal (
             }
         }
 
-        loopEndLabels.pop()
+        loopEndLabels.removeLast()
     }
 
     private fun repeatWordCount(iterations: Int, stmt: PtRepeatLoop) {
-        require(iterations in 257..65535) { "invalid repeat count ${stmt.position}" }
+        require(iterations in 257..65536) { "invalid repeat count ${stmt.position}" }
         val repeatLabel = makeLabel("repeat")
         val counterVar = createRepeatCounterVar(DataType.UWORD, isTargetCpu(CpuType.CPU65c02), stmt)
-        val loopcount = if(iterations and 0x00ff == 0) iterations else iterations + 0x0100   // so that the loop can simply use a double-dec
+        val loopcount = if(iterations==65536) 0 else if(iterations and 0x00ff == 0) iterations else iterations + 0x0100   // so that the loop can simply use a double-dec
         out("""
             ldy  #>$loopcount
             lda  #<$loopcount
@@ -1038,7 +1047,7 @@ $repeatLabel""")
         }
     }
 
-    private fun translate(ret: PtReturn, withRts: Boolean=true) {
+    private fun translate(ret: PtReturn) {
         ret.value?.let { returnvalue ->
             val sub = ret.definingSub()!!
             val returnReg = sub.returnRegister()!!
@@ -1055,9 +1064,7 @@ $repeatLabel""")
                 }
             }
         }
-
-        if(withRts)
-            out("  rts")
+        out("  rts")
     }
 
     private fun translate(asm: PtInlineAssembly) {
@@ -1082,7 +1089,7 @@ $repeatLabel""")
         out("  .binary \"$pathForAssembler\" $offset $length")
     }
 
-    private fun translate(brk: PtBreakpoint) {
+    private fun translateBrk() {
         val label = "_prog8_breakpoint_${breakpointLabels.size+1}"
         breakpointLabels.add(label)
         out(label)
@@ -1361,11 +1368,9 @@ $repeatLabel""")
             extra
     }
 
-    private var generatedLabelSequenceNumber: Int = 0
-
     internal fun makeLabel(postfix: String): String {
         generatedLabelSequenceNumber++
-        return "label_asm_${generatedLabelSequenceNumber}_$postfix"
+        return "${PtLabel.GeneratedLabelPrefix}${generatedLabelSequenceNumber}_$postfix"
     }
 
     fun assignConstFloatToPointerAY(number: PtNumber) {
@@ -1379,6 +1384,10 @@ $repeatLabel""")
             pla
             jsr  floats.copy_float""")
     }
+
+    internal fun assignIfExpression(target: AsmAssignTarget, value: PtIfExpression) {
+        ifElseAsmgen.assignIfExpression(target, value)
+    }
 }
 
 /**

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

@@ -3,6 +3,7 @@ package prog8.codegen.cpu6502
 import prog8.code.StConstant
 import prog8.code.StMemVar
 import prog8.code.SymbolTable
+import prog8.code.ast.PtLabel
 import prog8.code.core.IMachineDefinition
 
 
@@ -361,15 +362,16 @@ or *_afterif labels.
 This gets generated after certain if conditions, and only the branch instruction is needed in these cases.
          */
 
+        val autoLabelPrefix = PtLabel.GeneratedLabelPrefix
         if(first=="beq  +" && second=="lda  #1" && third=="+") {
-            if((fourth.startsWith("beq  label_") || fourth.startsWith("bne  label_")) &&
+            if((fourth.startsWith("beq  $autoLabelPrefix") || fourth.startsWith("bne  $autoLabelPrefix")) &&
                 (fourth.endsWith("_shortcut") || fourth.endsWith("_afterif") || fourth.endsWith("_shortcut:") || fourth.endsWith("_afterif:"))) {
                 mods.add(Modification(lines[0].index, true, null))
                 mods.add(Modification(lines[1].index, true, null))
                 mods.add(Modification(lines[2].index, true, null))
             }
-            else if(fourth.startsWith("label_") && (fourth.endsWith("_shortcut") || fourth.endsWith("_shortcut:"))) {
-                if((fifth.startsWith("beq  label_") || fifth.startsWith("bne  label_")) &&
+            else if(fourth.startsWith(autoLabelPrefix) && (fourth.endsWith("_shortcut") || fourth.endsWith("_shortcut:"))) {
+                if((fifth.startsWith("beq  $autoLabelPrefix") || fifth.startsWith("bne  $autoLabelPrefix")) &&
                     (fifth.endsWith("_shortcut") || fifth.endsWith("_afterif") || fifth.endsWith("_shortcut:") || fifth.endsWith("_afterif:"))) {
                     mods.add(Modification(lines[0].index, true, null))
                     mods.add(Modification(lines[1].index, true, null))
@@ -516,34 +518,37 @@ private fun optimizeJsrRtsAndOtherCombinations(linesByFour: Sequence<List<Indexe
         val first = lines[0].value
         val second = lines[1].value
         val third = lines[2].value
-        if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {
-            mods += Modification(lines[0].index, false, lines[0].value.replace("jsr", "jmp"))
-            mods += Modification(lines[1].index, true, null)
-        }
-        else if (" rts" in first || "\trts" in first) {
-            if (" jmp" in second || "\tjmp" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bra" in second || "\tbra" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bcc" in second || "\tbcc" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bcs" in second || "\tbcs" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" beq" in second || "\tbeq" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bne" in second || "\tbne" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bmi" in second || "\tbmi" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bpl" in second || "\tbpl" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bvs" in second || "\tbvs" in second)
-                mods += Modification(lines[1].index, true, null)
-            else if (" bvc" in second || "\tbvc" in second)
-                mods += Modification(lines[1].index, true, null)
-        }
 
-        if (!haslabel(second)) {
+        if(!haslabel(second)) {
+            if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {
+                if("floats.pushFAC" !in first && "floats.popFAC" !in first) {       // these 2 routines depend on being called with JSR!!
+                    mods += Modification(lines[0].index, false, lines[0].value.replace("jsr", "jmp"))
+                    mods += Modification(lines[1].index, true, null)
+                }
+            }
+            else if (" rts" in first || "\trts" in first) {
+                if (" jmp" in second || "\tjmp" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bra" in second || "\tbra" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bcc" in second || "\tbcc" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bcs" in second || "\tbcs" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" beq" in second || "\tbeq" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bne" in second || "\tbne" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bmi" in second || "\tbmi" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bpl" in second || "\tbpl" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bvs" in second || "\tbvs" in second)
+                    mods += Modification(lines[1].index, true, null)
+                else if (" bvc" in second || "\tbvc" in second)
+                    mods += Modification(lines[1].index, true, null)
+            }
+
             if ((" lda" in first || "\tlda" in first) && (" cmp  #0" in second || "\tcmp  #0" in second) ||
                 (" ldx" in first || "\tldx" in first) && (" cpx  #0" in second || "\tcpx  #0" in second) ||
                 (" ldy" in first || "\tldy" in first) && (" cpy  #0" in second || "\tcpy  #0" in second)

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

@@ -11,18 +11,22 @@ fun asmsub6502ArgsEvalOrder(sub: PtAsmSub): List<Int> {
     //  1) cx16 virtual word registers,
     //  2) paired CPU registers,
     //  3) single CPU registers (order Y,X,A),
-    //  4) CPU Carry status flag
+    //  4) floating point registers (FAC1, FAC2),
+    //  5) CPU Carry status flag
     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 (singleRegs, rest) = args3.partition { it.value.first.registerOrPair != null }
+    val (singleRegsMixed, rest) = args3.partition { it.value.first.registerOrPair != null }
+    val (singleCpuRegs, floatRegs) = singleRegsMixed.partition {it.value.first.registerOrPair != RegisterOrPair.FAC1 && it.value.first.registerOrPair != RegisterOrPair.FAC2  }
 
     cx16regs.forEach { order += it.index }
     pairedRegs.forEach { order += it.index }
-    singleRegs.sortedBy { it.value.first.registerOrPair!!.asCpuRegister() }.asReversed().forEach { order += it.index }
+    singleCpuRegs.sortedBy { it.value.first.registerOrPair!!.asCpuRegister() }.asReversed().forEach { order += it.index }
     require(rest.all { it.value.first.registerOrPair==null && it.value.first.statusflag!=null})
+    floatRegs.forEach { order += it.index }
     rest.forEach { order += it.index }
     require(order.size==sub.parameters.size)
+
     return order
 }

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

@@ -1,13 +1,16 @@
 package prog8.codegen.cpu6502
 
+import prog8.code.ast.PtLabel
 import prog8.code.core.*
+import prog8.code.target.AtariTarget
 import prog8.code.target.C64Target
+import prog8.code.target.Neo6502Target
 import java.nio.file.Path
 
 
 internal class AssemblyProgram(
         override val name: String,
-        outputDir: Path,
+        private val outputDir: Path,
         private val compTarget: ICompilationTarget) : IAssemblyProgram {
 
     private val assemblyFile = outputDir.resolve("$name.asm")
@@ -16,6 +19,7 @@ internal class AssemblyProgram(
     private val binFile = outputDir.resolve("$name.bin")
     private val viceMonListFile = outputDir.resolve(C64Target.viceMonListName(name))
     private val listFile = outputDir.resolve("$name.list")
+    private val targetWithoutBreakpointsForEmulator = setOf(AtariTarget.NAME, Neo6502Target.NAME)
 
     override fun assemble(options: CompilationOptions, errors: IErrorReporter): Boolean {
 
@@ -96,12 +100,47 @@ internal class AssemblyProgram(
                 command.addAll(listOf("--output", outFile.toString(), assemblyFile.toString()))
                 assemblerCommand = command
             }
+            "neo" -> {
+                // Neo6502 raw program generation.
+
+                if(options.output!=OutputType.RAW || options.loadAddress!=0x0800u || options.launcher!=CbmPrgLauncherType.NONE) {
+                    throw AssemblyError("invalid program compilation options. Neo6502 requires %output raw, %launcher none, %address $0800")
+                }
+
+                // TODO are these options okay for neo?
+                val command = mutableListOf("64tass", "--case-sensitive", "--long-branch",
+                    "-Wall", // "-Werror", "-Wno-strict-bool"
+                    "--no-monitor"
+                )
+
+                if(options.warnSymbolShadowing)
+                    command.add("-Wshadow")
+                else
+                    command.add("-Wno-shadow")
+
+                if(options.asmQuiet)
+                    command.add("--quiet")
+
+                if(options.asmListfile)
+                    command.add("--list=$listFile")
+
+                val outFile = when (options.output) {
+                    OutputType.RAW -> {
+                        command.add("--nostart")
+                        println("\nCreating raw binary for target ${compTarget.name}.")
+                        binFile
+                    }
+                    else -> throw AssemblyError("invalid output type, need 'raw'")
+                }
+                command.addAll(listOf("--output", outFile.toString(), assemblyFile.toString()))
+                assemblerCommand = command
+            }
             else -> throw AssemblyError("invalid compilation target")
         }
 
         val proc = ProcessBuilder(assemblerCommand).inheritIO().start()
         val result = proc.waitFor()
-        if (result == 0 && compTarget.name!="atari") {
+        if (result == 0 && compTarget.name !in targetWithoutBreakpointsForEmulator) {
             removeGeneratedLabelsFromMonlist()
             generateBreakpointList()
         }
@@ -109,7 +148,7 @@ internal class AssemblyProgram(
     }
 
     private fun removeGeneratedLabelsFromMonlist() {
-        val pattern = Regex("""al (\w+) \S+prog8_label_.+?""")
+        val pattern = Regex("""al (\w+) \S+${PtLabel.GeneratedLabelPrefix}.+?""")
         val lines = viceMonListFile.toFile().readLines()
         viceMonListFile.toFile().outputStream().bufferedWriter().use {
             for (line in lines) {

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

@@ -1,7 +1,5 @@
 package prog8.codegen.cpu6502
 
-import prog8.code.StMemVar
-import prog8.code.StStaticVariable
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.codegen.cpu6502.assignment.*
@@ -64,120 +62,32 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "rrestore" -> funcRrestore()
             "cmp" -> funcCmp(fcall)
             "callfar" -> funcCallFar(fcall, resultRegister)
+            "callfar2" -> funcCallFar2(fcall, resultRegister)
             "call" -> funcCall(fcall)
             "prog8_ifelse_bittest_set" -> throw AssemblyError("prog8_ifelse_bittest_set() should have been translated as part of an ifElse statement")
             "prog8_ifelse_bittest_notset" -> throw AssemblyError("prog8_ifelse_bittest_notset() should have been translated as part of an ifElse statement")
             "prog8_lib_stringcompare" -> funcStringCompare(fcall, resultRegister)
             "prog8_lib_square_byte" -> funcSquare(fcall, DataType.UBYTE, resultRegister)
             "prog8_lib_square_word" -> funcSquare(fcall, DataType.UWORD, resultRegister)
-            "prog8_lib_arraycopy" -> funcArrayCopy(fcall)
             else -> throw AssemblyError("missing asmgen for builtin func ${fcall.name}")
         }
 
         return BuiltinFunctions.getValue(fcall.name).returnType
     }
 
-    private fun funcArrayCopy(fcall: PtBuiltinFunctionCall) {
-        val source = fcall.args[0] as PtIdentifier
-        val target = fcall.args[1] as PtIdentifier
-
-        val numElements = when(val sourceSymbol = asmgen.symbolTable.lookup(source.name)) {
-            is StStaticVariable -> sourceSymbol.length!!
-            is StMemVar -> sourceSymbol.length!!
-            else -> 0
-        }
-        val sourceAsm = asmgen.asmVariableName(source)
-        val targetAsm = asmgen.asmVariableName(target)
-
-        if(source.type in SplitWordArrayTypes && target.type in SplitWordArrayTypes) {
-            // split -> split words (copy lsb and msb arrays separately)
-            asmgen.out("""
-                lda  #<${sourceAsm}_lsb
-                ldy  #>${sourceAsm}_lsb
-                sta  P8ZP_SCRATCH_W1
-                sty  P8ZP_SCRATCH_W1+1
-                lda  #<${targetAsm}_lsb
-                ldy  #>${targetAsm}_lsb
-                sta  P8ZP_SCRATCH_W2
-                sty  P8ZP_SCRATCH_W2+1
-                ldy  #${numElements and 255}
-                jsr  prog8_lib.memcopy_small
-                lda  #<${sourceAsm}_msb
-                ldy  #>${sourceAsm}_msb
-                sta  P8ZP_SCRATCH_W1
-                sty  P8ZP_SCRATCH_W1+1
-                lda  #<${targetAsm}_msb
-                ldy  #>${targetAsm}_msb
-                sta  P8ZP_SCRATCH_W2
-                sty  P8ZP_SCRATCH_W2+1
-                ldy  #${numElements and 255}
-                jsr  prog8_lib.memcopy_small""")
-        }
-        else if(source.type in SplitWordArrayTypes) {
-            // split word array to normal word array (copy lsb and msb arrays separately)
-            require(target.type==DataType.ARRAY_UW || target.type==DataType.ARRAY_W)
-            asmgen.out("""
-                lda  #<${sourceAsm}_lsb
-                ldy  #>${sourceAsm}_lsb
-                sta  P8ZP_SCRATCH_W1
-                sty  P8ZP_SCRATCH_W1+1
-                lda  #<${sourceAsm}_msb
-                ldy  #>${sourceAsm}_msb
-                sta  P8ZP_SCRATCH_W2
-                sty  P8ZP_SCRATCH_W2+1
-                lda  #<${targetAsm}
-                ldy  #>${targetAsm}
-                ldx  #${numElements and 255}
-                jsr  prog8_lib.arraycopy_split_to_normal_words""")
-        }
-        else if(target.type in SplitWordArrayTypes) {
-            // normal word array to split array
-            require(source.type==DataType.ARRAY_UW || source.type==DataType.ARRAY_W)
-            asmgen.out("""
-                lda  #<${targetAsm}_lsb
-                ldy  #>${targetAsm}_lsb
-                sta  P8ZP_SCRATCH_W1
-                sty  P8ZP_SCRATCH_W1+1
-                lda  #<${targetAsm}_msb
-                ldy  #>${targetAsm}_msb
-                sta  P8ZP_SCRATCH_W2
-                sty  P8ZP_SCRATCH_W2+1
-                lda  #<${sourceAsm}
-                ldy  #>${sourceAsm}
-                ldx  #${numElements and 255}
-                jsr  prog8_lib.arraycopy_normal_to_split_words""")
-        }
-        else {
-            // normal array to array copy, various element types
-            val eltsize = asmgen.options.compTarget.memorySize(ArrayToElementTypes.getValue(source.type))
-            val numBytes = numElements * eltsize
-            asmgen.out("""
-                lda  #<${sourceAsm}
-                ldy  #>${sourceAsm}
-                sta  P8ZP_SCRATCH_W1
-                sty  P8ZP_SCRATCH_W1+1
-                lda  #<${targetAsm}
-                ldy  #>${targetAsm}
-                sta  P8ZP_SCRATCH_W2
-                sty  P8ZP_SCRATCH_W2+1
-                ldy  #${numBytes and 255}
-                jsr  prog8_lib.memcopy_small""")
-        }
-    }
-
     private fun funcSquare(fcall: PtBuiltinFunctionCall, resultType: DataType, resultRegister: RegisterOrPair?) {
         // 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")
+                asmgen.out("  tay  |  jsr  prog8_math.multiply_bytes")
                 if(resultRegister!=null)  {
                     assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), CpuRegister.A, false, false)
                 }
             }
             DataType.UWORD -> {
                 asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
-                asmgen.out("  jsr  math.square")
+                asmgen.out("  jsr  prog8_math.square")
                 if(resultRegister!=null)  {
                     assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), RegisterOrPair.AY)
                 }
@@ -194,7 +104,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         assignAsmGen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.Y, false)
         asmgen.restoreRegisterStack(CpuRegister.A ,false)
         // math.divmod_ub_asm: -- divide A by Y, result quotient in Y, remainder in A   (unsigned)
-        asmgen.out("  jsr  math.divmod_ub_asm")
+        asmgen.out("  jsr  prog8_math.divmod_ub_asm")
         val var2name = asmgen.asmVariableName(fcall.args[2] as PtIdentifier)
         val var3name = asmgen.asmVariableName(fcall.args[3] as PtIdentifier)
         val divisionTarget = AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UBYTE, fcall.definingISub(), fcall.args[2].position, var2name)
@@ -208,7 +118,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         // math.divmod_uw_asm: -- divide two unsigned words (16 bit each) into 16 bit results
         //    input:  P8ZP_SCRATCH_W1 in ZP: 16 bit number, A/Y: 16 bit divisor
         //    output: P8ZP_SCRATCH_W2 in ZP: 16 bit remainder, A/Y: 16 bit division result
-        asmgen.out("  jsr  math.divmod_uw_asm")
+        asmgen.out("  jsr  prog8_math.divmod_uw_asm")
         val var2name = asmgen.asmVariableName(fcall.args[2] as PtIdentifier)
         val divisionTarget = AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UBYTE, fcall.definingISub(), fcall.args[2].position, var2name)
         val remainderVar = asmgen.asmVariableName(fcall.args[3] as PtIdentifier)
@@ -266,33 +176,124 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
     }
 
     private fun funcCall(fcall: PtBuiltinFunctionCall) {
+        // note: the routine can return a word value (in AY)
         val constAddr = fcall.args[0].asConstInteger()
         if(constAddr!=null) {
             asmgen.out("  jsr  ${constAddr.toHex()}")
-        } else {
-            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)     // jump address
-            asmgen.out("""
-              sta  (+)+1
-              sty  (+)+2
-+             jsr  0       ; modified""")
+            return
         }
 
-        // note: the routine can return a word value (in AY)
+        val identifier = fcall.args[0] as? PtIdentifier
+        if(identifier!=null) {
+            asmgen.out("""
+                ; push a return address so the jmp becomes indirect jsr
+                lda  #>((+)-1)
+                pha
+                lda  #<((+)-1)
+                pha
+                jmp  (${asmgen.asmSymbolName(identifier)})
++""")
+            return
+        }
+
+        asmgen.assignExpressionToVariable(fcall.args[0], asmgen.asmVariableName("P8ZP_SCRATCH_W2"), DataType.UWORD)     // jump address
+        asmgen.out("""
+                ; push a return address so the jmp becomes indirect jsr
+                lda  #>((+)-1)
+                pha
+                lda  #<((+)-1)
+                pha
+                jmp  (P8ZP_SCRATCH_W2)
++""")
     }
 
     private fun funcCallFar(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
-        if(asmgen.options.compTarget.name != "cx16")
-            throw AssemblyError("callfar only works on cx16 target at this time")
+        val targetName = asmgen.options.compTarget.name
+        if(targetName !in arrayOf("cx16", "c64", "c128"))
+            throw AssemblyError("callfar only works on cx16, c64 and c128 targets at this time")
+
+        val jsrfar = when(targetName) {
+            "cx16" -> "cx16.JSRFAR"
+            "c64" -> "c64.x16jsrfar"
+            "c128" -> "c128.x16jsrfar"
+            else -> TODO("jsrfar routine")
+        }
+        val constBank = fcall.args[0].asConstInteger()
+        val constAddress = fcall.args[1].asConstInteger()
+        if(constBank!=null && constAddress!=null) {
+            asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
+            asmgen.out("""
+                jsr  $jsrfar
+                .word  ${constAddress.toHex()}
+                .byte  $constBank""")
+        } else {
+            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
+            asmgen.out("  sta  (++)+0")
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
+            asmgen.out("  sta  (+)+0 |  sty  (+)+1")
+            asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
+            asmgen.out("""
+                jsr  $jsrfar
++               .word  0
++               .byte  0""")
+        }
+
+        // note that by convention the values in A+Y registers are now the return value of the call.
+        if(resultRegister!=null)  {
+            assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), RegisterOrPair.AY)
+        }
+    }
+
+    private fun funcCallFar2(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        val targetName = asmgen.options.compTarget.name
+        if(targetName !in arrayOf("cx16", "c64", "c128"))
+            throw AssemblyError("callfar2 only works on cx16, c64 and c128 targets at this time")
+
+        fun assignArgs() {
+            fun assign(value: PtExpression, register: Char) {
+                when(value) {
+                    is PtBool -> asmgen.out("  ld$register  #${value.asInt()}")
+                    is PtNumber -> asmgen.out("  ld$register  #${value.number.toInt()}")
+                    is PtIdentifier -> asmgen.out("  ld$register  ${asmgen.asmVariableName(value)}")
+                    else -> TODO("callfar2: support non-simple expressions for arguments")
+                }
+            }
+            assign(fcall.args[2], 'a')
+            assign(fcall.args[3], 'x')
+            assign(fcall.args[4], 'y')
+            val carry = fcall.args[5].asConstInteger()
+            if(carry!=null)
+                asmgen.out(if(carry==0) "  clc" else "  sec")
+            else
+                TODO("callfar2: support non-const argument values")
+        }
+
+        val jsrfar = when(targetName) {
+            "cx16" -> "cx16.JSRFAR"
+            "c64" -> "c64.x16jsrfar"
+            "c128" -> "c128.x16jsrfar"
+            else -> TODO("jsrfar routine")
+        }
+        val constBank = fcall.args[0].asConstInteger()
+        val constAddress = fcall.args[1].asConstInteger()
+        if(constBank!=null && constAddress!=null) {
+            assignArgs()
+            asmgen.out("""
+                jsr  $jsrfar
+                .word  ${constAddress.toHex()}
+                .byte  $constBank""")
+        } else {
+            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
+            asmgen.out("  sta  (++)+0")
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
+            asmgen.out("  sta  (+)+0 |  sty  (+)+1")
+            assignArgs()
+            asmgen.out("""
+                jsr  $jsrfar
++               .word  0
++               .byte  0""")
+        }
 
-        asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
-        asmgen.out("  sta  (++)+0")
-        asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
-        asmgen.out("  sta  (+)+0 |  sty  (+)+1")
-        asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
-        asmgen.out("""
-            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.
         if(resultRegister!=null)  {
             assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), RegisterOrPair.AY)
@@ -1122,7 +1123,44 @@ 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 msb byte out of the word array
+                if(arg.splitWords) {
+                    val arrayVar = asmgen.asmVariableName(arg.variable)+"_msb"
+                    when(resultRegister) {
+                        null, RegisterOrPair.A -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
+                            asmgen.out("  lda  $arrayVar,y")
+                        }
+                        RegisterOrPair.Y -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.X)
+                            asmgen.out("  ldy  $arrayVar,x")
+                        }
+                        RegisterOrPair.X -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
+                            asmgen.out("  ldx  $arrayVar,y")
+                        }
+                        else -> throw AssemblyError("invalid reg")
+                    }
+                } else {
+                    val arrayVar = asmgen.asmVariableName(arg.variable)
+                    when(resultRegister) {
+                        null, RegisterOrPair.A -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
+                            asmgen.out("  lda  $arrayVar+1,y")
+                        }
+                        RegisterOrPair.Y -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.X)
+                            asmgen.out("  ldy  $arrayVar+1,x")
+                        }
+                        RegisterOrPair.X -> {
+                            asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
+                            asmgen.out("  ldx  $arrayVar+1,y")
+                        }
+                        else -> throw AssemblyError("invalid reg")
+                    }
+                }
+            } else when(resultRegister) {
                 null, RegisterOrPair.A -> {
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  tya")
@@ -1137,6 +1175,23 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  pla")
                 }
+                RegisterOrPair.AY -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
+                    asmgen.out("  tya |  ldy  #0 |  cmp  #0")
+                }
+                RegisterOrPair.AX -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AX)
+                    asmgen.out("  txa |  ldx  #0 |  cmp  #0")
+                }
+                RegisterOrPair.XY -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.XY)
+                    asmgen.out("  tya |  tax |  ldy  #0 |  cpx  #0")
+                }
+                in Cx16VirtualRegisters -> {
+                    val reg = "cx16.${resultRegister.toString().lowercase()}"
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
+                    asmgen.out("  sty  ${reg}L |  lda  #0 |  sta  ${reg}H |  lda  ${reg}L")
+                }
                 else -> throw AssemblyError("invalid reg")
             }
         }
@@ -1178,7 +1233,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     }
                     RegisterOrPair.Y -> {
                         asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.X)
-                        asmgen.out("  lda  $arrayVar,x")
+                        asmgen.out("  ldy  $arrayVar,x")
                     }
                     RegisterOrPair.X -> {
                         asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
@@ -1216,33 +1271,15 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.out("  ldy  #0 |  cpx  #0")
                 }
                 in Cx16VirtualRegisters -> {
-                    asmgen.assignExpressionToRegister(arg, resultRegister)
-                    val zero = PtNumber(DataType.UBYTE, 0.0, Position.DUMMY)
-                    zero.parent=fcall
-                    assignAsmGen.assignExpressionToVariable(zero, "cx16.${resultRegister.toString().lowercase()}H", DataType.UBYTE)
-                    asmgen.out("  lda  cx16.r0L")
+                    val reg = "cx16.${resultRegister.toString().lowercase()}"
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
+                    asmgen.out("  sta  ${reg}L |  ldy  #0 |  sty  ${reg}H |  cmp  #0")
                 }
                 else -> throw AssemblyError("invalid reg")
             }
         }
     }
 
-    private fun outputAddressAndLengthOfArray(arg: PtIdentifier) {
-        // address goes in P8ZP_SCRATCH_W1,  number of elements in A
-        val numElements = when(val symbol = asmgen.symbolTable.lookup(arg.name)) {
-            is StStaticVariable -> symbol.length!!
-            is StMemVar -> symbol.length!!
-            else -> 0
-        }
-        val identifierName = asmgen.asmVariableName(arg)
-        asmgen.out("""
-                    lda  #<$identifierName
-                    ldy  #>$identifierName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #${numElements and 255}""")
-    }
-
     private fun translateArguments(call: PtBuiltinFunctionCall, scope: IPtSubroutine?) {
         val signature = BuiltinFunctions.getValue(call.name)
         val callConv = signature.callConvention(call.args.map { it.type})

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

@@ -37,7 +37,7 @@ internal class ForLoopsAsmGen(
         val endLabel = asmgen.makeLabel("for_end")
         val modifiedLabel = asmgen.makeLabel("for_modified")
         val modifiedLabel2 = asmgen.makeLabel("for_modifiedb")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         val stepsize=range.step.asConstInteger()!!
 
         if(stepsize < -1) {
@@ -48,25 +48,43 @@ internal class ForLoopsAsmGen(
 
         when(iterableDt) {
             DataType.ARRAY_B, DataType.ARRAY_UB -> {
-                if (stepsize==1 || stepsize==-1) {
-
+                val varname = asmgen.asmVariableName(stmt.variable)
+                asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt))
+                if (stepsize==-1 && range.to.asConstInteger()==0) {
+                    // simple loop downto 0 step -1
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.statements)
+                    asmgen.out("""
+                        dec  $varname
+                        lda  $varname
+                        cmp  #255
+                        bne  $loopLabel""")
+                }
+                else if (stepsize==-1 && range.to.asConstInteger()==1) {
+                    // simple loop downto 1 step -1
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.statements)
+                    asmgen.out("""
+                        dec  $varname
+                        bne  $loopLabel""")
+                }
+                else if (stepsize==1 || stepsize==-1) {
                     // bytes array, step 1 or -1
 
                     val incdec = if(stepsize==1) "inc" else "dec"
                     // loop over byte range via loopvar
-                    val varname = asmgen.asmVariableName(stmt.variable)
-                    asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt))
                     asmgen.assignExpressionToRegister(range.to, RegisterOrPair.A, false)
                     // pre-check for end already reached
                     if(iterableDt==DataType.ARRAY_B) {
                         asmgen.out("  sta  $modifiedLabel+1")
-                        if(stepsize<0)
+                        if(stepsize<0) {
                             asmgen.out("""
                                 clc
                                 sbc  $varname
                                 bvc  +
                                 eor  #${'$'}80
 +                               bpl  $endLabel""")
+                        }
                         else
                             asmgen.out("""
                                 sec
@@ -75,12 +93,13 @@ internal class ForLoopsAsmGen(
                                 eor  #${'$'}80
 +                               bmi  $endLabel""")
                     } else {
-                        if(stepsize<0)
+                        if(stepsize<0) {
                             asmgen.out("""
                                 cmp  $varname
                                 beq  +
                                 bcs  $endLabel
 +""")
+                        }
                         else
                             asmgen.out("  cmp  $varname  |  bcc  $endLabel")
                         asmgen.out("  sta  $modifiedLabel+1")
@@ -100,8 +119,6 @@ $modifiedLabel          cmp  #0         ; modified
                     // bytes, step >= 2 or <= -2
 
                     // loop over byte range via loopvar
-                    val varname = asmgen.asmVariableName(stmt.variable)
-                    asmgen.assignExpressionToVariable(range.from, varname, ArrayToElementTypes.getValue(iterableDt))
                     asmgen.assignExpressionToRegister(range.to, RegisterOrPair.A, false)
                     // pre-check for end already reached
                     if(iterableDt==DataType.ARRAY_B) {
@@ -155,58 +172,85 @@ $modifiedLabel              cmp  #0     ; modified
                 }
             }
             DataType.ARRAY_W, DataType.ARRAY_UW -> {
-                when {
-
+                val varname = asmgen.asmVariableName(stmt.variable)
+                assignLoopvarWord(stmt, range)
+                if(stepsize==-1 && range.to.asConstInteger()==0) {
+                    // simple loop downto 0 step -1 (words)
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.statements)
+                    asmgen.out("""
+                        lda  $varname
+                        bne  ++
+                        lda  $varname+1
+                        beq  $endLabel
++                       lda  $varname
+                        bne  +
+                        dec  $varname+1
++                       dec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
+                }
+                else if (stepsize==-1 && range.to.asConstInteger()==1) {
+                    // simple loop downto 1 step -1 (words)
+                    asmgen.out(loopLabel)
+                    asmgen.translate(stmt.statements)
+                    asmgen.out("""
+                        lda  $varname
+                        cmp  #1
+                        bne  +
+                        lda  $varname+1
+                        beq  $endLabel
++                       lda  $varname
+                        bne  +
+                        dec  $varname+1
++                       dec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
+                }
+                else if (stepsize == 1 || stepsize == -1) {
                     // words, step 1 or -1
-
-                    stepsize == 1 || stepsize == -1 -> {
-                        val varname = asmgen.asmVariableName(stmt.variable)
-                        assignLoopvarWord(stmt, range)
-                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
-                        precheckFromToWord(iterableDt, stepsize, varname, endLabel)
-                        asmgen.out("""
-                            sty  $modifiedLabel+1
-                            sta  $modifiedLabel2+1
+                    asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                    precheckFromToWord(iterableDt, stepsize, varname, endLabel)
+                    asmgen.out("""
+                sty  $modifiedLabel+1
+                sta  $modifiedLabel2+1
 $loopLabel""")
-                        asmgen.translate(stmt.statements)
-                        asmgen.out("""
+                    asmgen.translate(stmt.statements)
+                    asmgen.out("""
                 lda  $varname+1
 $modifiedLabel  cmp  #0    ; modified 
                 bne  +
                 lda  $varname
 $modifiedLabel2 cmp  #0    ; modified 
                 beq  $endLabel""")
-                        if(stepsize==1) {
-                            asmgen.out("""
+                    if(stepsize==1) {
+                        asmgen.out("""
 +               inc  $varname
                 bne  $loopLabel
                 inc  $varname+1""")
-                            asmgen.jmp(loopLabel)
-                        } else {
-                            asmgen.out("""
+                        asmgen.jmp(loopLabel)
+                    } else {
+                        asmgen.out("""
 +               lda  $varname
                 bne  +
                 dec  $varname+1
 +               dec  $varname""")
-                            asmgen.jmp(loopLabel)
-                        }
-                        asmgen.out(endLabel)
+                        asmgen.jmp(loopLabel)
                     }
-                    stepsize > 0 -> {
-
-                        // (u)words, step >= 2
-                        val varname = asmgen.asmVariableName(stmt.variable)
-                        assignLoopvarWord(stmt, range)
-                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
-                        precheckFromToWord(iterableDt, stepsize, varname, endLabel)
-                        asmgen.out("""
-                            sty  $modifiedLabel+1
-                            sta  $modifiedLabel2+1
+                    asmgen.out(endLabel)
+                }
+                else if (stepsize > 0) {
+                    // (u)words, step >= 2
+                    asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                    precheckFromToWord(iterableDt, stepsize, varname, endLabel)
+                    asmgen.out("""
+                sty  $modifiedLabel+1
+                sta  $modifiedLabel2+1
 $loopLabel""")
-                        asmgen.translate(stmt.statements)
+                    asmgen.translate(stmt.statements)
 
-                        if (iterableDt == DataType.ARRAY_UW) {
-                            asmgen.out("""
+                    if (iterableDt == DataType.ARRAY_UW) {
+                        asmgen.out("""
                 lda  $varname
                 clc
                 adc  #<$stepsize
@@ -222,8 +266,8 @@ $modifiedLabel2 lda  #0     ; modified
                 bcc  $endLabel
                 bcs  $loopLabel
 $endLabel""")
-                        } else {
-                            asmgen.out("""
+                    } else {
+                        asmgen.out("""
                 lda  $varname
                 clc
                 adc  #<$stepsize
@@ -239,22 +283,20 @@ $modifiedLabel  lda  #0   ; modified
                 eor  #$80
 +               bpl  $loopLabel                
 $endLabel""")
-                        }
                     }
-                    else -> {
+                }
+                else {
 
-                        // (u)words, step <= -2
-                        val varname = asmgen.asmVariableName(stmt.variable)
-                        assignLoopvarWord(stmt, range)
-                        asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
-                        precheckFromToWord(iterableDt, stepsize, varname, endLabel)
-                        asmgen.out("""
-                            sty  $modifiedLabel+1
-                            sta  $modifiedLabel2+1
+                    // (u)words, step <= -2
+                    asmgen.assignExpressionToRegister(range.to, RegisterOrPair.AY)
+                    precheckFromToWord(iterableDt, stepsize, varname, endLabel)
+                    asmgen.out("""
+                sty  $modifiedLabel+1
+                sta  $modifiedLabel2+1
 $loopLabel""")
-                        asmgen.translate(stmt.statements)
+                    asmgen.translate(stmt.statements)
 
-                            asmgen.out("""
+                    asmgen.out("""
                 lda  $varname
                 sec
                 sbc  #<${stepsize.absoluteValue}
@@ -271,13 +313,12 @@ $modifiedLabel  sbc  #0    ; modified
                 eor  #$80
 +               bpl  $loopLabel                
 $endLabel""")
-                    }
                 }
             }
             else -> throw AssemblyError("range expression can only be byte or word")
         }
 
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
     private fun precheckFromToWord(iterableDt: DataType, stepsize: Int, fromVar: String, endLabel: String) {
@@ -333,7 +374,7 @@ $endLabel""")
     private fun translateForOverIterableVar(stmt: PtForLoop, iterableDt: DataType, ident: PtIdentifier) {
         val loopLabel = asmgen.makeLabel("for_loop")
         val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         val iterableName = asmgen.asmVariableName(ident)
         val numElements = when(val symbol = asmgen.symbolTable.lookup(ident.name)) {
             is StStaticVariable -> symbol.length!!
@@ -477,17 +518,19 @@ $loopLabel          sty  $indexVar
             }
             else -> throw AssemblyError("can't iterate over $iterableDt")
         }
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
     private fun translateForOverConstRange(stmt: PtForLoop, iterableDt: DataType, range: IntProgression) {
         if (range.isEmpty() || range.step==0)
             throw AssemblyError("empty range or step 0")
         if(iterableDt==DataType.ARRAY_B || iterableDt==DataType.ARRAY_UB) {
+            if(range.last==range.first) return translateForSimpleByteRangeAsc(stmt, range)
             if(range.step==1 && range.last>range.first) return translateForSimpleByteRangeAsc(stmt, range)
-            if(range.step==-1 && range.last<range.first) return translateForSimpleByteRangeDesc(stmt, range)
+            if(range.step==-1 && range.last<range.first) return translateForSimpleByteRangeDesc(stmt, range, iterableDt==DataType.ARRAY_UB)
         }
         else if(iterableDt==DataType.ARRAY_W || iterableDt==DataType.ARRAY_UW) {
+            if(range.last==range.first) return translateForSimpleWordRangeAsc(stmt, range)
             if(range.step==1 && range.last>range.first) return translateForSimpleWordRangeAsc(stmt, range)
             if(range.step==-1 && range.last<range.first) return translateForSimpleWordRangeDesc(stmt, range)
         }
@@ -495,7 +538,7 @@ $loopLabel          sty  $indexVar
         // not one of the easy cases, generate more complex code...
         val loopLabel = asmgen.makeLabel("for_loop")
         val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         when(iterableDt) {
             DataType.ARRAY_B, DataType.ARRAY_UB -> {
                 // loop over byte range via loopvar, step >= 2 or <= -2
@@ -507,7 +550,7 @@ $loopLabel""")
                 asmgen.translate(stmt.statements)
                 when (range.step) {
                     0, 1, -1 -> {
-                        throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $stmt")
+                        throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $range ${stmt.position}")
                     }
                     2 -> {
                         if(range.last==255 || range.last==254) {
@@ -601,13 +644,13 @@ $loopLabel""")
             }
             else -> throw AssemblyError("range expression can only be byte or word")
         }
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
     private fun translateForSimpleByteRangeAsc(stmt: PtForLoop, range: IntProgression) {
         val loopLabel = asmgen.makeLabel("for_loop")
         val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         val varname = asmgen.asmVariableName(stmt.variable)
         asmgen.out("""
                 lda  #${range.first}
@@ -627,50 +670,50 @@ $endLabel""")
                 bne  $loopLabel
 $endLabel""")
         }
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
-    private fun translateForSimpleByteRangeDesc(stmt: PtForLoop, range: IntProgression) {
+    private fun translateForSimpleByteRangeDesc(stmt: PtForLoop, range: IntProgression, unsigned: Boolean) {
         val loopLabel = asmgen.makeLabel("for_loop")
-        val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
         val varname = asmgen.asmVariableName(stmt.variable)
         asmgen.out("""
-                    lda  #${range.first}
-                    sta  $varname
+            lda  #${range.first}
+            sta  $varname
 $loopLabel""")
         asmgen.translate(stmt.statements)
         when (range.last) {
             0 -> {
-                asmgen.out("""
-                    lda  $varname
-                    beq  $endLabel
-                    dec  $varname""")
-                asmgen.jmp(loopLabel)
-                asmgen.out(endLabel)
+                if(!unsigned || range.first<=127) {
+                    asmgen.out("""
+                        dec  $varname
+                        bpl  $loopLabel""")
+                } else {
+                    asmgen.out("""
+                        dec  $varname
+                        lda  $varname
+                        cmp  #255
+                        bne  $loopLabel""")
+                }
             }
             1 -> {
                 asmgen.out("""
                     dec  $varname
-                    bne  $loopLabel
-$endLabel""")
+                    bne  $loopLabel""")
             }
             else -> {
                 asmgen.out("""
                     dec  $varname
                     lda  $varname
                     cmp  #${range.last-1}
-                    bne  $loopLabel
-$endLabel""")
+                    bne  $loopLabel""")
             }
         }
-        asmgen.loopEndLabels.pop()
     }
 
     private fun translateForSimpleWordRangeAsc(stmt: PtForLoop, range: IntProgression) {
         val loopLabel = asmgen.makeLabel("for_loop")
         val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         val varname = asmgen.asmVariableName(stmt.variable)
         asmgen.out("""
             lda  #<${range.first}
@@ -691,13 +734,13 @@ $loopLabel""")
             inc  $varname+1""")
         asmgen.jmp(loopLabel)
         asmgen.out(endLabel)
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
     private fun translateForSimpleWordRangeDesc(stmt: PtForLoop, range: IntProgression) {
         val loopLabel = asmgen.makeLabel("for_loop")
         val endLabel = asmgen.makeLabel("for_end")
-        asmgen.loopEndLabels.push(endLabel)
+        asmgen.loopEndLabels.add(endLabel)
         val varname = asmgen.asmVariableName(stmt.variable)
         asmgen.out("""
             lda  #<${range.first}
@@ -706,20 +749,29 @@ $loopLabel""")
             sty  $varname+1
 $loopLabel""")
         asmgen.translate(stmt.statements)
+        if(range.last==0) {
+            asmgen.out("""
+                lda  $varname
+                bne  ++
+                lda  $varname+1
+                beq  $endLabel""")
+        } else {
+            asmgen.out("""
+                lda  $varname
+                cmp  #<${range.last}
+                bne  +
+                lda  $varname+1
+                cmp  #>${range.last}
+                beq  $endLabel""")
+        }
         asmgen.out("""
-            lda  $varname
-            cmp  #<${range.last}
-            bne  +
-            lda  $varname+1
-            cmp  #>${range.last}
-            beq  $endLabel
 +           lda  $varname
             bne  +
             dec  $varname+1
 +           dec  $varname""")
         asmgen.jmp(loopLabel)
         asmgen.out(endLabel)
-        asmgen.loopEndLabels.pop()
+        asmgen.loopEndLabels.removeLast()
     }
 
     private fun assignLoopvarWord(stmt: PtForLoop, range: PtRange) =

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

@@ -15,6 +15,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
         // just ignore any result values from the function call.
     }
 
+    // TODO tweak subroutine call convention to also make use of X register to pass a third byte?
     internal fun optimizeIntArgsViaRegisters(sub: PtSub) =
         (sub.parameters.size==1 && sub.parameters[0].type in IntegerDatatypesWithBoolean)
                 || (sub.parameters.size==2 && sub.parameters[0].type in ByteDatatypesWithBoolean && sub.parameters[1].type in ByteDatatypesWithBoolean)
@@ -40,7 +41,84 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                 sub.children.forEach { asmgen.translate(it as PtInlineAssembly) }
                 asmgen.out("  \t; inlined routine end: ${sub.name}")
             } else {
-                asmgen.out("  jsr  $subAsmName")
+                val bank = sub.address?.constbank?.toString()
+                if(bank==null) {
+                    val varbank = if(sub.address?.varbank==null) null else asmgen.asmVariableName(sub.address!!.varbank!!)
+                    if(varbank!=null) {
+                        when(asmgen.options.compTarget.name) {
+                            "cx16" -> {
+                                // JSRFAR can jump to a banked RAM address as well!
+                                asmgen.out("""
+                                php
+                                pha
+                                lda  $varbank
+                                sta  +
+                                pla
+                                plp
+                                jsr  cx16.JSRFAR
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
++                               .byte  0    ; modified"""
+                                )
+                            }
+                            "c64" -> {
+                                asmgen.out("""
+                                php
+                                pha
+                                lda  $varbank
+                                sta  +
+                                pla
+                                plp
+                                jsr  c64.x16jsrfar
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
++                               .byte  0    ; modified"""
+                                )
+                            }
+                            "c128" -> {
+                                asmgen.out("""
+                                php
+                                pha
+                                lda  $varbank
+                                sta  +
+                                pla
+                                plp
+                                jsr  c128.x16jsrfar
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
++                               .byte  0    ; modified"""
+                                )
+                            }
+                            else -> throw AssemblyError("callfar is not supported on the selected compilation target")
+                        }
+                    } else {
+                        asmgen.out("  jsr  $subAsmName")
+                    }
+                }
+                else {
+                    when(asmgen.options.compTarget.name) {
+                        "cx16" -> {
+                            // JSRFAR can jump to a banked RAM address as well!
+                            asmgen.out("""
+                                jsr cx16.JSRFAR
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
+                                .byte  $bank"""
+                            )
+                        }
+                        "c64" -> {
+                            asmgen.out("""
+                                jsr  c64.x16jsrfar
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
+                                .byte  $bank"""
+                            )
+                        }
+                        "c128" -> {
+                            asmgen.out("""
+                                jsr  c128.x16jsrfar
+                                .word  $subAsmName    ; ${sub.address!!.address.toHex()}
+                                .byte  $bank"""
+                            )
+                        }
+                        else -> throw AssemblyError("callfar is not supported on the selected compilation target")
+                    }
+                }
             }
         }
         else if(sub is PtSub) {
@@ -82,7 +160,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
             is PtAddressOf -> false
             is PtIdentifier -> false
             is PtIrRegister -> false
-            is PtMemoryByte -> true     // TODO might not actually need extra registers if the value has to end up in A
+            is PtMemoryByte -> arg.address !is PtNumber && arg.address !is PtIdentifier
             is PtNumber -> false
             is PtBool -> false
             else -> true
@@ -205,7 +283,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                     if(value is PtIdentifier) {
                         val addr = PtAddressOf(Position.DUMMY)
                         addr.add(value)
-                        addr.parent = sub as PtNode
+                        addr.parent = scope as PtNode
                         AsmAssignSource.fromAstSource(addr, program, asmgen).adjustSignedUnsigned(target)
                     } else {
                         AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
@@ -213,7 +291,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                 } else {
                     AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
                 }
-                asmgen.translateNormalAssignment(AsmAssignment(src, target, program.memsizer, Position.DUMMY), scope)
+                asmgen.translateNormalAssignment(AsmAssignment(src, target, program.memsizer, target.position), scope)
             }
             return RegisterOrStatusflag(register, null)
         }

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

@@ -1,6 +1,6 @@
 package prog8.codegen.cpu6502
 
-import prog8.code.StRomSub
+import prog8.code.StExtSub
 import prog8.code.SymbolTable
 import prog8.code.ast.*
 import prog8.code.core.*
@@ -17,7 +17,7 @@ internal class IfElseAsmGen(private val program: PtProgram,
 
     fun translate(stmt: PtIfElse) {
         require(stmt.condition.type== DataType.BOOL)
-        checkNotRomsubReturnsStatusReg(stmt.condition)
+        checkNotExtsubReturnsStatusReg(stmt.condition)
 
         val jumpAfterIf = stmt.ifScope.children.singleOrNull() as? PtJump
 
@@ -43,23 +43,105 @@ internal class IfElseAsmGen(private val program: PtProgram,
 
         val prefixCond = stmt.condition as? PtPrefix
         if(prefixCond?.operator=="not") {
-            checkNotRomsubReturnsStatusReg(prefixCond.value)
-            assignConditionValueToRegisterAndTest(prefixCond.value)
-            return if(jumpAfterIf!=null)
-                translateJumpElseBodies("beq", "bne", jumpAfterIf, stmt.elseScope)
-            else
-                translateIfElseBodies("bne", stmt)
+            if(stmt.hasElse())
+                throw AssemblyError("not prefix in ifelse should have been replaced by swapped if-else blocks")
+            else {
+                checkNotExtsubReturnsStatusReg(prefixCond.value)
+                assignConditionValueToRegisterAndTest(prefixCond.value)
+                return if (jumpAfterIf != null)
+                    translateJumpElseBodies("beq", "bne", jumpAfterIf, stmt.elseScope)
+                else
+                    translateIfElseBodies("bne", stmt)
+            }
         }
 
         throw AssemblyError("weird non-boolean condition node type ${stmt.condition} at ${stmt.condition.position}")
     }
 
-    private fun checkNotRomsubReturnsStatusReg(condition: PtExpression) {
+    internal fun assignIfExpression(target: AsmAssignTarget, expr: PtIfExpression) {
+        // this is NOT for the if-else STATEMENT, but this is code for the IF-EXPRESSION.
+        require(target.datatype==expr.type)
+        val falseLabel = asmgen.makeLabel("ifexpr_false")
+        val endLabel = asmgen.makeLabel("ifexpr_end")
+        evalConditonAndBranchWhenFalse(expr.condition, falseLabel)
+        when(expr.type) {
+            in ByteDatatypesWithBoolean -> {
+                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.A, false)
+                asmgen.jmp(endLabel)
+                asmgen.out(falseLabel)
+                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.A, false)
+                asmgen.out(endLabel)
+                assignmentAsmGen.assignRegisterByte(target, CpuRegister.A, false, false)
+            }
+            in WordDatatypes -> {
+                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.AY, false)
+                asmgen.jmp(endLabel)
+                asmgen.out(falseLabel)
+                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.AY, false)
+                asmgen.out(endLabel)
+                assignmentAsmGen.assignRegisterpairWord(target, RegisterOrPair.AY)
+            }
+            DataType.FLOAT -> {
+                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.FAC1, true)
+                asmgen.jmp(endLabel)
+                asmgen.out(falseLabel)
+                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.FAC1, true)
+                asmgen.out(endLabel)
+                asmgen.assignRegister(RegisterOrPair.FAC1, target)
+            }
+            else -> throw AssemblyError("weird dt")
+        }
+    }
+
+    private fun evalConditonAndBranchWhenFalse(condition: PtExpression, falseLabel: String) {
+        if (condition is PtBinaryExpression) {
+            return when(condition.right.type) {
+                in ByteDatatypesWithBoolean -> translateIfByteConditionBranch(condition, falseLabel)
+                in WordDatatypes -> translateIfWordConditionBranch(condition, falseLabel)
+                DataType.FLOAT -> translateFloatConditionBranch(condition, falseLabel)
+                else -> throw AssemblyError("weird dt")
+            }
+        }
+        else if(condition is PtPrefix && condition.operator=="not") {
+            throw AssemblyError("not prefix in ifexpression should have been replaced by swapped values")
+        } else {
+            // 'simple' condition, check if it is a byte bittest
+            val bittest = condition as? PtBuiltinFunctionCall
+            if(bittest!=null && bittest.name.startsWith("prog8_ifelse_bittest_")) {
+                val variable = bittest.args[0] as PtIdentifier
+                val bitnumber = (bittest.args[1] as PtNumber).number.toInt()
+                val testForBitSet = bittest.name.endsWith("_set")
+                when (bitnumber) {
+                    7 -> {
+                        // test via bit + N flag
+                        asmgen.out("  bit  ${variable.name}")
+                        if(testForBitSet) asmgen.out("  bpl  $falseLabel")
+                        else asmgen.out("  bmi  $falseLabel")
+                        return
+                    }
+                    6 -> {
+                        // test via bit + V flag
+                        asmgen.out("  bit  ${variable.name}")
+                        if(testForBitSet) asmgen.out("  bvc  $falseLabel")
+                        else asmgen.out("  bvs  $falseLabel")
+                        return
+                    }
+                    else -> throw AssemblyError("prog8_ifelse_bittest can only work on bits 7 and 6")
+                }
+            }
+
+            // the condition is "simple" enough to just assign its 0/1 value to a register and branch on that
+            assignConditionValueToRegisterAndTest(condition)
+            asmgen.out("  beq  $falseLabel")
+        }
+    }
+
+    private fun checkNotExtsubReturnsStatusReg(condition: PtExpression) {
         val fcall = condition as? PtFunctionCall
         if(fcall!=null && fcall.type==DataType.BOOL) {
-            val romsub = st.lookup(fcall.name) as? StRomSub
-            if(romsub!=null && romsub.returns.any { it.register.statusflag!=null }) {
-                throw AssemblyError("if romsub() that returns a status register boolean should have been changed into a Conditional branch such as if_cc")
+            val extsub = st.lookup(fcall.name) as? StExtSub
+            if(extsub!=null && extsub.returns.any { it.register.statusflag!=null }) {
+                throw AssemblyError("if extsub() returning a status register boolean should have been changed into a Conditional branch such as if_cc")
             }
         }
     }
@@ -73,6 +155,7 @@ internal class IfElseAsmGen(private val program: PtProgram,
             is PtIrRegister,
             is PtArrayIndexer,
             is PtPrefix,
+            is PtIfExpression,
             is PtBinaryExpression -> { /* no cmp necessary the lda has been done just prior */ }
             is PtTypeCast -> {
                 if(condition.value.type !in ByteDatatypes && condition.value.type !in WordDatatypes)
@@ -276,6 +359,230 @@ internal class IfElseAsmGen(private val program: PtProgram,
         }
     }
 
+    private fun translateIfByteConditionBranch(condition: PtBinaryExpression, falseLabel: String) {
+        val signed = condition.left.type in SignedDatatypes
+        val constValue = condition.right.asConstInteger()
+        if(constValue==0) {
+            return translateIfCompareWithZeroByteBranch(condition, signed, falseLabel)
+        }
+
+        when(condition.operator) {
+            "==" -> {
+                // if X==value
+                asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.A, signed)
+                cmpAwithByteValue(condition.right, false)
+                asmgen.out("  bne  $falseLabel")
+            }
+            "!=" -> {
+                // if X!=value
+                asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.A, signed)
+                cmpAwithByteValue(condition.right, false)
+                asmgen.out("  beq  $falseLabel")
+            }
+            in LogicalOperators -> {
+                val regAtarget = AsmAssignTarget(TargetStorageKind.REGISTER, asmgen, DataType.BOOL, condition.definingISub(), condition.position, register=RegisterOrPair.A)
+                if (assignmentAsmGen.optimizedLogicalExpr(condition, regAtarget)) {
+                    asmgen.out("  beq  $falseLabel")
+                } else {
+                    errors.warn("SLOW FALLBACK FOR 'IFEXPR' CODEGEN - ask for support", condition.position)      //  should not occur ;-)
+                    assignConditionValueToRegisterAndTest(condition)
+                    asmgen.out("  beq  $falseLabel")
+                }
+            }
+            else -> {
+                // TODO don't store condition as expression result but just use the flags, like a normal PtIfElse translation does
+                // TODO: special cases for <, <=, >, >= above.
+                assignConditionValueToRegisterAndTest(condition)
+                asmgen.out("  beq  $falseLabel")
+            }
+        }
+    }
+
+    private fun translateIfWordConditionBranch(condition: PtBinaryExpression, falseLabel: String) {
+        // TODO can we reuse this whole thing from IfElse ?
+        val constValue = condition.right.asConstInteger()
+        if(constValue!=null) {
+            if (constValue == 0) {
+                when (condition.operator) {
+                    "==" -> return translateWordExprIsZero(condition.left, falseLabel)
+                    "!=" -> return translateWordExprIsNotZero(condition.left, falseLabel)
+                }
+            }
+            if (constValue != 0) {
+                when (condition.operator) {
+                    "==" -> return translateWordExprEqualsNumber(condition.left, constValue, falseLabel)
+                    "!=" -> return translateWordExprNotEqualsNumber(condition.left, constValue, falseLabel)
+                }
+            }
+        }
+        val variable = condition.right as? PtIdentifier
+        if(variable!=null) {
+            when (condition.operator) {
+                "==" -> return translateWordExprEqualsVariable(condition.left, variable, falseLabel)
+                "!=" -> return translateWordExprNotEqualsVariable(condition.left, variable, falseLabel)
+            }
+        }
+
+        // TODO don't store condition as expression result but just use the flags, like a normal PtIfElse translation does
+        assignConditionValueToRegisterAndTest(condition)
+        asmgen.out("  beq  $falseLabel")
+    }
+
+    private fun translateWordExprNotEqualsVariable(expr: PtExpression, variable: PtIdentifier, falseLabel: String) {
+        // if w!=variable
+        // TODO reuse code from ifElse?
+        val varRight = asmgen.asmVariableName(variable)
+        if(expr is PtIdentifier) {
+            val varLeft = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varLeft
+                cmp  $varRight
+                bne  +
+                lda  $varLeft+1
+                cmp  $varRight+1
+                beq  $falseLabel
++""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out("""
+                cmp  $varRight
+                bne  +
+                cpy  $varRight+1
+                beq  $falseLabel
++""")
+        }
+    }
+
+    private fun translateWordExprEqualsVariable(expr: PtExpression, variable: PtIdentifier, falseLabel: String) {
+        // if w==variable
+        // TODO reuse code from ifElse?
+        val varRight = asmgen.asmVariableName(variable)
+        if(expr is PtIdentifier) {
+            val varLeft = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varLeft
+                cmp  $varRight
+                bne  $falseLabel
+                lda  $varLeft+1
+                cmp  $varRight+1
+                bne  $falseLabel""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out("""
+                cmp  $varRight
+                bne  $falseLabel
+                cpy  $varRight+1
+                bne  $falseLabel""")
+        }
+    }
+
+    private fun translateWordExprNotEqualsNumber(expr: PtExpression, number: Int, falseLabel: String) {
+        // if w!=number
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                cmp  #<$number
+                bne  +
+                lda  $varname+1
+                cmp  #>$number
+                beq  $falseLabel
++""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out("""
+                cmp  #<$number
+                bne  +
+                cpy  #>$number
+                beq  $falseLabel
++""")
+        }
+    }
+
+    private fun translateWordExprEqualsNumber(expr: PtExpression, number: Int, falseLabel: String) {
+        // if w==number
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                cmp  #<$number
+                bne  $falseLabel
+                lda  $varname+1
+                cmp  #>$number
+                bne  $falseLabel""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out(                """
+                cmp  #<$number
+                bne  $falseLabel
+                cpy  #>$number
+                bne  $falseLabel""")
+        }
+    }
+
+    private fun translateWordExprIsNotZero(expr: PtExpression, falseLabel: String) {
+        // if w!=0
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                ora  $varname+1
+                beq  $falseLabel""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out("  sty  P8ZP_SCRATCH_REG |  ora  P8ZP_SCRATCH_REG |  beq  $falseLabel")
+        }
+    }
+
+    private fun translateWordExprIsZero(expr: PtExpression, falseLabel: String) {
+        // if w==0
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                ora  $varname+1
+                bne  $falseLabel""")
+        } else {
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.AY, false)
+            asmgen.out("  sty  P8ZP_SCRATCH_REG |  ora  P8ZP_SCRATCH_REG |  bne  $falseLabel")
+        }
+    }
+
+    private fun translateIfCompareWithZeroByteBranch(condition: PtBinaryExpression, signed: Boolean, falseLabel: String) {
+        // optimized code for byte comparisons with 0
+        assignConditionValueToRegisterAndTest(condition.left)
+        when (condition.operator) {
+            "==" -> asmgen.out("  bne  $falseLabel")
+            "!=" -> asmgen.out("  beq  $falseLabel")
+            ">" -> {
+                if(signed) asmgen.out("  bmi  $falseLabel |  beq  $falseLabel")
+                else asmgen.out("  beq  $falseLabel")
+            }
+            ">=" -> {
+                if(signed) asmgen.out("  bmi  $falseLabel")
+                else { /* always true for unsigned */ }
+            }
+            "<" -> {
+                if(signed) asmgen.out("  bpl  $falseLabel")
+                else asmgen.jmp(falseLabel)
+            }
+            "<=" -> {
+                if(signed) {
+                    // inverted '>'
+                    asmgen.out("""
+                        beq  +
+                        bpl  $falseLabel
++""")
+                } else asmgen.out("  bne  $falseLabel")
+            }
+            else -> throw AssemblyError("expected comparison operator")
+        }
+    }
+
     private fun translateIfCompareWithZeroByte(stmt: PtIfElse, signed: Boolean, jumpAfterIf: PtJump?) {
         // optimized code for byte comparisons with 0
         val condition = stmt.condition as PtBinaryExpression
@@ -1741,6 +2048,51 @@ _jump                       jmp  ($asmLabel)
         }
     }
 
+    private fun translateFloatConditionBranch(condition: PtBinaryExpression, elseLabel: String) {
+        val constValue = (condition.right as? PtNumber)?.number
+        if(constValue==0.0) {
+            if (condition.operator == "==") {
+                // if FL==0.0
+                asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.FAC1, true)
+                asmgen.out("  jsr  floats.SIGN |  cmp  #0 |  bne  $elseLabel")
+                return
+            } else if(condition.operator=="!=") {
+                // if FL!=0.0
+                asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.FAC1, true)
+                asmgen.out("  jsr  floats.SIGN |  cmp  #0 |  beq  $elseLabel")
+                return
+            }
+        }
+
+        when(condition.operator) {
+            "==" -> {
+                translateFloatsEqualsConditionIntoA(condition.left, condition.right)
+                asmgen.out("  beq  $elseLabel")
+            }
+            "!=" -> {
+                translateFloatsEqualsConditionIntoA(condition.left, condition.right)
+                asmgen.out("  bne  $elseLabel")
+            }
+            "<" -> {
+                translateFloatsLessConditionIntoA(condition.left, condition.right, false)
+                asmgen.out("  beq  $elseLabel")
+            }
+            "<=" -> {
+                translateFloatsLessConditionIntoA(condition.left, condition.right, true)
+                asmgen.out("  beq  $elseLabel")
+            }
+            ">" -> {
+                translateFloatsLessConditionIntoA(condition.left, condition.right, true)
+                asmgen.out("  bne  $elseLabel")
+            }
+            ">=" -> {
+                translateFloatsLessConditionIntoA(condition.left, condition.right, false)
+                asmgen.out("  bne  $elseLabel")
+            }
+            else -> throw AssemblyError("expected comparison operator")
+        }
+    }
+
     private fun translateFloatsEqualsConditionIntoA(left: PtExpression, right: PtExpression) {
         fun equalf(leftName: String, rightName: String) {
             asmgen.out("""

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

@@ -32,10 +32,6 @@ internal class ProgramAndVarsGen(
 
     internal fun generate() {
         header()
-        val allBlocks = program.allBlocks()
-
-        if(allBlocks.first().name != "p8b_main" && allBlocks.first().name != "main")
-            throw AssemblyError("first block should be 'main' or 'p8b_main'")
 
         if(errors.noErrors())  {
             program.allBlocks().forEach { block2asm(it) }
@@ -75,6 +71,12 @@ internal class ProgramAndVarsGen(
         asmgen.out("P8ZP_SCRATCH_REG = ${zp.SCRATCH_REG}")
         asmgen.out("P8ZP_SCRATCH_W1 = ${zp.SCRATCH_W1}    ; word")
         asmgen.out("P8ZP_SCRATCH_W2 = ${zp.SCRATCH_W2}    ; word")
+        if(compTarget.name=="c64") {
+            if(options.floats)
+                asmgen.out("PROG8_C64_BANK_CONFIG=31  ; basic+IO+kernal")
+            else
+                asmgen.out("PROG8_C64_BANK_CONFIG=30  ; IO+kernal, no basic")
+        }
         asmgen.out(".weak")   // hack to allow user to override the following two with command line redefinition (however, just use '-esa' command line option instead!)
         asmgen.out(".endweak")
 
@@ -89,6 +91,13 @@ internal class ProgramAndVarsGen(
             OutputType.RAW -> {
                 asmgen.out("; ---- raw assembler program ----")
                 asmgen.out("* = ${options.loadAddress.toHex()}")
+                asmgen.out("prog8_program_start\t; start of program label")
+                asmgen.out("  cld")
+                asmgen.out("  tsx  ; save stackpointer for sys.exit()")
+                asmgen.out("  stx  prog8_lib.orig_stackpointer")
+                if(!options.noSysInit)
+                    asmgen.out("  jsr  p8_sys_startup.init_system")
+                asmgen.out("  jsr  p8_sys_startup.init_system_phase2")
             }
             OutputType.PRG -> {
                 when(options.launcher) {
@@ -98,30 +107,43 @@ internal class ProgramAndVarsGen(
                         }
                         asmgen.out("; ---- basic program with sys call ----")
                         asmgen.out("* = ${options.loadAddress.toHex()}")
+                        asmgen.out("prog8_program_start\t; start of program label")
                         val year = LocalDate.now().year
                         asmgen.out("  .word  (+), $year")
                         asmgen.out("  .null  $9e, format(' %d ', prog8_entrypoint), $3a, $8f, ' prog8'")
                         asmgen.out("+\t.word  0")
-                        asmgen.out("prog8_entrypoint\t; assembly code starts here")
+                        asmgen.out("prog8_entrypoint")
+                        asmgen.out("  cld")
+                        asmgen.out("  tsx  ; save stackpointer for sys.exit()")
+                        asmgen.out("  stx  prog8_lib.orig_stackpointer")
                         if(!options.noSysInit)
-                            asmgen.out("  jsr  sys.init_system")
-                        asmgen.out("  jsr  sys.init_system_phase2")
+                            asmgen.out("  jsr  p8_sys_startup.init_system")
+                        asmgen.out("  jsr  p8_sys_startup.init_system_phase2")
                     }
                     CbmPrgLauncherType.NONE -> {
+                        // this is the same as RAW
                         asmgen.out("; ---- program without basic sys call ----")
                         asmgen.out("* = ${options.loadAddress.toHex()}")
+                        asmgen.out("prog8_program_start\t; start of program label")
+                        asmgen.out("  cld")
+                        asmgen.out("  tsx  ; save stackpointer for sys.exit()")
+                        asmgen.out("  stx  prog8_lib.orig_stackpointer")
                         if(!options.noSysInit)
-                            asmgen.out("  jsr  sys.init_system")
-                        asmgen.out("  jsr  sys.init_system_phase2")
+                            asmgen.out("  jsr  p8_sys_startup.init_system")
+                        asmgen.out("  jsr  p8_sys_startup.init_system_phase2")
                     }
                 }
             }
             OutputType.XEX -> {
                 asmgen.out("; ---- atari xex program ----")
                 asmgen.out("* = ${options.loadAddress.toHex()}")
+                asmgen.out("prog8_program_start\t; start of program label")
+                asmgen.out("  cld")
+                asmgen.out("  tsx  ; save stackpointer for sys.exit()")
+                asmgen.out("  stx  prog8_lib.orig_stackpointer")
                 if(!options.noSysInit)
-                    asmgen.out("  jsr  sys.init_system")
-                asmgen.out("  jsr  sys.init_system_phase2")
+                    asmgen.out("  jsr  p8_sys_startup.init_system")
+                asmgen.out("  jsr  p8_sys_startup.init_system_phase2")
             }
         }
 
@@ -139,17 +161,20 @@ internal class ProgramAndVarsGen(
                 if(options.floats)
                     asmgen.out("  lda  #4 |  sta  $01")    // to use floats, make sure Basic rom is banked in
                 asmgen.out("  jsr  p8b_main.p8s_start")
-                asmgen.out("  jmp  sys.cleanup_at_exit")
+                asmgen.out("  jmp  p8_sys_startup.cleanup_at_exit")
             }
             "c64" -> {
-                asmgen.out("  jsr  p8b_main.p8s_start |  lda  #31 |  sta  $01")
-                asmgen.out("  jmp  sys.cleanup_at_exit")
+                asmgen.out("  jsr  p8b_main.p8s_start")
+                asmgen.out("  jmp  p8_sys_startup.cleanup_at_exit")
             }
             "c128" -> {
-                asmgen.out("  jsr  p8b_main.p8s_start |  lda  #0 |  sta ${"$"}ff00")
-                asmgen.out("  jmp  sys.cleanup_at_exit")
+                asmgen.out("  jsr  p8b_main.p8s_start")
+                asmgen.out("  jmp  p8_sys_startup.cleanup_at_exit")
+            }
+            else -> {
+                asmgen.out("  jsr  p8b_main.p8s_start")
+                asmgen.out("  jmp  p8_sys_startup.cleanup_at_exit")
             }
-            else -> asmgen.jmp("p8b_main.p8s_start")
         }
     }
 
@@ -265,6 +290,8 @@ internal class ProgramAndVarsGen(
                 asmgen.out("  .cerror * > ${relocatedBssEnd.toHex()}, \"too many data for slabs_BSS section\"")
             }
         }
+        asmgen.out("  ; memtop check")
+        asmgen.out("  .cerror * >= ${options.memtopAddress.toHex()}, \"Program too long by \", * - ${(options.memtopAddress-1u).toHex()}, \" bytes, memtop=${options.memtopAddress.toHex()}\"")
     }
 
     private fun block2asm(block: PtBlock) {
@@ -272,12 +299,6 @@ internal class ProgramAndVarsGen(
         asmgen.out("; ---- block: '${block.name}' ----")
         if(block.options.address!=null)
             asmgen.out("* = ${block.options.address!!.toHex()}")
-        else {
-            if(block.options.alignment==PtBlock.BlockAlignment.WORD)
-                asmgen.out("\t.align 2")
-            else if(block.options.alignment==PtBlock.BlockAlignment.PAGE)
-                asmgen.out("\t.align $100")
-        }
 
         asmgen.out("${block.name}\t" + (if(block.options.forceOutput) ".block" else ".proc"))
         asmgen.outputSourceLine(block)
@@ -451,22 +472,11 @@ internal class ProgramAndVarsGen(
     }
 
     private fun entrypointInitialization() {
-        asmgen.out("; program startup initialization")
-        asmgen.out("  cld |  tsx |  stx  prog8_lib.orig_stackpointer    ; required for sys.exit()")
-        // set full BSS area to zero
-        asmgen.out("""
-    .if  prog8_bss_section_size>0
-    ; reset all variables in BSS section to zero
-	lda  #<prog8_bss_section_start
-	ldy  #>prog8_bss_section_start
-	sta  P8ZP_SCRATCH_W1
-	sty  P8ZP_SCRATCH_W1+1
-	ldx  #<prog8_bss_section_size
-	ldy  #>prog8_bss_section_size
-	lda  #0
-	jsr  prog8_lib.memset
-    .endif""")
+        // zero out the BSS area first, before setting the variable init values
+        // this is mainly to make sure the arrays are all zero'd out at program startup
+        asmgen.out("  jsr  prog8_lib.program_startup_clear_bss")
 
+        // initialize block-level (global) variables at program start
         blockVariableInitializers.forEach {
             if (it.value.isNotEmpty())
                 asmgen.out("  jsr  ${it.key.name}.prog8_init_vars")
@@ -509,17 +519,15 @@ internal class ProgramAndVarsGen(
 
         stringVarsWithInitInZp.forEach {
             val varname = asmgen.asmVariableName(it.name)+"_init_value"
-            outputStringvar(varname, it.value.second, it.value.first)
+            outputStringvar(varname, 0, it.value.second, it.value.first)
         }
 
         arrayVarsWithInitInZp.forEach {
             val varname = asmgen.asmVariableName(it.name)+"_init_value"
-            arrayVariable2asm(varname, it.alloc.dt, it.value, null)
+            arrayVariable2asm(varname, it.alloc.dt, 0, it.value, null)
         }
 
-        asmgen.out("""+        
-                    clv
-                    clc""")
+        asmgen.out("+")
     }
 
     private class ZpStringWithInitial(
@@ -540,8 +548,8 @@ internal class ProgramAndVarsGen(
         for (variable in vars) {
             val scopedName = variable.key
             val svar = symboltable.lookup(scopedName) as? StStaticVariable
-            if(svar?.onetimeInitializationStringValue!=null)
-                result.add(ZpStringWithInitial(scopedName, variable.value, svar.onetimeInitializationStringValue!!))
+            if(svar?.initializationStringValue!=null)
+                result.add(ZpStringWithInitial(scopedName, variable.value, svar.initializationStringValue!!))
         }
         return result
     }
@@ -552,8 +560,8 @@ internal class ProgramAndVarsGen(
         for (variable in vars) {
             val scopedName = variable.key
             val svar = symboltable.lookup(scopedName) as? StStaticVariable
-            if(svar?.onetimeInitializationArrayValue!=null)
-                result.add(ZpArrayWithInitial(scopedName, variable.value, svar.onetimeInitializationArrayValue!!))
+            if(svar?.initializationArrayValue!=null)
+                result.add(ZpArrayWithInitial(scopedName, variable.value, svar.initializationArrayValue!!))
         }
         return result
     }
@@ -579,25 +587,44 @@ internal class ProgramAndVarsGen(
         asmgen.out("")
         val (varsNoInit, varsWithInit) = variables.partition { it.uninitialized }
         if(varsNoInit.isNotEmpty()) {
-            asmgen.out("; non-zeropage variables without initialization value")
+            asmgen.out("; non-zeropage variables")
             asmgen.out("  .section BSS")
-            varsNoInit.sortedWith(compareBy<StStaticVariable> { it.name }.thenBy { it.dt }).forEach {
+            val (notAligned, aligned) = varsNoInit.partition { it.align==0 }
+            notAligned.sortedWith(compareBy<StStaticVariable> { it.name }.thenBy { it.dt }).forEach {
+                uninitializedVariable2asm(it)
+            }
+            aligned.sortedWith(compareBy<StStaticVariable> { it.align }.thenBy { it.name }.thenBy { it.dt }).forEach {
                 uninitializedVariable2asm(it)
             }
             asmgen.out("  .send BSS")
         }
 
         if(varsWithInit.isNotEmpty()) {
-            asmgen.out("; non-zeropage variables")
+            asmgen.out("; non-zeropage variables with init value")
             val (stringvars, othervars) = varsWithInit.sortedBy { it.name }.partition { it.dt == DataType.STR }
-            stringvars.forEach {
+            val (notAlignedStrings, alignedStrings) = stringvars.partition { it.align==0 }
+            val (notAlignedOther, alignedOther) = othervars.partition { it.align==0 }
+            notAlignedStrings.forEach {
                 outputStringvar(
                     it.name,
-                    it.onetimeInitializationStringValue!!.second,
-                    it.onetimeInitializationStringValue!!.first
+                    it.align,
+                    it.initializationStringValue!!.second,
+                    it.initializationStringValue!!.first
                 )
             }
-            othervars.sortedBy { it.type }.forEach {
+            alignedStrings.sortedBy { it.align }.forEach {
+                outputStringvar(
+                    it.name,
+                    it.align,
+                    it.initializationStringValue!!.second,
+                    it.initializationStringValue!!.first
+                )
+            }
+
+            notAlignedOther.sortedBy { it.type }.forEach {
+                staticVariable2asm(it)
+            }
+            alignedOther.sortedBy { it.align }.sortedBy { it.type }.forEach {
                 staticVariable2asm(it)
             }
         }
@@ -611,11 +638,13 @@ internal class ProgramAndVarsGen(
             DataType.WORD -> asmgen.out("${variable.name}\t.sint  ?")
             DataType.FLOAT -> asmgen.out("${variable.name}\t.fill  ${compTarget.machine.FLOAT_MEM_SIZE}")
             in SplitWordArrayTypes -> {
+                alignVar(variable.align)
                 val numbytesPerHalf = compTarget.memorySize(variable.dt, variable.length!!) / 2
                 asmgen.out("${variable.name}_lsb\t.fill  $numbytesPerHalf")
                 asmgen.out("${variable.name}_msb\t.fill  $numbytesPerHalf")
             }
             in ArrayDatatypes -> {
+                alignVar(variable.align)
                 val numbytes = compTarget.memorySize(variable.dt, variable.length!!)
                 asmgen.out("${variable.name}\t.fill  $numbytes")
             }
@@ -625,13 +654,18 @@ internal class ProgramAndVarsGen(
         }
     }
 
+    private fun alignVar(align: Int) {
+        if(align > 1)
+            asmgen.out("  .align  ${align.toHex()}")
+    }
+
     private fun staticVariable2asm(variable: StStaticVariable) {
         val initialValue: Number =
-            if(variable.onetimeInitializationNumericValue!=null) {
+            if(variable.initializationNumericValue!=null) {
                 if(variable.dt== DataType.FLOAT)
-                    variable.onetimeInitializationNumericValue!!
+                    variable.initializationNumericValue!!
                 else
-                    variable.onetimeInitializationNumericValue!!.toInt()
+                    variable.initializationNumericValue!!.toInt()
             } else 0
 
         when (variable.dt) {
@@ -650,14 +684,17 @@ internal class ProgramAndVarsGen(
             DataType.STR -> {
                 throw AssemblyError("all string vars should have been interned into prog")
             }
-            in ArrayDatatypes -> arrayVariable2asm(variable.name, variable.dt, variable.onetimeInitializationArrayValue, variable.length)
+            in ArrayDatatypes -> {
+                arrayVariable2asm(variable.name, variable.dt, variable.align, variable.initializationArrayValue, variable.length)
+            }
             else -> {
                 throw AssemblyError("weird dt")
             }
         }
     }
 
-    private fun arrayVariable2asm(varname: String, dt: DataType, value: StArray?, orNumberOfZeros: Int?) {
+    private fun arrayVariable2asm(varname: String, dt: DataType, align: Int, value: StArray?, orNumberOfZeros: Int?) {
+        alignVar(align)
         when(dt) {
             DataType.ARRAY_UB, DataType.ARRAY_BOOL -> {
                 val data = makeArrayFillDataUnsigned(dt, value, orNumberOfZeros)
@@ -749,11 +786,16 @@ internal class ProgramAndVarsGen(
             .filter { it is PtAsmSub && it.address!=null }
             .forEach { asmsub ->
                 asmsub as PtAsmSub
-                asmgen.out("  ${asmsub.name} = ${asmsub.address!!.toHex()}")
+                val address = asmsub.address!!
+                val bank = if(address.constbank!=null) "; @bank ${address.constbank}"
+                    else if(address.varbank!=null) "; @bank ${address.varbank?.name}"
+                    else ""
+                asmgen.out("  ${asmsub.name} = ${address.address.toHex()} $bank")
             }
     }
 
-    private fun outputStringvar(varname: String, encoding: Encoding, value: String) {
+    private fun outputStringvar(varname: String, align: Int, encoding: Encoding, value: String) {
+        alignVar(align)
         asmgen.out("$varname\t; $encoding:\"${value.escape().replace("\u0000", "<NULL>")}\"", false)
         val bytes = compTarget.encodeString(value, encoding).plus(0.toUByte())
         val outputBytes = bytes.map { "$" + it.toString(16).padStart(2, '0') }

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

@@ -49,8 +49,8 @@ internal class VariableAllocator(private val symboltable: SymbolTable,
         val varsRequiringZp = allVariables.filter { it.zpwish == ZeropageWish.REQUIRE_ZEROPAGE }
         val varsPreferringZp = allVariables.filter { it.zpwish == ZeropageWish.PREFER_ZEROPAGE }
         val varsNotZp = allVariables.filter { it.zpwish == ZeropageWish.NOT_IN_ZEROPAGE }
-        val varsDontCare = allVariables.filter { it.zpwish == ZeropageWish.DONTCARE }
-        require(varsDontCare.size + varsRequiringZp.size + varsPreferringZp.size + varsNotZp.size == numberOfAllocatableVariables)
+        val (varsDontCareWithoutAlignment, varsDontCareWithAlignment) = allVariables.filter { it.zpwish == ZeropageWish.DONTCARE }.partition { it.align == 0 }
+        require(varsDontCareWithAlignment.size + varsDontCareWithoutAlignment.size + varsRequiringZp.size + varsPreferringZp.size + varsNotZp.size == numberOfAllocatableVariables)
 
         var numVariablesAllocatedInZP = 0
         var numberOfNonIntegerVariables = 0
@@ -89,7 +89,7 @@ internal class VariableAllocator(private val symboltable: SymbolTable,
             // try to allocate the "don't care" interger variables into the zeropage until it is full.
             // TODO some form of intelligent priorization? most often used variables first? loopcounter vars first? ...?
             if(errors.noErrors()) {
-                val sortedList = varsDontCare.sortedByDescending { it.scopedName }
+                val sortedList = varsDontCareWithoutAlignment.sortedByDescending { it.scopedName }
                 for (variable in sortedList) {
                     if(variable.dt in IntegerDatatypesWithBoolean) {
                         if(zeropage.free.isEmpty()) {

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

@@ -175,17 +175,14 @@ 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()
-                }
+                // C64 math library has a quirk: you have always make sure FAC2/ARG is loaded last (done using CONUPK)
+                // otherwise the result of certain floating point operations such as FDIVT will be wrong.
+                // see https://www.c64-wiki.com/wiki/CONUPK
+                // Unfortunately this means we have to push and pop an intermediary floating point value to and from memory.
+                asmgen.assignExpressionToRegister(right, RegisterOrPair.FAC1, true)
+                asmgen.pushFAC1()
+                asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)
+                asmgen.popFAC2()
             }
             Cx16Target.NAME -> {
                 asmgen.assignExpressionToRegister(left, RegisterOrPair.FAC1, true)

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

@@ -1,8 +1,8 @@
 package prog8.codegen.cpu6502.assignment
 
 import prog8.code.StMemVar
-import prog8.code.StRomSub
-import prog8.code.StRomSubParameter
+import prog8.code.StExtSub
+import prog8.code.StExtSubParameter
 import prog8.code.StStaticVariable
 import prog8.code.ast.*
 import prog8.code.core.*
@@ -11,11 +11,12 @@ import prog8.codegen.cpu6502.VariableAllocator
 import prog8.codegen.cpu6502.returnsWhatWhere
 
 
-internal class AssignmentAsmGen(private val program: PtProgram,
-                                private val asmgen: AsmGen6502Internal,
-                                private val options: CompilationOptions,
-                                private val anyExprGen: AnyExprAsmGen,
-                                private val allocator: VariableAllocator) {
+internal class AssignmentAsmGen(
+    private val program: PtProgram,
+    private val asmgen: AsmGen6502Internal,
+    private val anyExprGen: AnyExprAsmGen,
+    private val allocator: VariableAllocator
+) {
     private val augmentableAsmGen = AugmentableAssignmentAsmGen(program, this, asmgen, allocator)
 
     fun translate(assignment: PtAssignment) {
@@ -38,7 +39,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
         val values = assignment.value as? PtFunctionCall
             ?: throw AssemblyError("only function calls can return multiple values in a multi-assign")
 
-        val sub = asmgen.symbolTable.lookup(values.name) as? StRomSub
+        val sub = asmgen.symbolTable.lookup(values.name) as? StExtSub
             ?: throw AssemblyError("only asmsubs can return multiple values")
 
         require(sub.returns.size>=2)
@@ -47,7 +48,27 @@ internal class AssignmentAsmGen(private val program: PtProgram,
 
         asmgen.translate(values)
 
-        fun needsToSaveA(registersResults: List<Pair<StRomSubParameter, PtNode>>): Boolean =
+        val assignmentTargets = assignment.children.dropLast(1)
+        if(sub.returns.size==assignmentTargets.size) {
+            // because we can only handle integer results right now we can just zip() it all up
+            val (statusFlagResults, registersResults) = sub.returns.zip(assignmentTargets).partition { it.first.register.statusflag!=null }
+            if (statusFlagResults.isEmpty())
+                assignRegisterResults(registersResults)
+            else if(registersResults.isEmpty())
+                assignOnlyTheStatusFlagsResults(false, statusFlagResults)
+            else
+                assignStatusFlagsAndRegistersResults(statusFlagResults, registersResults)
+        } else {
+            throw AssemblyError("number of values and targets don't match")
+        }
+    }
+
+    private fun assignStatusFlagsAndRegistersResults(
+        statusFlagResults: List<Pair<StExtSubParameter, PtNode>>,
+        registersResults: List<Pair<StExtSubParameter, PtNode>>
+    ) {
+
+        fun needsToSaveA(registersResults: List<Pair<StExtSubParameter, PtNode>>): Boolean =
             if(registersResults.isEmpty())
                 false
             else if(registersResults.all { (it.second as PtAssignTarget).identifier!=null})
@@ -55,124 +76,123 @@ internal class AssignmentAsmGen(private val program: PtProgram,
             else
                 true
 
-        fun assignCarryFlagResult(target: PtAssignTarget, saveA: Boolean) {
-            if(saveA) asmgen.out("  pha")
-            asmgen.out("  lda  #0  |  rol  a")
-            val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
-            assignRegisterByte(tgt, CpuRegister.A, false, false)
-            if(saveA) asmgen.out("  pla")
+        if(registersResults.all {
+                val tgt = it.second as PtAssignTarget
+                tgt.void || tgt.identifier!=null})
+        {
+            // all other results are just stored into identifiers directly so first handle those
+            // (simple store instructions that don't modify the carry flag)
+            assignRegisterResults(registersResults)
+            assignOnlyTheStatusFlagsResults(false, statusFlagResults)
+        } else {
+            val saveA = needsToSaveA(registersResults)
+            assignOnlyTheStatusFlagsResults(saveA, statusFlagResults)
+            assignRegisterResults(registersResults)
+        }
+    }
+
+    private fun assignOnlyTheStatusFlagsResults(saveA: Boolean, statusFlagResults: List<Pair<StExtSubParameter, PtNode>>) {
+        // assigning flags to their variables targets requires load-instructions that destroy flags
+        // so if there's more than 1, we need to save and restore the flags
+        val saveFlags = statusFlagResults.size>1
+
+        fun hasFlag(statusFlagResults: List<Pair<StExtSubParameter, PtNode>>, flag: Statusflag): PtAssignTarget? {
+            for ((returns, target) in statusFlagResults) {
+                if(returns.register.statusflag!! == flag)
+                    return target as PtAssignTarget
+            }
+            return null
         }
 
-        fun assignZeroFlagResult(target: PtAssignTarget, saveA: Boolean) {
-            if(saveA) asmgen.out("  pha")
-            asmgen.out("""
+        val targetCarry = hasFlag(statusFlagResults, Statusflag.Pc)
+        val targetZero = hasFlag(statusFlagResults, Statusflag.Pz)
+        val targetNeg = hasFlag(statusFlagResults, Statusflag.Pn)
+        val targetOverflow = hasFlag(statusFlagResults, Statusflag.Pv)
+
+        if(saveA) asmgen.out("  pha")
+        if(targetZero!=null && !targetZero.void)
+            assignZeroFlagResult(targetZero, saveFlags)
+        if(targetNeg!=null && !targetNeg.void)
+            assignNegativeFlagResult(targetNeg, saveFlags)
+        if(targetCarry!=null && !targetCarry.void)
+            assignCarryFlagResult(targetCarry)
+        if(targetOverflow!=null && !targetOverflow.void)
+            assignOverflowFlagResult(targetOverflow)
+        if(saveA) asmgen.out("  pla")
+    }
+
+    private fun assignRegisterResults(registersResults: List<Pair<StExtSubParameter, PtNode>>) {
+        registersResults.forEach { (returns, target) ->
+            target as PtAssignTarget
+            if(!target.void) {
+                val targetIdent = target.identifier
+                val targetMem = target.memory
+                if(targetIdent!=null || targetMem!=null) {
+                    val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
+                    when(returns.type) {
+                        in ByteDatatypesWithBoolean -> {
+                            if(returns.register.registerOrPair in Cx16VirtualRegisters) {
+                                assignVirtualRegister(tgt, returns.register.registerOrPair!!)
+                            } else {
+                                assignRegisterByte(tgt, returns.register.registerOrPair!!.asCpuRegister(), false, false)
+                            }
+                        }
+                        in WordDatatypes -> {
+                            assignRegisterpairWord(tgt, returns.register.registerOrPair!!)
+                        }
+                        else -> throw AssemblyError("weird dt")
+                    }
+                }
+                else TODO("array target for multi-value assignment")        // Not done yet due to result register clobbering complexity
+            }
+        }
+    }
+
+    private fun assignCarryFlagResult(target: PtAssignTarget) {
+        // overflow is not clobbered so no need to save/restore it
+        asmgen.out("  lda  #0  |  rol  a")
+        val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
+        assignRegisterByte(tgt, CpuRegister.A, false, false)
+    }
+
+    private fun assignZeroFlagResult(target: PtAssignTarget, saveFlags: Boolean) {
+        if(saveFlags) asmgen.out("  php")
+        asmgen.out("""
                 beq  +
                 lda  #0
                 beq  ++
 +               lda  #1
 +""")
-            val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
-            assignRegisterByte(tgt, CpuRegister.A, false, false)
-            if(saveA) asmgen.out("  pla")
-        }
+        val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
+        assignRegisterByte(tgt, CpuRegister.A, false, false)
+        if(saveFlags) asmgen.out("  plp")
+    }
 
-        fun assignNegativeFlagResult(target: PtAssignTarget, saveA: Boolean) {
-            if(saveA) asmgen.out("  pha")
-            asmgen.out("""
+    private fun assignNegativeFlagResult(target: PtAssignTarget, saveFlags: Boolean) {
+        if(saveFlags) asmgen.out("  php")
+        asmgen.out("""
                 bmi  +
                 lda  #0
                 beq  ++
 +               lda  #1
 +""")
-            val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
-            assignRegisterByte(tgt, CpuRegister.A, false, false)
-            if(saveA) asmgen.out("  pla")
-        }
+        val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
+        assignRegisterByte(tgt, CpuRegister.A, false, false)
+        if(saveFlags) asmgen.out("  plp")
+    }
 
-        fun assignOverflowFlagResult(target: PtAssignTarget, saveA: Boolean) {
-            if(saveA) asmgen.out("  pha")
-            asmgen.out("""
+    private fun assignOverflowFlagResult(target: PtAssignTarget) {
+        // overflow is not clobbered so no need to save/restore it
+        asmgen.out("""
                 bvs  +
                 lda  #0
                 beq  ++
 +               lda  #1
 +""")
-            val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
-            assignRegisterByte(tgt, CpuRegister.A, false, false)
-            if(saveA) asmgen.out("  pla")
-        }
-
-        fun assignRegisterResults(registersResults: List<Pair<StRomSubParameter, PtNode>>) {
-            registersResults.forEach { (returns, target) ->
-                target as PtAssignTarget
-                if(!target.void) {
-                    val targetIdent = target.identifier
-                    val targetMem = target.memory
-                    if(targetIdent!=null || targetMem!=null) {
-                        val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
-                        when(returns.type) {
-                            in ByteDatatypesWithBoolean -> {
-                                if(returns.register.registerOrPair in Cx16VirtualRegisters) {
-                                    assignVirtualRegister(tgt, returns.register.registerOrPair!!)
-                                } else {
-                                    assignRegisterByte(tgt, returns.register.registerOrPair!!.asCpuRegister(), false, false)
-                                }
-                            }
-                            in WordDatatypes -> {
-                                assignRegisterpairWord(tgt, returns.register.registerOrPair!!)
-                            }
-                            else -> throw AssemblyError("weird dt")
-                        }
-                    }
-                    else TODO("array target for multi-value assignment")        // Not done yet due to result register clobbering complexity
-                }
-            }
-        }
-
-        val assignmentTargets = assignment.children.dropLast(1)
-        if(sub.returns.size==assignmentTargets.size) {
-            // because we can only handle integer results right now we can just zip() it all up
-            val (statusFlagResults, registersResults) = sub.returns.zip(assignmentTargets).partition { it.first.register.statusflag!=null }
-            if(statusFlagResults.isNotEmpty()) {
-                if(statusFlagResults.size>1)
-                    TODO("handle multiple status flag results")
-                val (returns, target) = statusFlagResults.single()
-                target as PtAssignTarget
-                if(target.void) {
-                    // forget about the Carry status flag, only assign the normal return values
-                    assignRegisterResults(registersResults)
-                    return
-                }
-                if(registersResults.all {
-                    val tgt = it.second as PtAssignTarget
-                    tgt.void || tgt.identifier!=null})
-                {
-                    // all other results are just stored into identifiers directly so first handle those
-                    // (simple store instructions that don't modify the carry flag)
-                    assignRegisterResults(registersResults)
-                    return when(returns.register.statusflag!!) {
-                        Statusflag.Pc -> assignCarryFlagResult(target, false)
-                        Statusflag.Pz -> assignZeroFlagResult(target, false)
-                        Statusflag.Pv -> assignOverflowFlagResult(target, false)
-                        Statusflag.Pn -> assignNegativeFlagResult(target, false)
-                    }
-                }
-
-                val saveA = needsToSaveA(registersResults)
-                when(returns.register.statusflag!!) {
-                    Statusflag.Pc -> assignCarryFlagResult(target, saveA)
-                    Statusflag.Pz -> assignZeroFlagResult(target, saveA)
-                    Statusflag.Pv -> assignOverflowFlagResult(target, saveA)
-                    Statusflag.Pn -> assignNegativeFlagResult(target, saveA)
-                }
-            }
-            assignRegisterResults(registersResults)
-        } else {
-            throw AssemblyError("number of values and targets don't match")
-        }
+        val tgt = AsmAssignTarget.fromAstAssignment(target, target.definingISub(), asmgen)
+        assignRegisterByte(tgt, CpuRegister.A, false, false)
     }
 
-
     fun translateNormalAssignment(assign: AsmAssignment, scope: IPtSubroutine?) {
         when(assign.source.kind) {
             SourceStorageKind.LITERALBOOLEAN -> {
@@ -201,7 +221,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 val variable = assign.source.asmVarname
                 when (assign.target.datatype) {
                     DataType.BOOL -> {
-                        if (assign.source.datatype == DataType.BOOL || !options.strictBool) assignVariableByte(assign.target, variable)
+                        if (assign.source.datatype == DataType.BOOL) assignVariableByte(assign.target, variable)
                         else throw AssemblyError("assigning non-bool variable to boolean, should have been typecasted")
                     }
                     DataType.UBYTE, DataType.BYTE -> assignVariableByte(assign.target, variable)
@@ -335,7 +355,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                             return
                         }
                         is PtNumber -> {
-                            val indexValue = index.number.toString()
+                            val indexValue = index.number.toInt().toString()
                             asmgen.out("""
                                 lda  $pointer
                                 sta  P8ZP_SCRATCH_W2
@@ -384,7 +404,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                             return
                         }
                         is PtNumber -> {
-                            val indexValue = index.number.toString()
+                            val indexValue = index.number.toInt().toString()
                             asmgen.out("""
                                 tax
                                 lda  $pointer
@@ -619,6 +639,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                     throw AssemblyError("Expression is too complex to translate into assembly. Split it up into several separate statements, introduce a temporary variable, or otherwise rewrite it. Location: $pos")
                 }
             }
+            is PtIfExpression -> asmgen.assignIfExpression(assign.target, value)
             else -> throw AssemblyError("weird assignment value type $value")
         }
     }
@@ -915,7 +936,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 if(!directIntoY(expr.right)) asmgen.out("  pha")
                 assignExpressionToRegister(expr.right, RegisterOrPair.Y, false)
                 if(!directIntoY(expr.right)) asmgen.out("  pla")
-                asmgen.out("  jsr  math.divmod_ub_asm")
+                asmgen.out("  jsr  prog8_math.divmod_ub_asm")
                 if(target.register==RegisterOrPair.A)
                     asmgen.out("  cmp  #0")     // fix the status register
                 else
@@ -924,7 +945,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
             }
             DataType.UWORD -> {
                 asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "P8ZP_SCRATCH_W1")
-                asmgen.out("  jsr  math.divmod_uw_asm")
+                asmgen.out("  jsr  prog8_math.divmod_uw_asm")
                 assignVariableWord(target, "P8ZP_SCRATCH_W2", DataType.UWORD)
                 return true
             }
@@ -940,7 +961,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 if(!directIntoY(expr.right)) asmgen.out("  pha")
                 assignExpressionToRegister(expr.right, RegisterOrPair.Y, false)
                 if(!directIntoY(expr.right)) asmgen.out("  pla")
-                asmgen.out("  jsr  math.divmod_ub_asm")
+                asmgen.out("  jsr  prog8_math.divmod_ub_asm")
                 assignRegisterByte(target, CpuRegister.Y, false, true)
                 return true
             }
@@ -949,19 +970,19 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 if(!directIntoY(expr.right)) asmgen.out("  pha")
                 assignExpressionToRegister(expr.right, RegisterOrPair.Y, true)
                 if(!directIntoY(expr.right)) asmgen.out("  pla")
-                asmgen.out("  jsr  math.divmod_b_asm")
+                asmgen.out("  jsr  prog8_math.divmod_b_asm")
                 assignRegisterByte(target, CpuRegister.Y, true, true)
                 return true
             }
             DataType.UWORD -> {
                 asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "P8ZP_SCRATCH_W1")
-                asmgen.out("  jsr  math.divmod_uw_asm")
+                asmgen.out("  jsr  prog8_math.divmod_uw_asm")
                 assignRegisterpairWord(target, RegisterOrPair.AY)
                 return true
             }
             DataType.WORD -> {
                 asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "P8ZP_SCRATCH_W1")
-                asmgen.out("  jsr  math.divmod_w_asm")
+                asmgen.out("  jsr  prog8_math.divmod_w_asm")
                 assignRegisterpairWord(target, RegisterOrPair.AY)
                 return true
             }
@@ -978,7 +999,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                     if(!directIntoY(expr.right)) asmgen.out("  pha")
                     assignExpressionToRegister(expr.right, RegisterOrPair.Y, expr.type in SignedDatatypes)
                     if(!directIntoY(expr.right)) asmgen.out("  pla")
-                    asmgen.out("  jsr  math.multiply_bytes")
+                    asmgen.out("  jsr  prog8_math.multiply_bytes")
                     assignRegisterByte(target, CpuRegister.A, false, true)
                     return true
                 }
@@ -1001,8 +1022,8 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                         assignRegisterpairWord(target, RegisterOrPair.AY)
                         return true
                     } else {
-                        asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "math.multiply_words.multiplier")
-                        asmgen.out("  jsr  math.multiply_words")
+                        asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "prog8_math.multiply_words.multiplier")
+                        asmgen.out("  jsr  prog8_math.multiply_words")
                         assignRegisterpairWord(target, RegisterOrPair.AY)
                     }
                     return true
@@ -1014,16 +1035,16 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 in ByteDatatypes -> {
                     assignExpressionToRegister(expr.left, RegisterOrPair.A, expr.type in SignedDatatypes)
                     if (value in asmgen.optimizedByteMultiplications)
-                        asmgen.out("  jsr  math.mul_byte_${value}")
+                        asmgen.out("  jsr  prog8_math.mul_byte_${value}")
                     else
-                        asmgen.out("  ldy  #$value |  jsr  math.multiply_bytes")
+                        asmgen.out("  ldy  #$value |  jsr  prog8_math.multiply_bytes")
                     assignRegisterByte(target, CpuRegister.A, false, true)
                     return true
                 }
                 in WordDatatypes -> {
                     if (value in asmgen.optimizedWordMultiplications) {
                         assignExpressionToRegister(expr.left, RegisterOrPair.AY, expr.type in SignedDatatypes)
-                        asmgen.out("  jsr  math.mul_word_${value}")
+                        asmgen.out("  jsr  prog8_math.mul_word_${value}")
                     }
                     else {
                         if(expr.definingBlock()!!.options.veraFxMuls){
@@ -1034,8 +1055,8 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                                 sty  cx16.r0+1
                                 jsr  verafx.muls""")
                         } else {
-                            asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "math.multiply_words.multiplier")
-                            asmgen.out("  jsr  math.multiply_words")
+                            asmgen.assignWordOperandsToAYAndVar(expr.right, expr.left, "prog8_math.multiply_words.multiplier")
+                            asmgen.out("  jsr  prog8_math.multiply_words")
                         }
                     }
                     assignRegisterpairWord(target, RegisterOrPair.AY)
@@ -1072,11 +1093,11 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 asmgen.restoreRegisterStack(CpuRegister.Y, true)
                 if(expr.operator==">>")
                     if(signed)
-                        asmgen.out("  jsr  math.lsr_byte_A")
+                        asmgen.out("  jsr  prog8_math.lsr_byte_A")
                     else
-                        asmgen.out("  jsr  math.lsr_ubyte_A")
+                        asmgen.out("  jsr  prog8_math.lsr_ubyte_A")
                 else
-                    asmgen.out("  jsr  math.asl_byte_A")
+                    asmgen.out("  jsr  prog8_math.asl_byte_A")
                 assignRegisterByte(target, CpuRegister.A, signed, true)
                 return true
             } else {
@@ -1084,11 +1105,11 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 asmgen.restoreRegisterStack(CpuRegister.X, true)
                 if(expr.operator==">>")
                     if(signed)
-                        asmgen.out("  jsr  math.lsr_word_AY")
+                        asmgen.out("  jsr  prog8_math.lsr_word_AY")
                     else
-                        asmgen.out("  jsr  math.lsr_uword_AY")
+                        asmgen.out("  jsr  prog8_math.lsr_uword_AY")
                 else
-                    asmgen.out("  jsr  math.asl_word_AY")
+                    asmgen.out("  jsr  prog8_math.asl_word_AY")
                 assignRegisterpairWord(target, RegisterOrPair.AY)
                 return true
             }
@@ -1108,7 +1129,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                                 if(shifts==1)
                                     asmgen.out("  cmp  #$80 |  ror  a")
                                 else
-                                    asmgen.out("  ldy  #$shifts |  jsr  math.lsr_byte_A")
+                                    asmgen.out("  ldy  #$shifts |  jsr  prog8_math.lsr_byte_A")
                             } else {
                                 repeat(shifts) {
                                     asmgen.out("  lsr  a")
@@ -1120,7 +1141,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                     }
                     else -> {
                         if(signed && expr.operator==">>") {
-                            asmgen.out("  ldy  #$shifts |  jsr  math.lsr_byte_A")
+                            asmgen.out("  ldy  #$shifts |  jsr  prog8_math.lsr_byte_A")
                         } else {
                             asmgen.out("  lda  #0")
                         }
@@ -1153,7 +1174,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                                         ror  a""")
                                 }
                                 else
-                                    asmgen.out("  ldx  #$shifts |  jsr  math.lsr_word_AY")
+                                    asmgen.out("  ldx  #$shifts |  jsr  prog8_math.lsr_word_AY")
                             } else {
                                 if(shifts>0) {
                                     asmgen.out("  sty  P8ZP_SCRATCH_B1")
@@ -1177,16 +1198,16 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                         } else {
                             asmgen.out("  ldx  #$shifts")
                             if(signed)
-                                asmgen.out("  jsr  math.lsr_word_AY")
+                                asmgen.out("  jsr  prog8_math.lsr_word_AY")
                             else
-                                asmgen.out("  jsr  math.lsr_uword_AY")
+                                asmgen.out("  jsr  prog8_math.lsr_uword_AY")
                         }
                         assignRegisterpairWord(target, RegisterOrPair.AY)
                         return true
                     }
                     else -> {
                         if(signed && expr.operator==">>") {
-                            asmgen.out("  ldx  #$shifts |  jsr  math.lsr_word_AY")
+                            asmgen.out("  ldx  #$shifts |  jsr  prog8_math.lsr_word_AY")
                         } else {
                             asmgen.out("  lda  #0 |  ldy  #0")
                         }
@@ -1483,11 +1504,16 @@ internal class AssignmentAsmGen(private val program: PtProgram,
         // special optimization for  bytevalue +/- pointervar[y]  (actually: bytevalue +/-  @(address) )
         val address = mem.address as? PtBinaryExpression
         if(address is PtBinaryExpression) {
+            val constOffset = address.right.asConstInteger()
+            // check that the offset is actually a byte (so that it fits in a single register)
+            if(constOffset==null && address.right.type !in ByteDatatypes)
+                return false
+            if(constOffset!=null && constOffset !in -128..255)
+                return false
             val ptrVar = address.left as? PtIdentifier
             if(ptrVar!=null && asmgen.isZpVar(ptrVar)) {
                 assignExpressionToRegister(value, RegisterOrPair.A, false)
                 val pointername = asmgen.asmVariableName(ptrVar)
-                val constOffset = address.right.asConstInteger()
                 if (constOffset != null && constOffset < 256) {
                     // we have value + @(zpptr + 255), or value - @(zpptr+255)
                     asmgen.out("  ldy  #$constOffset")
@@ -1805,9 +1831,56 @@ internal class AssignmentAsmGen(private val program: PtProgram,
     }
 
     private fun containmentCheckIntoA(containment: PtContainmentCheck) {
-        val elementDt = containment.element.type
-        val symbol = asmgen.symbolTable.lookup(containment.iterable.name)!!
-        val symbolName = asmgen.asmVariableName(symbol, containment.definingSub())
+        val elementDt = containment.needle.type
+
+        if(containment.haystackValues!=null) {
+            val haystack = containment.haystackValues!!.children.map {
+                if(it is PtBool) it.asInt()
+                else (it as PtNumber).number.toInt()
+            }
+            when(elementDt) {
+                in ByteDatatypesWithBoolean -> {
+                    require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_BYTE)
+                    assignExpressionToRegister(containment.needle, RegisterOrPair.A, elementDt == DataType.BYTE)
+                    for(number in haystack) {
+                        asmgen.out("""
+                            cmp  #$number
+                            beq  +""")
+                    }
+                    asmgen.out("""
+                        lda  #0
+                        beq  ++
++                       lda  #1
++""")
+                }
+                in WordDatatypes -> {
+                    require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_WORD)
+                    assignExpressionToRegister(containment.needle, RegisterOrPair.AY, elementDt == DataType.WORD)
+                    val gottemLabel = asmgen.makeLabel("gottem")
+                    val endLabel = asmgen.makeLabel("end")
+                    for(number in haystack) {
+                        asmgen.out("""
+                            cmp  #<$number
+                            bne  +
+                            cpy  #>$number
+                            beq  $gottemLabel
++                       """)
+                    }
+                    asmgen.out("""
+                        lda  #0
+                        beq  $endLabel
+$gottemLabel            lda  #1
+$endLabel""")
+                }
+                DataType.FLOAT -> throw AssemblyError("containmentchecks for floats should always be done on an array variable with subroutine")
+                else -> throw AssemblyError("weird dt $elementDt")
+            }
+
+            return
+        }
+
+        val symbol = asmgen.symbolTable.lookup(containment.haystackHeapVar!!.name)!!
+        val symbolName = asmgen.asmVariableName(symbol, containment.definingISub())
         val (dt, numElements) = when(symbol) {
             is StStaticVariable  -> symbol.dt to symbol.length!!
             is StMemVar -> symbol.dt to symbol.length!!
@@ -1815,7 +1888,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
         }
         when(dt) {
             DataType.STR -> {
-                assignExpressionToRegister(containment.element, RegisterOrPair.A, elementDt == DataType.BYTE)
+                assignExpressionToRegister(containment.needle, RegisterOrPair.A, elementDt == DataType.BYTE)
                 asmgen.out("  pha")     // need to keep the scratch var safe so we have to do it in this order
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position,"P8ZP_SCRATCH_W1"), symbolName, null, null)
                 asmgen.out("  pla")
@@ -1823,13 +1896,13 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 asmgen.out("  jsr  prog8_lib.containment_bytearray")
             }
             DataType.ARRAY_F -> {
-                assignExpressionToRegister(containment.element, RegisterOrPair.FAC1, true)
+                assignExpressionToRegister(containment.needle, RegisterOrPair.FAC1, true)
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position, "P8ZP_SCRATCH_W1"), symbolName, null, null)
                 asmgen.out("  ldy  #$numElements")
                 asmgen.out("  jsr  floats.containment_floatarray")
             }
-            DataType.ARRAY_B, DataType.ARRAY_UB -> {
-                assignExpressionToRegister(containment.element, RegisterOrPair.A, elementDt == DataType.BYTE)
+            DataType.ARRAY_B, DataType.ARRAY_UB, DataType.ARRAY_BOOL -> {
+                assignExpressionToRegister(containment.needle, RegisterOrPair.A, elementDt == DataType.BYTE)
                 asmgen.out("  pha")     // need to keep the scratch var safe so we have to do it in this order
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position, "P8ZP_SCRATCH_W1"), symbolName, null, null)
                 asmgen.out("  pla")
@@ -1837,7 +1910,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 asmgen.out("  jsr  prog8_lib.containment_bytearray")
             }
             DataType.ARRAY_W, DataType.ARRAY_UW -> {
-                assignExpressionToVariable(containment.element, "P8ZP_SCRATCH_W1", elementDt)
+                assignExpressionToVariable(containment.needle, "P8ZP_SCRATCH_W1", elementDt)
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position, "P8ZP_SCRATCH_W2"), symbolName, null, null)
                 asmgen.out("  ldy  #$numElements")
                 asmgen.out("  jsr  prog8_lib.containment_wordarray")

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

@@ -225,7 +225,7 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                             SourceStorageKind.LITERALNUMBER -> inplacemodificationSplitWordWithLiteralval(target.asmVarname, target.datatype, index, operator, value.number!!.number.toInt())
                             else -> {
                                 // TODO: more optimized code for VARIABLE, REGISTER, MEMORY, ARRAY, EXPRESSION in the case of split-word arrays
-                                val scope = target.origAstTarget?.definingSub()
+                                val scope = target.origAstTarget?.definingISub()
                                 val regTarget = AsmAssignTarget.fromRegisters(RegisterOrPair.R0, false, target.position, scope, asmgen)
                                 val assignToReg = AsmAssignment(value, regTarget, program.memsizer, target.position)
                                 assignmentAsmGen.translateNormalAssignment(assignToReg, scope)
@@ -904,9 +904,9 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
         when (operator) {
             "+" -> asmgen.out("  clc |  adc  $otherName")
             "-" -> asmgen.out("  sec |  sbc  $otherName")
-            "*" -> asmgen.out("  ldy  $otherName |  jsr  math.multiply_bytes")
-            "/" -> asmgen.out("  ldy  $otherName |  jsr  math.divmod_ub_asm |  tya")
-            "%" -> asmgen.out("  ldy  $otherName |  jsr  math.divmod_ub_asm")
+            "*" -> asmgen.out("  ldy  $otherName |  jsr  prog8_math.multiply_bytes")
+            "/" -> asmgen.out("  ldy  $otherName |  jsr  prog8_math.divmod_ub_asm |  tya")
+            "%" -> asmgen.out("  ldy  $otherName |  jsr  prog8_math.divmod_ub_asm")
             "<<" -> {
                 asmgen.out("""
                         ldy  $otherName
@@ -981,23 +981,23 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
             "*" -> {
                 val sourceName = asmgen.loadByteFromPointerIntoA(pointervar)
                 if(value in asmgen.optimizedByteMultiplications)
-                    asmgen.out("  jsr  math.mul_byte_${value}")
+                    asmgen.out("  jsr  prog8_math.mul_byte_${value}")
                 else
-                    asmgen.out("  ldy  #$value |  jsr  math.multiply_bytes")
+                    asmgen.out("  ldy  #$value |  jsr  prog8_math.multiply_bytes")
                 asmgen.storeAIntoZpPointerVar(sourceName, false)
             }
             "/" -> {
                 val sourceName = asmgen.loadByteFromPointerIntoA(pointervar)
                 if(value==0)
                     throw AssemblyError("division by zero")
-                asmgen.out("  ldy  #$value |  jsr  math.divmod_ub_asm |  tya")
+                asmgen.out("  ldy  #$value |  jsr  prog8_math.divmod_ub_asm |  tya")
                 asmgen.storeAIntoZpPointerVar(sourceName, false)
             }
             "%" -> {
                 val sourceName = asmgen.loadByteFromPointerIntoA(pointervar)
                 if(value==0)
                     throw AssemblyError("division by zero")
-                asmgen.out("  ldy  #$value |  jsr  math.divmod_ub_asm")
+                asmgen.out("  ldy  #$value |  jsr  prog8_math.divmod_ub_asm")
                 asmgen.storeAIntoZpPointerVar(sourceName, false)
             }
             "<<" -> {
@@ -1143,17 +1143,17 @@ $shortcutLabel:""")
         when (operator) {
             "+" -> asmgen.out("  clc |  adc  $variable")
             "-" -> asmgen.out("  sec |  sbc  $variable")
-            "*" -> asmgen.out("  ldy  $variable  |  jsr  math.multiply_bytes")
+            "*" -> asmgen.out("  ldy  $variable  |  jsr  prog8_math.multiply_bytes")
             "/" -> {
                 if(signed)
-                    asmgen.out("  ldy  $variable  |  jsr  math.divmod_b_asm |  tya")
+                    asmgen.out("  ldy  $variable  |  jsr  prog8_math.divmod_b_asm |  tya")
                 else
-                    asmgen.out("  ldy  $variable  |  jsr  math.divmod_ub_asm |  tya")
+                    asmgen.out("  ldy  $variable  |  jsr  prog8_math.divmod_ub_asm |  tya")
             }
             "%" -> {
                 if(signed)
                     throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
-                asmgen.out("  ldy  $variable  |  jsr  math.divmod_ub_asm")
+                asmgen.out("  ldy  $variable  |  jsr  prog8_math.divmod_ub_asm")
             }
             "<<" -> {
                 asmgen.out("""
@@ -1318,14 +1318,14 @@ $shortcutLabel:""")
             }
             "/" -> {
                 if(signed)
-                    asmgen.out("  tay |  lda  $variable  |  jsr  math.divmod_b_asm |  tya")
+                    asmgen.out("  tay |  lda  $variable  |  jsr  prog8_math.divmod_b_asm |  tya")
                 else
-                    asmgen.out("  tay |  lda  $variable  |  jsr  math.divmod_ub_asm |  tya")
+                    asmgen.out("  tay |  lda  $variable  |  jsr  prog8_math.divmod_ub_asm |  tya")
             }
             "%" -> {
                 if(signed)
                     throw AssemblyError("remainder of signed integers is not properly defined/implemented, use unsigned instead")
-                asmgen.out("  tay |  lda  $variable  |  jsr  math.divmod_ub_asm")
+                asmgen.out("  tay |  lda  $variable  |  jsr  prog8_math.divmod_ub_asm")
             }
             "<<" -> {
                 asmgen.out("""
@@ -1478,16 +1478,16 @@ $shortcutLabel:""")
             }
             "*" -> {
                 if(value in asmgen.optimizedByteMultiplications)
-                    asmgen.out("  lda  $name |  jsr  math.mul_byte_$value |  sta  $name")
+                    asmgen.out("  lda  $name |  jsr  prog8_math.mul_byte_$value |  sta  $name")
                 else
-                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  math.multiply_bytes |  sta  $name")
+                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  prog8_math.multiply_bytes |  sta  $name")
             }
             "/" -> {
                 // replacing division by shifting is done in an optimizer step.
                 if (dt == DataType.UBYTE)
-                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  math.divmod_ub_asm |  sty  $name")
+                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  prog8_math.divmod_ub_asm |  sty  $name")
                 else
-                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  math.divmod_b_asm |  sty  $name")
+                    asmgen.out("  lda  $name |  ldy  #$value |  jsr  prog8_math.divmod_b_asm |  sty  $name")
             }
             "%" -> {
                 if(dt==DataType.BYTE)
@@ -1495,7 +1495,7 @@ $shortcutLabel:""")
                 asmgen.out("""
                     lda  $name
                     ldy  #$value
-                    jsr  math.divmod_ub_asm
+                    jsr  prog8_math.divmod_ub_asm
                     sta  $name""")
             }
             "<<" -> {
@@ -1529,7 +1529,7 @@ $shortcutLabel:""")
                             value>3 -> asmgen.out("""
                                 lda  $name
                                 ldy  #$value
-                                jsr  math.lsr_byte_A
+                                jsr  prog8_math.lsr_byte_A
                                 sta  $name""")
                             else -> repeat(value) { asmgen.out("  lda  $name | asl  a |  ror  $name") }
                         }
@@ -1802,7 +1802,7 @@ $shortcutLabel:""")
             "*" -> {
                 // the mul code works for both signed and unsigned
                 if(value in asmgen.optimizedWordMultiplications) {
-                    asmgen.out("  lda  $lsb |  ldy  $msb |  jsr  math.mul_word_$value |  sta  $lsb |  sty  $msb")
+                    asmgen.out("  lda  $lsb |  ldy  $msb |  jsr  prog8_math.mul_word_$value |  sta  $lsb |  sty  $msb")
                 } else {
                     if(block?.options?.veraFxMuls==true)
                         // cx16 verafx hardware mul
@@ -1821,12 +1821,12 @@ $shortcutLabel:""")
                     else
                         asmgen.out("""
                             lda  $lsb
-                            sta  math.multiply_words.multiplier
+                            sta  prog8_math.multiply_words.multiplier
                             lda  $msb
-                            sta  math.multiply_words.multiplier+1
+                            sta  prog8_math.multiply_words.multiplier+1
                             lda  #<$value
                             ldy  #>$value
-                            jsr  math.multiply_words
+                            jsr  prog8_math.multiply_words
                             sta  $lsb
                             sty  $msb""")
                 }
@@ -1844,7 +1844,7 @@ $shortcutLabel:""")
                             sty  P8ZP_SCRATCH_W1+1
                             lda  #<$value
                             ldy  #>$value
-                            jsr  math.divmod_w_asm
+                            jsr  prog8_math.divmod_w_asm
                             sta  $lsb
                             sty  $msb
                         """)
@@ -1857,7 +1857,7 @@ $shortcutLabel:""")
                             sty  P8ZP_SCRATCH_W1+1
                             lda  #<$value
                             ldy  #>$value
-                            jsr  math.divmod_uw_asm
+                            jsr  prog8_math.divmod_uw_asm
                             sta  $lsb
                             sty  $msb
                         """)
@@ -1876,7 +1876,7 @@ $shortcutLabel:""")
                     sty  P8ZP_SCRATCH_W1+1
                     lda  #<$value
                     ldy  #>$value
-                    jsr  math.divmod_uw_asm
+                    jsr  prog8_math.divmod_uw_asm
                     lda  P8ZP_SCRATCH_W2
                     ldy  P8ZP_SCRATCH_W2+1
                     sta  $lsb
@@ -2297,20 +2297,20 @@ $shortcutLabel:""")
                                 sty  $name+1""")
                         } else {
                             if(valueDt==DataType.UBYTE) {
-                                asmgen.out("  lda  $otherName |  sta  math.multiply_words.multiplier")
+                                asmgen.out("  lda  $otherName |  sta  prog8_math.multiply_words.multiplier")
                                 if(asmgen.isTargetCpu(CpuType.CPU65c02))
-                                    asmgen.out("  stz  math.multiply_words.multiplier+1")
+                                    asmgen.out("  stz  prog8_math.multiply_words.multiplier+1")
                                 else
-                                    asmgen.out("  lda  #0 |  sta  math.multiply_words.multiplier+1")
+                                    asmgen.out("  lda  #0 |  sta  prog8_math.multiply_words.multiplier+1")
                             } else {
                                 asmgen.out("  lda  $otherName")
                                 asmgen.signExtendAYlsb(valueDt)
-                                asmgen.out("  sta  math.multiply_words.multiplier |  sty  math.multiply_words.multiplier+1")
+                                asmgen.out("  sta  prog8_math.multiply_words.multiplier |  sty  prog8_math.multiply_words.multiplier+1")
                             }
                             asmgen.out("""
                                     lda  $name
                                     ldy  $name+1
-                                    jsr  math.multiply_words
+                                    jsr  prog8_math.multiply_words
                                     sta  $name
                                     sty  $name+1""")
                         }
@@ -2324,7 +2324,7 @@ $shortcutLabel:""")
                                 sty  P8ZP_SCRATCH_W1+1
                                 lda  $otherName
                                 ldy  #0
-                                jsr  math.divmod_uw_asm
+                                jsr  prog8_math.divmod_uw_asm
                                 sta  $name
                                 sty  $name+1
                             """)
@@ -2336,7 +2336,7 @@ $shortcutLabel:""")
                                 sty  P8ZP_SCRATCH_W1+1
                                 lda  $otherName
                                 ldy  #0
-                                jsr  math.divmod_w_asm
+                                jsr  prog8_math.divmod_w_asm
                                 sta  $name
                                 sty  $name+1
                             """)
@@ -2352,7 +2352,7 @@ $shortcutLabel:""")
                             sty  P8ZP_SCRATCH_W1+1
                             lda  $otherName
                             ldy  #0
-                            jsr  math.divmod_uw_asm
+                            jsr  prog8_math.divmod_uw_asm
                             lda  P8ZP_SCRATCH_W2
                             sta  $name
                             lda  P8ZP_SCRATCH_W2+1
@@ -2460,11 +2460,11 @@ $shortcutLabel:""")
                             asmgen.out("""
                                 lda  $otherName
                                 ldy  $otherName+1
-                                sta  math.multiply_words.multiplier
-                                sty  math.multiply_words.multiplier+1
+                                sta  prog8_math.multiply_words.multiplier
+                                sty  prog8_math.multiply_words.multiplier+1
                                 lda  $name
                                 ldy  $name+1
-                                jsr  math.multiply_words
+                                jsr  prog8_math.multiply_words
                                 sta  $name
                                 sty  $name+1""")
                     }
@@ -2477,7 +2477,7 @@ $shortcutLabel:""")
                                 sty  P8ZP_SCRATCH_W1+1
                                 lda  $otherName
                                 ldy  $otherName+1
-                                jsr  math.divmod_w_asm
+                                jsr  prog8_math.divmod_w_asm
                                 sta  $name
                                 sty  $name+1""")
                         }
@@ -2489,7 +2489,7 @@ $shortcutLabel:""")
                                 sty  P8ZP_SCRATCH_W1+1
                                 lda  $otherName
                                 ldy  $otherName+1
-                                jsr  math.divmod_uw_asm
+                                jsr  prog8_math.divmod_uw_asm
                                 sta  $name
                                 sty  $name+1""")
                         }
@@ -2504,7 +2504,7 @@ $shortcutLabel:""")
                             sty  P8ZP_SCRATCH_W1+1
                             lda  $otherName
                             ldy  $otherName+1
-                            jsr  math.divmod_uw_asm
+                            jsr  prog8_math.divmod_uw_asm
                             lda  P8ZP_SCRATCH_W2
                             sta  $name
                             lda  P8ZP_SCRATCH_W2+1
@@ -2661,11 +2661,11 @@ $shortcutLabel:""")
                 """)
             else
                 asmgen.out("""
-                    sta  math.multiply_words.multiplier
-                    sty  math.multiply_words.multiplier+1
+                    sta  prog8_math.multiply_words.multiplier
+                    sty  prog8_math.multiply_words.multiplier+1
                     lda  $name
                     ldy  $name+1
-                    jsr  math.multiply_words
+                    jsr  prog8_math.multiply_words
                     sta  $name
                     sty  $name+1
                 """)
@@ -2680,9 +2680,9 @@ $shortcutLabel:""")
                     sta  P8ZP_SCRATCH_W1+1
                     txa""")
             if (dt == DataType.WORD)
-                asmgen.out("  jsr  math.divmod_w_asm")
+                asmgen.out("  jsr  prog8_math.divmod_w_asm")
             else
-                asmgen.out("  jsr  math.divmod_uw_asm")
+                asmgen.out("  jsr  prog8_math.divmod_uw_asm")
             asmgen.out("  sta  $name |  sty  $name+1")
         }
 
@@ -2696,7 +2696,7 @@ $shortcutLabel:""")
                 lda  $name+1
                 sta  P8ZP_SCRATCH_W1+1
                 txa
-                jsr  math.divmod_uw_asm
+                jsr  prog8_math.divmod_uw_asm
                 lda  P8ZP_SCRATCH_W2
                 ldy  P8ZP_SCRATCH_W2+1
                 sta  $name

codeGenCpu6502/test/TestCodegen.kt

@@ -26,7 +26,8 @@ class TestCodegen: FunSpec({
             floats = true,
             noSysInit = false,
             compTarget = target,
-            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS
+            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS,
+            memtopAddress = 0xffffu
         )
     }
 
@@ -43,14 +44,46 @@ class TestCodegen: FunSpec({
 //        xx += cx16.r0
 //    }
 //}
-        val codegen = AsmGen6502(prefixSymbols = false)
+        val codegen = AsmGen6502(prefixSymbols = false, 0)
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main",false, SourceCode.Generated("test"), PtBlock.Options(), 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))
-        sub.add(PtVariable("particleDX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
-        sub.add(PtVariable("xx", DataType.WORD, ZeropageWish.DONTCARE, PtNumber(DataType.WORD, 1.0, Position.DUMMY), null, Position.DUMMY))
+        sub.add(PtVariable(
+            "pi",
+            DataType.UBYTE,
+            ZeropageWish.DONTCARE,
+            0u,
+            PtNumber(DataType.UBYTE, 0.0, Position.DUMMY),
+            null,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "particleX",
+            DataType.ARRAY_UB,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            3u,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "particleDX",
+            DataType.ARRAY_UB,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            3u,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "xx",
+            DataType.WORD,
+            ZeropageWish.DONTCARE,
+            0u,
+            PtNumber(DataType.WORD, 1.0, Position.DUMMY),
+            null,
+            Position.DUMMY
+        ))
 
         val assign = PtAugmentedAssign("+=", Position.DUMMY)
         val target = PtAssignTarget(false, Position.DUMMY).also {

codeGenIntermediate/build.gradle

@@ -31,6 +31,7 @@ dependencies {
 
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.1'
     testImplementation 'org.junit.jupiter:junit-jupiter:5.9.2'
+    testImplementation 'io.kotest:kotest-framework-datatest:5.9.1'
     testRuntimeOnly 'org.junit.platform:junit-platform-launcher'
 
 }

codeGenIntermediate/codeGenIntermediate.iml

@@ -14,6 +14,7 @@
     <orderEntry type="module" module-name="intermediate" />
     <orderEntry type="library" name="io.kotest.assertions.core.jvm" level="project" />
     <orderEntry type="library" name="io.kotest.runner.junit5.jvm" level="project" />
+    <orderEntry type="library" name="io.kotest.framework.datatest" level="project" />
     <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
   </component>
 </module>

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

@@ -1,7 +1,7 @@
 package prog8.codegen.intermediate
 
-import prog8.code.StRomSub
-import prog8.code.StRomSubParameter
+import prog8.code.StExtSub
+import prog8.code.StExtSubParameter
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.intermediate.*
@@ -14,7 +14,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             val values = assignment.value as? PtFunctionCall
                 ?: throw AssemblyError("only function calls can return multiple values in a multi-assign")
 
-            val sub = codeGen.symbolTable.lookup(values.name) as? StRomSub
+            val sub = codeGen.symbolTable.lookup(values.name) as? StExtSub
                 ?: throw AssemblyError("only asmsubs can return multiple values")
 
             val result = mutableListOf<IRCodeChunkBase>()
@@ -46,7 +46,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         }
     }
 
-    private fun assignCpuRegister(returns: StRomSubParameter, regNum: Int, target: PtAssignTarget): IRCodeChunks {
+    private fun assignCpuRegister(returns: StExtSubParameter, regNum: Int, target: PtAssignTarget): IRCodeChunks {
         val result = mutableListOf<IRCodeChunkBase>()
         val loadCpuRegInstr = when(returns.register.registerOrPair) {
             RegisterOrPair.A -> IRInstruction(Opcode.LOADHA, IRDataType.BYTE, reg1=regNum)
@@ -57,10 +57,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             RegisterOrPair.XY -> IRInstruction(Opcode.LOADHXY, IRDataType.WORD, reg1=regNum)
             in Cx16VirtualRegisters -> IRInstruction(Opcode.LOADM, IRDataType.WORD, reg1=regNum, labelSymbol = "cx16.${returns.register.registerOrPair.toString().lowercase()}")
             null -> {
-                when(returns.register.statusflag) {
-                    Statusflag.Pc -> IRInstruction(Opcode.LOADHA, IRDataType.BYTE, reg1=regNum)
-                    else -> throw AssemblyError("weird statusflag as returnvalue")
-                }
+                TODO("assign CPU status flag ${returns.register.statusflag!!}")
             }
             else -> throw AssemblyError("cannot load register")
         }
@@ -281,7 +278,6 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 }
             }
             "not" -> {
-                // TODO: in boolean branch, is 'not' handled ok like this?
                 val register = codeGen.registers.nextFree()
                 if(constIndex!=null) {
                     result += IRCodeChunk(null, null).also {
@@ -363,7 +359,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             val variable = targetArray.variable.name
             val itemsize = codeGen.program.memsizer.memorySize(targetArray.type)
 
-            val fixedIndex = constIntValue(targetArray.index)
+            val fixedIndex = targetArray.index.asConstInteger()
             val arrayLength = codeGen.symbolTable.getLength(targetArray.variable.name)
             if(zero) {
                 if(fixedIndex!=null) {
@@ -887,7 +883,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
     }
 
     private fun operatorMultiplyInplaceSplitArray(array: PtArrayIndexer, operand: PtExpression): IRCodeChunks? {
-        return null //    TODO("inplace split word array *")
+        return null  // TODO("inplace split word array *")
     }
 
     private fun operatorMinusInplaceSplitArray(array: PtArrayIndexer, operand: PtExpression): IRCodeChunks? {
@@ -1374,7 +1370,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
 
     private fun createInplaceArrayComparison(array: PtArrayIndexer, value: PtExpression, comparisonOperator: String): IRCodeChunks? {
         if(array.type==DataType.FLOAT)
-            return null  // TODO("optimized in-place compare on float arrays"))   // TODO?
+            return null  // TODO("optimized in-place compare on float arrays"))
 
         val eltSize = codeGen.program.memsizer.memorySize(array.type)
         val result = mutableListOf<IRCodeChunkBase>()
@@ -1393,10 +1389,10 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                     it += when(comparisonOperator) {
                         "==" -> IRInstruction(Opcode.SZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
                         "!=" -> IRInstruction(Opcode.SNZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
-                        "<" -> return null  // TODO("array <0 inplace")) // TODO?
-                        "<=" -> return null  // TODO("array <=0 inplace")) // TODO?
-                        ">" -> return null  // TODO("array >0 inplace")) // TODO?
-                        ">=" -> return null  // TODO("array >=0 inplace")) // TODO?
+                        "<" -> return null  // TODO("array <0 inplace"))
+                        "<=" -> return null  // TODO("array <=0 inplace"))
+                        ">" -> return null  // TODO("array >0 inplace"))
+                        ">=" -> return null  // TODO("array >=0 inplace"))
                         else -> throw AssemblyError("invalid operator")
                     }
                     it += IRInstruction(Opcode.STOREM, vmDt, reg1 = cmpResultReg, labelSymbol = array.variable.name, symbolOffset = constIndex*eltSize)
@@ -1416,10 +1412,10 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                     it += when(comparisonOperator) {
                         "==" -> IRInstruction(Opcode.SZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
                         "!=" -> IRInstruction(Opcode.SNZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
-                        "<" -> return null  // TODO("array <0 inplace")) // TODO?
-                        "<=" -> return null  // TODO("array <=0 inplace")) // TODO?
-                        ">" -> return null  // TODO("array >0 inplace")) // TODO?
-                        ">=" -> return null  // TODO("array >=0 inplace")) // TODO?
+                        "<" -> return null  // TODO("array <0 inplace"))
+                        "<=" -> return null  // TODO("array <=0 inplace"))
+                        ">" -> return null  // TODO("array >0 inplace"))
+                        ">=" -> return null  // TODO("array >=0 inplace"))
                         else -> throw AssemblyError("invalid operator")
                     }
                     it += IRInstruction(Opcode.STOREX, vmDt, reg1=valueReg, reg2=indexTr.resultReg, labelSymbol = array.variable.name)

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

@@ -17,6 +17,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "divmod__uword" -> funcDivmod(call, IRDataType.WORD)
             "rsave", "rrestore" -> ExpressionCodeResult.EMPTY  // vm doesn't have registers to save/restore
             "callfar" -> funcCallfar(call)
+            "callfar2" -> funcCallfar2(call)
             "call" -> funcCall(call)
             "msb" -> funcMsb(call)
             "lsb" -> funcLsb(call)
@@ -38,77 +39,15 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "ror" -> funcRolRor(call)
             "rol2" -> funcRolRor(call)
             "ror2" -> funcRolRor(call)
-            "prog8_ifelse_bittest_set" -> throw AssemblyError("prog8_ifelse_bittest_set() should have been translated as part of an ifElse statement")
-            "prog8_ifelse_bittest_notset" -> throw AssemblyError("prog8_ifelse_bittest_notset() should have been translated as part of an ifElse statement")
+            "prog8_ifelse_bittest_set" -> throw AssemblyError("prog8_ifelse_bittest_set() should have been translated as part of an ifElse/ifExpression statement")
+            "prog8_ifelse_bittest_notset" -> throw AssemblyError("prog8_ifelse_bittest_notset() should have been translated as part of an ifElse/ifExpression statement")
             "prog8_lib_stringcompare" -> funcStringCompare(call)
             "prog8_lib_square_byte" -> funcSquare(call, IRDataType.BYTE)
             "prog8_lib_square_word" -> funcSquare(call, IRDataType.WORD)
-            "prog8_lib_arraycopy" -> funcArrayCopy(call)
             else -> throw AssemblyError("missing builtinfunc for ${call.name}")
         }
     }
 
-    private fun funcArrayCopy(call: PtBuiltinFunctionCall): ExpressionCodeResult {
-        val source = call.args[0] as PtIdentifier
-        val target = call.args[1] as PtIdentifier
-        val sourceLength = codeGen.symbolTable.getLength(source.name)!!
-        val targetLength = codeGen.symbolTable.getLength(target.name)!!
-        require(sourceLength==targetLength)
-        val result = mutableListOf<IRCodeChunkBase>()
-        val fromReg = codeGen.registers.nextFree()
-        val toReg = codeGen.registers.nextFree()
-        val countReg = codeGen.registers.nextFree()
-        if(source.type in SplitWordArrayTypes && target.type in SplitWordArrayTypes) {
-            // split words -> split words, copy lsb and msb arrays separately
-            result += IRCodeChunk(null, null).also {
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromReg, labelSymbol = source.name+"_lsb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toReg, labelSymbol = target.name+"_lsb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=countReg, immediate = sourceLength)
-                it += codeGen.makeSyscall(IMSyscall.MEMCOPY_SMALL, listOf(IRDataType.WORD to fromReg, IRDataType.WORD to toReg, IRDataType.BYTE to (countReg and 255)), returns = null)
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromReg, labelSymbol = source.name+"_msb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toReg, labelSymbol = target.name+"_msb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=countReg, immediate = sourceLength)
-                it += codeGen.makeSyscall(IMSyscall.MEMCOPY_SMALL, listOf(IRDataType.WORD to fromReg, IRDataType.WORD to toReg, IRDataType.BYTE to (countReg and 255)), returns = null)
-            }
-        }
-        else if(source.type in SplitWordArrayTypes) {
-            // split -> normal words
-            require(target.type==DataType.ARRAY_UW || target.type==DataType.ARRAY_W)
-            val fromRegMsb = codeGen.registers.nextFree()
-            result += IRCodeChunk(null, null).also {
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromReg, labelSymbol = source.name+"_lsb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromRegMsb, labelSymbol = source.name+"_msb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toReg, labelSymbol = target.name)
-                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=countReg, immediate = sourceLength)
-            }
-            result += codeGen.makeSyscall(IMSyscall.ARRAYCOPY_SPLITW_TO_NORMAL, listOf(IRDataType.WORD to fromReg, IRDataType.WORD to fromRegMsb, IRDataType.WORD to toReg, IRDataType.BYTE to countReg), returns = null)
-        }
-        else if(target.type in SplitWordArrayTypes) {
-            // normal -> split words
-            require(source.type==DataType.ARRAY_UW || source.type==DataType.ARRAY_W)
-            val toRegMsb = codeGen.registers.nextFree()
-            result += IRCodeChunk(null, null).also {
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromReg, labelSymbol = source.name)
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toReg, labelSymbol = target.name+"_lsb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toRegMsb, labelSymbol = target.name+"_msb")
-                it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=countReg, immediate = sourceLength)
-            }
-            result += codeGen.makeSyscall(IMSyscall.ARRAYCOPY_NORMAL_TO_SPLITW, listOf(IRDataType.WORD to fromReg, IRDataType.WORD to toReg, IRDataType.WORD to toRegMsb, IRDataType.BYTE to countReg), returns = null)
-        }
-        else {
-            // normal array to array copy (various element types)
-            val eltsize = codeGen.options.compTarget.memorySize(ArrayToElementTypes.getValue(source.type))
-            result += IRCodeChunk(null, null).also {
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=fromReg, labelSymbol = source.name)
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=toReg, labelSymbol = target.name)
-                it += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=countReg, immediate = sourceLength * eltsize)
-            }
-            result += codeGen.makeSyscall(IMSyscall.MEMCOPY_SMALL, listOf(IRDataType.WORD to fromReg, IRDataType.WORD to toReg, IRDataType.BYTE to (countReg and 255)), returns = null)
-        }
-
-        return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
-    }
-
     private fun funcSquare(call: PtBuiltinFunctionCall, resultType: IRDataType): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val valueTr = exprGen.translateExpression(call.args[0])
@@ -149,6 +88,29 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         return ExpressionCodeResult(result, IRDataType.WORD, argumentwordTr.resultReg, -1)
     }
 
+    private fun funcCallfar2(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 argumentA = exprGen.translateExpression(call.args[2])
+        val argumentX = exprGen.translateExpression(call.args[3])
+        val argumentY = exprGen.translateExpression(call.args[4])
+        val argumentCarry = exprGen.translateExpression(call.args[5])
+        addToResult(result, bankTr, bankTr.resultReg, -1)
+        addToResult(result, addressTr, addressTr.resultReg, -1)
+        addToResult(result, argumentA, argumentA.resultReg, -1)
+        addToResult(result, argumentX, argumentX.resultReg, -1)
+        addToResult(result, argumentY, argumentY.resultReg, -1)
+        addToResult(result, argumentCarry, argumentCarry.resultReg, -1)
+        result += codeGen.makeSyscall(IMSyscall.CALLFAR2, listOf(IRDataType.BYTE to bankTr.resultReg, IRDataType.WORD to addressTr.resultReg,
+            IRDataType.BYTE to argumentA.resultReg,
+            IRDataType.BYTE to argumentX.resultReg,
+            IRDataType.BYTE to argumentY.resultReg,
+            IRDataType.BYTE to argumentCarry.resultReg), IRDataType.WORD to addressTr.resultReg)
+        return ExpressionCodeResult(result, IRDataType.WORD, addressTr.resultReg, -1)
+    }
+
     private fun funcDivmod(call: PtBuiltinFunctionCall, type: IRDataType): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val number = call.args[0]
@@ -580,6 +542,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val arr = (arg as? PtArrayIndexer)
         val index = arr?.index?.asConstInteger()
         if(arr!=null && index!=null) {
+            if(arr.splitWords)  TODO("IR rol/ror on split words array")
             val variable = arr.variable.name
             val itemsize = codeGen.program.memsizer.memorySize(arr.type)
             addInstr(result, IRInstruction(opcodeMemAndReg.first, vmDt, labelSymbol = variable, symbolOffset = index*itemsize), null)

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

@@ -1,7 +1,7 @@
 package prog8.codegen.intermediate
 
 import prog8.code.StNode
-import prog8.code.StRomSub
+import prog8.code.StExtSub
 import prog8.code.StSub
 import prog8.code.ast.*
 import prog8.code.core.*
@@ -79,6 +79,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             is PtPrefix -> translate(expr)
             is PtArrayIndexer -> translate(expr)
             is PtBinaryExpression -> translate(expr)
+            is PtIfExpression -> translate(expr)
             is PtBuiltinFunctionCall -> codeGen.translateBuiltinFunc(expr)
             is PtFunctionCall -> translate(expr)
             is PtContainmentCheck -> translate(expr)
@@ -89,6 +90,41 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         }
     }
 
+    private fun translate(ifExpr: PtIfExpression): ExpressionCodeResult {
+
+        if((ifExpr.condition as? PtPrefix)?.operator=="not")
+            throw AssemblyError("not prefix in ifexpression should have been replaced by swapped values")
+
+        // TODO don't store condition as expression result but just use the flags, like a normal PtIfElse translation does
+        val condTr = translateExpression(ifExpr.condition)
+        val trueTr = translateExpression(ifExpr.truevalue)
+        val falseTr = translateExpression(ifExpr.falsevalue)
+        val irDt = irType(ifExpr.type)
+        val result = mutableListOf<IRCodeChunkBase>()
+        val falseLabel = codeGen.createLabelName()
+        val endLabel = codeGen.createLabelName()
+
+        addToResult(result, condTr, condTr.resultReg, -1)
+        addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=condTr.resultReg, immediate = 0), null)
+        addInstr(result, IRInstruction(Opcode.BSTEQ, labelSymbol = falseLabel), null)
+
+        if (irDt != IRDataType.FLOAT) {
+            addToResult(result, trueTr, trueTr.resultReg, -1)
+            addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
+            result += IRCodeChunk(falseLabel, null)
+            addToResult(result, falseTr, trueTr.resultReg, -1)
+            result += IRCodeChunk(endLabel, null)
+            return ExpressionCodeResult(result, irDt, trueTr.resultReg, -1)
+        } else {
+            addToResult(result, trueTr, -1, trueTr.resultFpReg)
+            addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = endLabel), null)
+            result += IRCodeChunk(falseLabel, null)
+            addToResult(result, falseTr, -1, trueTr.resultFpReg)
+            result += IRCodeChunk(endLabel, null)
+            return ExpressionCodeResult(result, irDt, -1, trueTr.resultFpReg)
+        }
+    }
+
     private fun translate(expr: PtAddressOf): ExpressionCodeResult {
         val vmDt = irType(expr.type)
         val symbol = expr.identifier.name
@@ -168,55 +204,93 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
     }
 
     private fun translate(check: PtContainmentCheck): ExpressionCodeResult {
+        val elementDt = check.needle.type
         val result = mutableListOf<IRCodeChunkBase>()
-        when(check.iterable.type) {
+
+        if(check.haystackValues!=null) {
+            val haystack = check.haystackValues!!.children.map {
+                if(it is PtBool) it.asInt()
+                else (it as PtNumber).number.toInt()
+            }
+            when(elementDt) {
+                in IntegerDatatypesWithBoolean -> {
+                    if (elementDt in ByteDatatypesWithBoolean) require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_BYTE)
+                    if (elementDt in WordDatatypes) require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_WORD)
+                    val gottemLabel = codeGen.createLabelName()
+                    val endLabel = codeGen.createLabelName()
+                    val elementTr = translateExpression(check.needle)
+                    addToResult(result, elementTr, elementTr.resultReg, -1)
+                    result += IRCodeChunk(null, null).also {
+                        for(value in haystack){
+                            it += IRInstruction(Opcode.CMPI, irType(elementDt), elementTr.resultReg, immediate = value)
+                            it += IRInstruction(Opcode.BSTEQ, labelSymbol = gottemLabel)
+                        }
+                        it += IRInstruction(Opcode.LOAD, IRDataType.BYTE, elementTr.resultReg, immediate = 0)
+                        it += IRInstruction(Opcode.JUMP, labelSymbol = endLabel)
+                    }
+                    addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, elementTr.resultReg, immediate = 1), gottemLabel)
+                    result += IRCodeChunk(endLabel, null)
+                    return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
+
+                }
+                DataType.FLOAT -> throw AssemblyError("containmentchecks for floats should always be done on an array variable with subroutine")
+                else -> throw AssemblyError("weird dt $elementDt")
+            }
+        }
+
+        val haystackVar = check.haystackHeapVar!!
+        when(haystackVar.type) {
             DataType.STR -> {
                 addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-                val elementTr = translateExpression(check.element)
+                val elementTr = translateExpression(check.needle)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
-                val iterableTr = translateExpression(check.iterable)
+                val iterableTr = translateExpression(haystackVar)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 result += codeGen.makeSyscall(IMSyscall.STRING_CONTAINS, listOf(IRDataType.BYTE to elementTr.resultReg, IRDataType.WORD to iterableTr.resultReg), IRDataType.BYTE to elementTr.resultReg)
+                addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=elementTr.resultReg, immediate = 0), null)
                 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)
+                val elementTr = translateExpression(check.needle)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
-                val iterableTr = translateExpression(check.iterable)
+                val iterableTr = translateExpression(haystackVar)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 val lengthReg = codeGen.registers.nextFree()
-                val iterableLength = codeGen.symbolTable.getLength(check.iterable.name)
+                val iterableLength = codeGen.symbolTable.getLength(haystackVar.name)
                 addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterableLength!!), null)
                 result += codeGen.makeSyscall(IMSyscall.BYTEARRAY_CONTAINS, listOf(IRDataType.BYTE to elementTr.resultReg, IRDataType.WORD to iterableTr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to elementTr.resultReg)
+                addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=elementTr.resultReg, immediate = 0), null)
                 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)
+                val elementTr = translateExpression(check.needle)
                 addToResult(result, elementTr, elementTr.resultReg, -1)
-                val iterableTr = translateExpression(check.iterable)
+                val iterableTr = translateExpression(haystackVar)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 val lengthReg = codeGen.registers.nextFree()
-                val iterableLength = codeGen.symbolTable.getLength(check.iterable.name)
+                val iterableLength = codeGen.symbolTable.getLength(haystackVar.name)
                 addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterableLength!!), null)
                 result += codeGen.makeSyscall(IMSyscall.WORDARRAY_CONTAINS, listOf(IRDataType.WORD to elementTr.resultReg, IRDataType.WORD to iterableTr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to elementTr.resultReg)
+                addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=elementTr.resultReg, immediate = 0), null)
                 return ExpressionCodeResult(result, IRDataType.BYTE, elementTr.resultReg, -1)
             }
             DataType.ARRAY_F -> {
                 addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 3), null)
-                val elementTr = translateExpression(check.element)
+                val elementTr = translateExpression(check.needle)
                 addToResult(result, elementTr, -1, elementTr.resultFpReg)
-                val iterableTr = translateExpression(check.iterable)
+                val iterableTr = translateExpression(haystackVar)
                 addToResult(result, iterableTr, iterableTr.resultReg, -1)
                 val lengthReg = codeGen.registers.nextFree()
                 val resultReg = codeGen.registers.nextFree()
-                val iterableLength = codeGen.symbolTable.getLength(check.iterable.name)
+                val iterableLength = codeGen.symbolTable.getLength(haystackVar.name)
                 addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1=lengthReg, immediate = iterableLength!!), null)
                 result += codeGen.makeSyscall(IMSyscall.FLOATARRAY_CONTAINS, listOf(IRDataType.FLOAT to elementTr.resultFpReg, IRDataType.WORD to iterableTr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to resultReg)
+                addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1=resultReg, immediate = 0), null)
                 return ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
             }
-            else -> throw AssemblyError("weird iterable dt ${check.iterable.type} for ${check.iterable.name}")
+            else -> throw AssemblyError("weird iterable dt ${haystackVar.type} for ${haystackVar.name}")
         }
     }
 
@@ -454,7 +528,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
     fun translate(fcall: PtFunctionCall): ExpressionCodeResult {
         val callTarget = codeGen.symbolTable.flat.getValue(fcall.name)
 
-        if(callTarget.scopedName in listOf("sys.push", "sys.pushw", "sys.pop", "sys.popw")) {
+        if(callTarget.scopedName in listOf("sys.push", "sys.pushw", "sys.pop", "sys.popw", "floats.push", "floats.pop")) {
             // special case, these should be inlined, or even use specialized instructions. Instead of doing a normal subroutine call.
             return translateStackFunctions(fcall, callTarget)
         }
@@ -514,7 +588,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 else
                     ExpressionCodeResult(result, returnRegSpec!!.dt, returnRegSpec.registerNum, -1)
             }
-            is StRomSub -> {
+            is StExtSub -> {
                 val result = mutableListOf<IRCodeChunkBase>()
                 addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = callTarget.parameters.size), null)
                 // assign the arguments
@@ -534,6 +608,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                         RegisterOrPair.AX -> addInstr(result, IRInstruction(Opcode.STOREHAX, IRDataType.WORD, reg1=tr.resultReg), null)
                         RegisterOrPair.AY -> addInstr(result, IRInstruction(Opcode.STOREHAY, IRDataType.WORD, reg1=tr.resultReg), null)
                         RegisterOrPair.XY -> addInstr(result, IRInstruction(Opcode.STOREHXY, IRDataType.WORD, reg1=tr.resultReg), null)
+                        RegisterOrPair.FAC1, RegisterOrPair.FAC2 -> TODO("floating point register parameters not supported")
                         in Cx16VirtualRegisters -> {
                             addInstr(result, IRInstruction(Opcode.STOREM, paramDt, reg1=tr.resultReg, labelSymbol = "cx16.${parameter.register.registerOrPair.toString().lowercase()}"), null)
                         }
@@ -547,7 +622,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 }
 
                 if(callTarget.returns.size>1)
-                    return callRomSubWithMultipleReturnValues(callTarget, fcall, argRegisters, result)
+                    return callExtSubWithMultipleReturnValues(callTarget, fcall, argRegisters, result)
 
                 // return a single value (or nothing)
                 val returnRegSpec = if(fcall.void) null else {
@@ -569,8 +644,18 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 val call =
                     if(callTarget.address==null)
                         IRInstruction(Opcode.CALL, labelSymbol = fcall.name, fcallArgs = FunctionCallArgs(argRegisters, returnRegs))
-                    else
-                        IRInstruction(Opcode.CALL, address = callTarget.address!!.toInt(), fcallArgs = FunctionCallArgs(argRegisters, returnRegs))
+                    else {
+                        val address = callTarget.address!!
+                        if(address.constbank==null && address.varbank==null) {
+                            IRInstruction(
+                                Opcode.CALL,
+                                address = address.address.toInt(),
+                                fcallArgs = FunctionCallArgs(argRegisters, returnRegs))
+                        }
+                        else {
+                            TODO("callfar into another bank is not implemented for the selected compilation target")
+                        }
+                    }
                 addInstr(result, call, null)
                 var finalReturnRegister = returnRegSpec?.registerNum ?: -1
 
@@ -649,12 +734,25 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 addInstr(chunk, IRInstruction(Opcode.POP, IRDataType.WORD, reg1=popReg), null)
                 return ExpressionCodeResult(chunk, IRDataType.WORD, popReg, -1)
             }
+            "floats.push" -> {
+                // push float
+                val tr = translateExpression(fcall.args.single())
+                chunk += tr.chunks
+                addInstr(chunk, IRInstruction(Opcode.PUSH, IRDataType.FLOAT, fpReg1 = tr.resultFpReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.FLOAT, -1, -1)
+            }
+            "floats.pop" -> {
+                // pop float
+                val popReg = codeGen.registers.nextFreeFloat()
+                addInstr(chunk, IRInstruction(Opcode.POP, IRDataType.FLOAT, fpReg1 = popReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.FLOAT, -1, resultFpReg = popReg)
+            }
             else -> throw AssemblyError("unknown stack subroutine called")
         }
     }
 
-    private fun callRomSubWithMultipleReturnValues(
-        callTarget: StRomSub,
+    private fun callExtSubWithMultipleReturnValues(
+        callTarget: StExtSub,
         fcall: PtFunctionCall,
         argRegisters: MutableList<FunctionCallArgs.ArgumentSpec>,
         result: MutableList<IRCodeChunkBase>
@@ -668,8 +766,17 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         val call =
             if(callTarget.address==null)
                 IRInstruction(Opcode.CALL, labelSymbol = fcall.name, fcallArgs = FunctionCallArgs(argRegisters, returnRegisters))
-            else
-                IRInstruction(Opcode.CALL, address = callTarget.address!!.toInt(), fcallArgs = FunctionCallArgs(argRegisters, returnRegisters))
+            else {
+                val address = callTarget.address!!
+                if(address.constbank==null && address.varbank==null) {
+                    IRInstruction(
+                        Opcode.CALL,
+                        address = address.address.toInt(),
+                        fcallArgs = FunctionCallArgs(argRegisters, returnRegisters)
+                    )
+                }
+                else TODO("extsub with banked address got called ${callTarget.name}")
+            }
         addInstr(result, call, null)
         val resultRegs = returnRegisters.filter{it.dt!=IRDataType.FLOAT}.map{it.registerNum}
         val resultFpRegs = returnRegisters.filter{it.dt==IRDataType.FLOAT}.map{it.registerNum}
@@ -788,7 +895,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
             if(binExpr.left.type==DataType.STR || binExpr.right.type==DataType.STR) {
                 throw AssemblyError("str compares should have been replaced with builtin function call to do the compare")
             } else {
-                return if(constValue(binExpr.right)==0.0) {
+                return if(binExpr.right.asConstValue()==0.0) {
                     val tr = translateExpression(binExpr.left)
                     addToResult(result, tr, tr.resultReg, -1)
                     val opcode = if (notEquals) Opcode.SNZ else Opcode.SZ

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

@@ -23,6 +23,7 @@ class IRCodeGen(
         makeAllNodenamesScoped(program)
         moveAllNestedSubroutinesToBlockScope(program)
         verifyNameScoping(program, symbolTable)
+        changeGlobalVarInits(symbolTable)
 
         val irSymbolTable = IRSymbolTable.fromStDuringCodegen(symbolTable)
         val irProg = IRProgram(program.name, irSymbolTable, options, program.encoding)
@@ -54,12 +55,43 @@ class IRCodeGen(
         return irProg
     }
 
+    private fun changeGlobalVarInits(symbolTable: SymbolTable) {
+        // Normally, block level (global) variables that have a numeric initialization value
+        // are initialized via an assignment statement.
+        val initsToRemove = mutableListOf<Pair<PtBlock, PtAssignment>>()
+
+        symbolTable.allVariables.forEach { variable ->
+            if(variable.uninitialized && variable.parent.type==StNodeType.BLOCK) {
+                val block = variable.parent.astNode as PtBlock
+                val initialization = (block.children.firstOrNull {
+                    it is PtAssignment && it.target.identifier?.name==variable.scopedName
+                } as PtAssignment?)
+                val initValue = initialization?.value
+                when(initValue){
+                    is PtBool -> {
+                        variable.setOnetimeInitNumeric(initValue.asInt().toDouble())
+                        initsToRemove += block to initialization
+                    }
+                    is PtNumber -> {
+                        variable.setOnetimeInitNumeric(initValue.number)
+                        initsToRemove += block to initialization
+                    }
+                    is PtArray, is PtString -> throw AssemblyError("array or string initialization values should already be part of the vardecl, not a separate assignment")
+                    else -> {}
+                }
+            }
+        }
+
+        for((block, assign) in initsToRemove) {
+            block.children.remove(assign)
+        }
+    }
+
     private fun verifyNameScoping(program: PtProgram, symbolTable: SymbolTable) {
         fun verifyPtNode(node: PtNode) {
             when (node) {
                 is PtBuiltinFunctionCall -> require('.' !in node.name) { "builtin function call name should not be scoped: ${node.name}" }
                 is PtFunctionCall -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
-                is PtIdentifier -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
                 is PtAsmSub -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
                 is PtBlock -> require('.' !in node.name) { "block name should not be scoped: ${node.name}" }
                 is PtConstant -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
@@ -69,8 +101,7 @@ class IRCodeGen(
                 is PtVariable -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
                 is PtProgram -> require('.' !in node.name) { "program name should not be scoped: ${node.name}" }
                 is PtSubroutineParameter -> require('.' in node.name) { "node $node name is not scoped: ${node.name}" }
-                else -> { /* node has no name */
-                }
+                else -> { /* node has no name or is ok to have no dots in the name */ }
             }
             node.children.forEach { verifyPtNode(it) }
         }
@@ -201,6 +232,7 @@ class IRCodeGen(
                 chunk += IRInstruction(Opcode.BREAKPOINT)
                 listOf(chunk)
             }
+            is PtAlign -> TODO("ir support for inline %align")
             is PtConditionalBranch -> translate(node)
             is PtInlineAssembly -> listOf(IRInlineAsmChunk(null, node.assembly, node.isIR, null))
             is PtIncludeBinary -> listOf(IRInlineBinaryChunk(null, readBinaryData(node), null))
@@ -221,6 +253,7 @@ class IRCodeGen(
             is PtBool,
             is PtArray,
             is PtBlock,
+            is PtDefer -> throw AssemblyError("should have been transformed")
             is PtString -> throw AssemblyError("should not occur as separate statement node ${node.position}")
             is PtSub -> throw AssemblyError("nested subroutines should have been flattened ${node.position}")
             else -> TODO("missing codegen for $node")
@@ -487,49 +520,76 @@ class IRCodeGen(
         val loopLabel = createLabelName()
         val result = mutableListOf<IRCodeChunkBase>()
 
-        val toTr = expressionEval.translateExpression(iterable.to)
-        addToResult(result, toTr, toTr.resultReg, -1)
-        val fromTr = expressionEval.translateExpression(iterable.from)
-        addToResult(result, fromTr, fromTr.resultReg, -1)
-
-        val labelAfterFor = createLabelName()
-        val precheckInstruction = if(loopvarDt in SignedDatatypes) {
-            if(step>0)
-                IRInstruction(Opcode.BGTSR, loopvarDtIr, fromTr.resultReg, toTr.resultReg, labelSymbol=labelAfterFor)
-            else
-                IRInstruction(Opcode.BGTSR, loopvarDtIr, toTr.resultReg, fromTr.resultReg, labelSymbol=labelAfterFor)
-        } else {
-            if(step>0)
-                IRInstruction(Opcode.BGTR, loopvarDtIr, fromTr.resultReg, toTr.resultReg, labelSymbol=labelAfterFor)
-            else
-                IRInstruction(Opcode.BGTR, loopvarDtIr, toTr.resultReg, fromTr.resultReg, labelSymbol=labelAfterFor)
-        }
-        addInstr(result, precheckInstruction, null)
-
-        addInstr(result, IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol=loopvarSymbol), null)
-        result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
-        if(step==1 || step==-1) {
-            // if endvalue == loopvar, stop loop, else iterate
+        if(loopvarDtIr==IRDataType.BYTE && step==-1 && iterable.to.asConstInteger()==0) {
+            // downto 0 optimization (byte)
+            val fromTr = expressionEval.translateExpression(iterable.from)
+            addToResult(result, fromTr, fromTr.resultReg, -1)
+            addInstr(result, IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol=loopvarSymbol), null)
+            result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
+            result += addConstMem(loopvarDtIr, null, loopvarSymbol, -1)
             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)
+                it += IRInstruction(Opcode.CMPI, loopvarDtIr, reg1 = fromTr.resultReg, immediate = 255)
+                it += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
             }
-            result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
-            addInstr(result, IRInstruction(Opcode.JUMP, labelSymbol = loopLabel), null)
-        } else {
-            // ind/dec index, then:
-            // ascending: if endvalue >= loopvar, iterate
-            // descending: if loopvar >= endvalue, iterate
-            result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
-            addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol = loopvarSymbol), null)
-            if(step > 0)
-                addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = toTr.resultReg, fromTr.resultReg), null)
-            else
-                addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = fromTr.resultReg, toTr.resultReg), null)
-            addInstr(result, IRInstruction(Opcode.BSTPOS, labelSymbol = loopLabel), null)
         }
-        result += IRCodeChunk(labelAfterFor, null)
+        else if(step==-1 && iterable.to.asConstInteger()==1) {
+            // downto 1 optimization (byte and word)
+            val fromTr = expressionEval.translateExpression(iterable.from)
+            addToResult(result, fromTr, fromTr.resultReg, -1)
+            addInstr(result, IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol=loopvarSymbol), null)
+            result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
+            result += addConstMem(loopvarDtIr, null, loopvarSymbol, -1)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = fromTr.resultReg, labelSymbol = loopvarSymbol)
+                it += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+            }
+        }
+        else {
+            val toTr = expressionEval.translateExpression(iterable.to)
+            addToResult(result, toTr, toTr.resultReg, -1)
+            val fromTr = expressionEval.translateExpression(iterable.from)
+            addToResult(result, fromTr, fromTr.resultReg, -1)
+
+            val labelAfterFor = createLabelName()
+            val precheckInstruction = if(loopvarDt in SignedDatatypes) {
+                if(step>0)
+                    IRInstruction(Opcode.BGTSR, loopvarDtIr, fromTr.resultReg, toTr.resultReg, labelSymbol=labelAfterFor)
+                else
+                    IRInstruction(Opcode.BGTSR, loopvarDtIr, toTr.resultReg, fromTr.resultReg, labelSymbol=labelAfterFor)
+            } else {
+                if(step>0)
+                    IRInstruction(Opcode.BGTR, loopvarDtIr, fromTr.resultReg, toTr.resultReg, labelSymbol=labelAfterFor)
+                else
+                    IRInstruction(Opcode.BGTR, loopvarDtIr, toTr.resultReg, fromTr.resultReg, labelSymbol=labelAfterFor)
+            }
+            addInstr(result, precheckInstruction, null)
+
+            addInstr(result, IRInstruction(Opcode.STOREM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol=loopvarSymbol), null)
+            result += labelFirstChunk(translateNode(forLoop.statements), loopLabel)
+            if(step==1 || step==-1) {
+                // if endvalue == loopvar, stop loop, else iterate
+                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 {
+                // ind/dec index, then:
+                // ascending: if endvalue >= loopvar, iterate
+                // descending: if loopvar >= endvalue, iterate
+                result += addConstMem(loopvarDtIr, null, loopvarSymbol, step)
+                addInstr(result, IRInstruction(Opcode.LOADM, loopvarDtIr, reg1=fromTr.resultReg, labelSymbol = loopvarSymbol), null)
+                if(step > 0)
+                    addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = toTr.resultReg, fromTr.resultReg), null)
+                else
+                    addInstr(result, IRInstruction(Opcode.CMP, loopvarDtIr, reg1 = fromTr.resultReg, toTr.resultReg), null)
+                addInstr(result, IRInstruction(Opcode.BSTPOS, labelSymbol = loopLabel), null)
+            }
+            result += IRCodeChunk(labelAfterFor, null)
+        }
         return result
     }
 
@@ -558,9 +618,25 @@ class IRCodeGen(
         result += chunk
         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.CMPI, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped)
-        chunk2 += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+        if(loopvarDtIr==IRDataType.BYTE && iterable.step==-1 && iterable.last==0) {
+            // downto 0 optimization (byte)
+            if(loopvarDt==DataType.BYTE || iterable.first<=127) {
+                chunk2 += IRInstruction(Opcode.BSTPOS, labelSymbol = loopLabel)
+            } else {
+                chunk2 += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = indexReg, labelSymbol = loopvarSymbol)
+                chunk2 += IRInstruction(Opcode.CMPI, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped)
+                chunk2 += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+            }
+        }
+        else if(iterable.step==-1 && iterable.last==1) {
+            // downto 1 optimization (byte and word)
+            chunk2 += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+        } else {
+            // downto some other value
+            chunk2 += IRInstruction(Opcode.LOADM, loopvarDtIr, reg1 = indexReg, labelSymbol = loopvarSymbol)
+            chunk2 += IRInstruction(Opcode.CMPI, loopvarDtIr, reg1 = indexReg, immediate = rangeEndExclusiveWrapped)
+            chunk2 += IRInstruction(Opcode.BSTNE, labelSymbol = loopLabel)
+        }
         result += chunk2
         return result
     }
@@ -1252,6 +1328,9 @@ class IRCodeGen(
     }
 
     private fun translateIfElse(ifElse: PtIfElse): IRCodeChunks {
+        if((ifElse.condition as? PtPrefix)?.operator=="not" && ifElse.hasElse())
+            throw AssemblyError("not prefix in ifelse should have been replaced by swapped if-else blocks")
+
         val condition = ifElse.condition as? PtBinaryExpression
         if(condition==null || condition.left.type != DataType.FLOAT) {
             return ifWithElse_IntegerCond(ifElse)
@@ -1317,12 +1396,14 @@ class IRCodeGen(
     private fun ifWithElse_IntegerCond(ifElse: PtIfElse): List<IRCodeChunkBase> {
         val result = mutableListOf<IRCodeChunkBase>()
 
-        fun translateSimple(condition: PtExpression, jumpFalseOpcode: Opcode) {
+        fun translateSimple(condition: PtExpression, jumpFalseOpcode: Opcode, addCmpiZero: Boolean) {
 
             if(condition is PtBuiltinFunctionCall && condition.name.startsWith("prog8_ifelse_bittest_"))
                 throw AssemblyError("IR codegen doesn't have special instructions for dedicated BIT tests and should just still use normal AND")
 
             val tr = expressionEval.translateExpression(condition)
+            if(addCmpiZero)
+                tr.chunks.last().instructions.add(IRInstruction(Opcode.CMPI, tr.dt, reg1 = tr.resultReg, immediate = 0))
             result += tr.chunks
             if(ifElse.hasElse()) {
                 val elseLabel = createLabelName()
@@ -1342,7 +1423,7 @@ class IRCodeGen(
 
         fun translateBinExpr(condition: PtBinaryExpression) {
             if(condition.operator in LogicalOperators)
-                return translateSimple(condition, Opcode.BSTEQ)
+                return translateSimple(condition, Opcode.BSTEQ, false)
 
             val signed = condition.left.type in SignedDatatypes
             val elseBranchFirstReg: Int
@@ -1481,18 +1562,21 @@ class IRCodeGen(
         when(val cond=ifElse.condition) {
             is PtBool -> {
                 // normally this will be optimized away, but not with -noopt
-                translateSimple(cond, Opcode.BSTEQ)
+                translateSimple(cond, Opcode.BSTEQ, false)
             }
             is PtTypeCast -> {
                 require(cond.type==DataType.BOOL && cond.value.type in NumericDatatypes)
-                translateSimple(cond, Opcode.BSTEQ)
+                translateSimple(cond, Opcode.BSTEQ, false)
             }
-            is PtIdentifier, is PtArrayIndexer, is PtBuiltinFunctionCall, is PtFunctionCall, is PtContainmentCheck -> {
-                translateSimple(cond, Opcode.BSTEQ)
+            is PtIdentifier, is PtArrayIndexer, is PtContainmentCheck -> {
+                translateSimple(cond, Opcode.BSTEQ, false)
+            }
+            is PtBuiltinFunctionCall, is PtFunctionCall -> {
+                translateSimple(cond, Opcode.BSTEQ, true)
             }
             is PtPrefix -> {
                 require(cond.operator=="not")
-                translateSimple(cond.value, Opcode.BSTNE)
+                translateSimple(cond.value, Opcode.BSTNE, false)
             }
             is PtBinaryExpression -> {
                 translateBinExpr(cond)
@@ -1503,7 +1587,7 @@ class IRCodeGen(
     }
 
     private fun translate(repeat: PtRepeatLoop): IRCodeChunks {
-        when (constIntValue(repeat.count)) {
+        when (repeat.count.asConstInteger()) {
             0 -> return emptyList()
             1 -> return translateGroup(repeat.children)
             256 -> {
@@ -1514,18 +1598,30 @@ class IRCodeGen(
 
         val repeatLabel = createLabelName()
         val skipRepeatLabel = createLabelName()
-        val irDt = irType(repeat.count.type)
+        val constRepeats = repeat.count.asConstInteger()
         val result = mutableListOf<IRCodeChunkBase>()
-        val countTr = expressionEval.translateExpression(repeat.count)
-        addToResult(result, countTr, countTr.resultReg, -1)
-        if(constIntValue(repeat.count)==null) {
-            // check if the counter is already zero
-            addInstr(result, IRInstruction(Opcode.BSTEQ, labelSymbol = skipRepeatLabel), null)
-        }
-        result += labelFirstChunk(translateNode(repeat.statements), repeatLabel)
-        result += IRCodeChunk(null, null).also {
-            it += IRInstruction(Opcode.DEC, irDt, reg1 = countTr.resultReg)  // sets status bits
-            it += IRInstruction(Opcode.BSTNE, labelSymbol = repeatLabel)
+        if(constRepeats==65536) {
+            // make use of the word wrap around to count to 65536
+            val resultRegister = registers.nextFree()
+            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=resultRegister, immediate = 0), null)
+            result += labelFirstChunk(translateNode(repeat.statements), repeatLabel)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.DEC, IRDataType.WORD, reg1 = resultRegister)  // sets status bits
+                it += IRInstruction(Opcode.BSTNE, labelSymbol = repeatLabel)
+            }
+        } else {
+            val irDt = irType(repeat.count.type)
+            val countTr = expressionEval.translateExpression(repeat.count)
+            addToResult(result, countTr, countTr.resultReg, -1)
+            if (repeat.count.asConstValue() == null) {
+                // check if the counter is already zero
+                addInstr(result, IRInstruction(Opcode.BSTEQ, labelSymbol = skipRepeatLabel), null)
+            }
+            result += labelFirstChunk(translateNode(repeat.statements), repeatLabel)
+            result += IRCodeChunk(null, null).also {
+                it += IRInstruction(Opcode.DEC, irDt, reg1 = countTr.resultReg)  // sets status bits
+                it += IRInstruction(Opcode.BSTNE, labelSymbol = repeatLabel)
+            }
         }
         result += IRCodeChunk(skipRepeatLabel, null)
         return result
@@ -1570,14 +1666,22 @@ class IRCodeGen(
             addInstr(result, IRInstruction(Opcode.RETURN), null)
         } else {
             if(value.type==DataType.FLOAT) {
-                val tr = expressionEval.translateExpression(value)
-                addToResult(result, tr, -1, tr.resultFpReg)
-                addInstr(result, IRInstruction(Opcode.RETURNR, IRDataType.FLOAT, fpReg1 = tr.resultFpReg), null)
+                if(value is PtNumber) {
+                    addInstr(result, IRInstruction(Opcode.RETURNI, IRDataType.FLOAT, immediateFp = value.number), null)
+                } else {
+                    val tr = expressionEval.translateExpression(value)
+                    addToResult(result, tr, -1, tr.resultFpReg)
+                    addInstr(result, IRInstruction(Opcode.RETURNR, IRDataType.FLOAT, fpReg1 = tr.resultFpReg), null)
+                }
             }
             else {
-                val tr = expressionEval.translateExpression(value)
-                addToResult(result, tr, tr.resultReg, -1)
-                addInstr(result, IRInstruction(Opcode.RETURNR, irType(value.type) , reg1=tr.resultReg), null)
+                if(value.asConstInteger()!=null) {
+                    addInstr(result, IRInstruction(Opcode.RETURNI, irType(value.type), immediate = value.asConstInteger()), null)
+                } else {
+                    val tr = expressionEval.translateExpression(value)
+                    addToResult(result, tr, tr.resultReg, -1)
+                    addInstr(result, IRInstruction(Opcode.RETURNR, irType(value.type), reg1 = tr.resultReg), null)
+                }
             }
         }
         return result
@@ -1590,14 +1694,14 @@ class IRCodeGen(
                 block.options.forceOutput,
                 block.options.noSymbolPrefixing,
                 block.options.veraFxMuls,
-                block.options.ignoreUnused,
-                translate(block.options.alignment)
+                block.options.ignoreUnused
             ), block.position)
         for (child in block.children) {
             when(child) {
                 is PtNop -> { /* nothing */ }
                 is PtAssignment, is PtAugmentedAssign -> { /* global variable initialization is done elsewhere */ }
                 is PtVariable, is PtConstant, is PtMemMapped -> { /* vars should be looked up via symbol table */ }
+                is PtAlign -> TODO("ir support for inline %align")
                 is PtSub -> {
                     val sub = IRSubroutine(child.name, translate(child.parameters), child.returntype, child.position)
                     for (subchild in child.children) {
@@ -1607,9 +1711,9 @@ class IRCodeGen(
                 }
                 is PtAsmSub -> {
                     if(child.address!=null) {
-                        // romsub. No codegen needed: calls to this are jumping straight to the address.
+                        // extmsub. No codegen needed: calls to this are jumping straight to the address.
                         require(child.children.isEmpty()) {
-                            "romsub should be empty at ${child.position}"
+                            "extsub should be empty at ${child.position}"
                         }
                     } else {
                         // regular asmsub
@@ -1622,7 +1726,7 @@ class IRCodeGen(
                         )
                         irBlock += IRAsmSubroutine(
                             child.name,
-                            child.address,
+                            null,
                             child.clobbers,
                             child.parameters.map { IRAsmSubroutine.IRAsmParam(it.first, it.second.type) },        // note: the name of the asmsub param is not used here anymore
                             child.returns.map { IRAsmSubroutine.IRAsmParam(it.first, it.second)},
@@ -1648,25 +1752,17 @@ class IRCodeGen(
 
     private fun translate(parameters: List<PtSubroutineParameter>) =
         parameters.map {
-            val flattenedName = it.definingSub()!!.name + "." + it.name
+            val flattenedName = it.definingISub()!!.name + "." + it.name
             if(symbolTable.lookup(flattenedName)==null)
                 TODO("fix missing lookup for: $flattenedName   parameter")
             val orig = symbolTable.lookup(flattenedName) as StStaticVariable
             IRSubroutine.IRParam(flattenedName, orig.dt)
         }
 
-    private fun translate(alignment: PtBlock.BlockAlignment): IRBlock.BlockAlignment {
-        return when(alignment) {
-            PtBlock.BlockAlignment.NONE -> IRBlock.BlockAlignment.NONE
-            PtBlock.BlockAlignment.WORD -> IRBlock.BlockAlignment.WORD
-            PtBlock.BlockAlignment.PAGE -> IRBlock.BlockAlignment.PAGE
-        }
-    }
-
     private var labelSequenceNumber = 0
     internal fun createLabelName(): String {
         labelSequenceNumber++
-        return "label_gen_$labelSequenceNumber"
+        return "${PtLabel.GeneratedLabelPrefix}$labelSequenceNumber"
     }
 
     internal fun translateBuiltinFunc(call: PtBuiltinFunctionCall): ExpressionCodeResult

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

@@ -64,7 +64,7 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
     private fun replaceConcatZeroMsbWithExt(chunk: IRCodeChunk, indexedInstructions: List<IndexedValue<IRInstruction>>): Boolean {
         var changed = false
         indexedInstructions.reversed().forEach { (idx, ins) ->
-            if (ins.opcode == Opcode.CONCAT) {
+            if (ins.opcode == Opcode.CONCAT && idx>0) {
                 // 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) {
@@ -299,7 +299,7 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
             }
 
             // remove useless RETURN
-            if(idx>0 && (ins.opcode == Opcode.RETURN || ins.opcode==Opcode.RETURNR)) {
+            if(idx>0 && (ins.opcode == Opcode.RETURN || ins.opcode==Opcode.RETURNR || ins.opcode==Opcode.RETURNI)) {
                 val previous = chunk.instructions[idx-1]
                 if(previous.opcode in OpcodesThatJump) {
                     chunk.instructions.removeAt(idx)
@@ -449,7 +449,7 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
     private fun removeDoubleLoadsAndStores(chunk: IRCodeChunk, indexedInstructions: List<IndexedValue<IRInstruction>>): Boolean {
         var changed = false
         indexedInstructions.forEach { (idx, ins) ->
-            if(ins.opcode==Opcode.STOREM) {
+            if(ins.opcode==Opcode.STOREM && idx>0) {
                 val prev = indexedInstructions[idx-1].value
                 if(prev.opcode==Opcode.LOADM) {
                     // loadm.X rX,something | storem.X rX,something ?? -> get rid of the store.

codeGenIntermediate/test/TestIRPeepholeOpt.kt

@@ -22,7 +22,8 @@ class TestIRPeepholeOpt: FunSpec({
             floats = false,
             noSysInit = true,
             compTarget = target,
-            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS
+            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS,
+            memtopAddress = 0xffffu
         )
         val prog = IRProgram("test", IRSymbolTable(), options, target)
         prog.addBlock(block)

codeGenIntermediate/test/TestVmCodeGen.kt

@@ -23,7 +23,8 @@ class TestVmCodeGen: FunSpec({
             floats = true,
             noSysInit = false,
             compTarget = target,
-            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS
+            loadAddress = target.machine.PROGRAM_LOAD_ADDRESS,
+            memtopAddress = 0xffffu
         )
     }
 
@@ -44,10 +45,42 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), 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))
-        sub.add(PtVariable("particleDX", DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, 3u, Position.DUMMY))
-        sub.add(PtVariable("xx", DataType.WORD, ZeropageWish.DONTCARE, PtNumber(DataType.WORD, 1.0, Position.DUMMY), null, Position.DUMMY))
+        sub.add(PtVariable(
+            "pi",
+            DataType.UBYTE,
+            ZeropageWish.DONTCARE,
+            0u,
+            PtNumber(DataType.UBYTE, 0.0, Position.DUMMY),
+            null,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "particleX",
+            DataType.ARRAY_UB,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            3u,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "particleDX",
+            DataType.ARRAY_UB,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            3u,
+            Position.DUMMY
+        ))
+        sub.add(PtVariable(
+            "xx",
+            DataType.WORD,
+            ZeropageWish.DONTCARE,
+            0u,
+            PtNumber(DataType.WORD, 1.0, Position.DUMMY),
+            null,
+            Position.DUMMY
+        ))
 
         val assign = PtAugmentedAssign("+=", Position.DUMMY)
         val target = PtAssignTarget(false, Position.DUMMY).also {
@@ -123,7 +156,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "f1",
+            DataType.FLOAT,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.f1", DataType.FLOAT, Position.DUMMY))
@@ -186,7 +227,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "f1",
+            DataType.FLOAT,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.f1", DataType.FLOAT, Position.DUMMY))
@@ -245,7 +294,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("f1", DataType.FLOAT, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "f1",
+            DataType.FLOAT,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.f1", DataType.FLOAT, Position.DUMMY))
@@ -292,7 +349,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "sb1",
+            DataType.BYTE,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.sb1", DataType.BYTE, Position.DUMMY))
@@ -355,7 +420,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("sb1", DataType.BYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "sb1",
+            DataType.BYTE,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.sb1", DataType.BYTE, Position.DUMMY))
@@ -414,7 +487,15 @@ class TestVmCodeGen: FunSpec({
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
-        sub.add(PtVariable("ub1", DataType.UBYTE, ZeropageWish.DONTCARE, null, null, Position.DUMMY))
+        sub.add(PtVariable(
+            "ub1",
+            DataType.UBYTE,
+            ZeropageWish.DONTCARE,
+            0u,
+            null,
+            null,
+            Position.DUMMY
+        ))
         val if1 = PtIfElse(Position.DUMMY)
         val cmp1 = PtBinaryExpression("==", DataType.BOOL, Position.DUMMY)
         cmp1.add(PtIdentifier("main.start.ub1", DataType.UBYTE, Position.DUMMY))
@@ -442,9 +523,9 @@ class TestVmCodeGen: FunSpec({
         irChunks.size shouldBe 1
     }
 
-    test("romsub allowed in ir-codegen") {
+    test("extsub allowed in ir-codegen") {
 //main {
-//    romsub $5000 = routine()
+//    extsub $5000 = routine()
 //
 //    sub start()  {
 //        routine()
@@ -453,8 +534,8 @@ class TestVmCodeGen: FunSpec({
         val codegen = VmCodeGen()
         val program = PtProgram("test", DummyMemsizer, DummyStringEncoder)
         val block = PtBlock("main", false, SourceCode.Generated("test"), PtBlock.Options(), Position.DUMMY)
-        val romsub = PtAsmSub("routine", 0x5000u, setOf(CpuRegister.Y), emptyList(), emptyList(), false, Position.DUMMY)
-        block.add(romsub)
+        val extsub = PtAsmSub("routine", PtAsmSub.Address(null, null, 0x5000u), setOf(CpuRegister.Y), emptyList(), emptyList(), false, Position.DUMMY)
+        block.add(extsub)
         val sub = PtSub("start", emptyList(), null, Position.DUMMY)
         val call = PtFunctionCall("main.routine", true, DataType.UNDEFINED, Position.DUMMY)
         sub.add(call)

codeOptimizers/src/prog8/optimizer/ConstExprEvaluator.kt

@@ -66,7 +66,7 @@ class ConstExprEvaluator {
 //            DataType.WORD -> result = result.toShort().toInt()
 //            else -> { /* keep as it is */ }
 //        }
-        return NumericLiteral.optimalNumeric(result.toDouble(), left.position)
+        return NumericLiteral.optimalNumeric(left.type, null, result.toDouble(), left.position)
     }
 
     private fun bitwiseXor(left: NumericLiteral, right: NumericLiteral): NumericLiteral {
@@ -133,7 +133,7 @@ class ConstExprEvaluator {
         val error = "cannot add $left and $right"
         return when (left.type) {
             in IntegerDatatypes -> when (right.type) {
-                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.number.toInt() + right.number.toInt(), left.position)
+                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.type, right.type, left.number.toInt() + right.number.toInt(), left.position)
                 DataType.FLOAT -> NumericLiteral(DataType.FLOAT, left.number.toInt() + right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
@@ -150,7 +150,7 @@ class ConstExprEvaluator {
         val error = "cannot subtract $left and $right"
         return when (left.type) {
             in IntegerDatatypes -> when (right.type) {
-                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.number.toInt() - right.number.toInt(), left.position)
+                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.type, right.type, left.number.toInt() - right.number.toInt(), left.position)
                 DataType.FLOAT -> NumericLiteral(DataType.FLOAT, left.number.toInt() - right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
@@ -167,7 +167,7 @@ class ConstExprEvaluator {
         val error = "cannot multiply ${left.type} and ${right.type}"
         return when (left.type) {
             in IntegerDatatypes -> when (right.type) {
-                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.number.toInt() * right.number.toInt(), left.position)
+                in IntegerDatatypes -> NumericLiteral.optimalInteger(left.type, right.type, left.number.toInt() * right.number.toInt(), left.position)
                 DataType.FLOAT -> NumericLiteral(DataType.FLOAT, left.number.toInt() * right.number, left.position)
                 else -> throw ExpressionError(error, left.position)
             }
@@ -190,7 +190,7 @@ class ConstExprEvaluator {
                 in IntegerDatatypes -> {
                     if(right.number.toInt()==0) divideByZeroError(right.position)
                     val result: Int = left.number.toInt() / right.number.toInt()
-                    NumericLiteral.optimalInteger(result, left.position)
+                    NumericLiteral.optimalInteger(left.type, right.type, result, left.position)
                 }
                 DataType.FLOAT -> {
                     if(right.number==0.0) divideByZeroError(right.position)
@@ -219,7 +219,7 @@ class ConstExprEvaluator {
             in IntegerDatatypes -> when (right.type) {
                 in IntegerDatatypes -> {
                     if(right.number.toInt()==0) divideByZeroError(right.position)
-                    NumericLiteral.optimalNumeric(left.number.toInt().toDouble() % right.number.toInt().toDouble(), left.position)
+                    NumericLiteral.optimalNumeric(left.type, right.type, left.number.toInt().toDouble() % right.number.toInt().toDouble(), left.position)
                 }
                 DataType.FLOAT -> {
                     if(right.number ==0.0) divideByZeroError(right.position)

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -2,6 +2,7 @@ package prog8.optimizer
 
 import prog8.ast.Node
 import prog8.ast.Program
+import prog8.ast.base.FatalAstException
 import prog8.ast.expressions.*
 import prog8.ast.maySwapOperandOrder
 import prog8.ast.statements.*
@@ -33,6 +34,14 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
     }
 
     override fun after(numLiteral: NumericLiteral, parent: Node): Iterable<IAstModification> {
+
+        if(numLiteral.type==DataType.LONG) {
+            // see if LONG values may be reduced to something smaller
+            val smaller = NumericLiteral.optimalInteger(numLiteral.number.toInt(), numLiteral.position)
+            if(smaller.type!=DataType.LONG)
+                return listOf(IAstModification.ReplaceNode(numLiteral, smaller, parent))
+        }
+
         if(parent is Assignment) {
             val iDt = parent.target.inferType(program)
             if(iDt.isKnown && !iDt.isBool && !iDt.istype(numLiteral.type)) {
@@ -104,6 +113,28 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
             }
         }
 
+        if(expr.left.inferType(program).isArray) {
+            if (expr.operator=="*" && rightconst!=null) {
+                if (expr.left is ArrayLiteral) {
+                    // concatenate array literal.
+                    val part = expr.left as ArrayLiteral
+                    if(part.value.isEmpty())
+                        errors.warn("resulting array has length zero", part.position)
+                    val tmp = mutableListOf<Expression>()
+                    repeat(rightconst.number.toInt()) {
+                        part.value.forEach { tmp += it.copy() }
+                    }
+                    val newArray = ArrayLiteral(part.type, tmp.toTypedArray(), part.position)
+                    return listOf(IAstModification.ReplaceNode(expr, newArray, parent))
+                }
+                else {
+                    val leftTarget = (expr.left as? IdentifierReference)?.targetVarDecl(program)
+                    if(leftTarget!=null && leftTarget.origin==VarDeclOrigin.ARRAYLITERAL)
+                        throw FatalAstException("shouldn't see an array literal converted to an autovar here")
+                }
+            }
+        }
+
         if(expr.operator=="==" && rightconst!=null) {
             val leftExpr = expr.left as? BinaryExpression
             // only do this shuffling when the LHS is not a constant itself (otherwise problematic nested replacements)
@@ -164,13 +195,13 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
         if(leftconst==null && rightconst!=null && rightconst.number<0.0) {
             if (expr.operator == "-") {
                 // X - -1 ---> X + 1
-                val posNumber = NumericLiteral.optimalNumeric(-rightconst.number, rightconst.position)
+                val posNumber = NumericLiteral.optimalNumeric(rightconst.type, null, -rightconst.number, rightconst.position)
                 val plusExpr = BinaryExpression(expr.left, "+", posNumber, expr.position)
                 return listOf(IAstModification.ReplaceNode(expr, plusExpr, parent))
             }
             else if (expr.operator == "+") {
                 // X + -1 ---> X - 1
-                val posNumber = NumericLiteral.optimalNumeric(-rightconst.number, rightconst.position)
+                val posNumber = NumericLiteral.optimalNumeric(rightconst.type, null, -rightconst.number, rightconst.position)
                 val plusExpr = BinaryExpression(expr.left, "-", posNumber, expr.position)
                 return listOf(IAstModification.ReplaceNode(expr, plusExpr, parent))
             }
@@ -290,6 +321,31 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
         return noModifications
     }
 
+    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
+        if(parent is VarDecl && parent.parent is Block) {
+            // only block level (global) initializers are considered here, because they're run just once at program startup
+
+            val const = arrayIndexedExpression.constValue(program)
+            if (const != null)
+                return listOf(IAstModification.ReplaceNode(arrayIndexedExpression, const, parent))
+
+            val constIndex = arrayIndexedExpression.indexer.constIndex()
+            if (constIndex != null) {
+                val arrayVar = arrayIndexedExpression.arrayvar.targetVarDecl(program)
+                if(arrayVar!=null) {
+                    val array =arrayVar.value as? ArrayLiteral
+                    if(array!=null) {
+                        val value = array.value[constIndex].constValue(program)
+                        if(value!=null) {
+                            return listOf(IAstModification.ReplaceNode(arrayIndexedExpression, value, parent))
+                        }
+                    }
+                }
+            }
+        }
+        return noModifications
+    }
+
     override fun after(functionCallExpr: FunctionCallExpression, parent: Node): Iterable<IAstModification> {
         val constvalue = functionCallExpr.constValue(program)
         return if(constvalue!=null)
@@ -384,12 +440,13 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
         val numval = decl.value as? NumericLiteral
         if(decl.type== VarDeclType.CONST && numval!=null) {
             val valueDt = numval.inferType(program)
+            if(valueDt istype DataType.LONG) {
+                return noModifications  // this is handled in the numericalvalue case
+            }
             if(valueDt isnot decl.datatype) {
-                if(decl.datatype!=DataType.BOOL || valueDt.isnot(DataType.UBYTE)) {
-                    val cast = numval.cast(decl.datatype, true)
-                    if (cast.isValid)
-                        return listOf(IAstModification.ReplaceNode(numval, cast.valueOrZero(), decl))
-                }
+                val cast = numval.cast(decl.datatype, true)
+                if (cast.isValid)
+                    return listOf(IAstModification.ReplaceNode(numval, cast.valueOrZero(), decl))
             }
         }
         return noModifications

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -76,7 +76,7 @@ class VarConstantValueTypeAdjuster(
                     if (declValue != null) {
                         // variable is never written to, so it can be replaced with a constant, IF the value is a constant
                         errors.info("variable '${decl.name}' is never written to and was replaced by a constant", decl.position)
-                        val const = VarDecl(VarDeclType.CONST, decl.origin, decl.datatype, decl.zeropage, decl.arraysize, decl.name, decl.names, declValue, decl.sharedWithAsm, decl.splitArray, decl.position)
+                        val const = VarDecl(VarDeclType.CONST, decl.origin, decl.datatype, decl.zeropage, decl.arraysize, decl.name, decl.names, declValue, decl.sharedWithAsm, decl.splitArray, decl.alignment, decl.dirty, decl.position)
                         decl.value = null
                         return listOf(
                             IAstModification.ReplaceNode(decl, const, parent)
@@ -96,7 +96,7 @@ class VarConstantValueTypeAdjuster(
                     }
                     // variable only has a single write and it is the initialization value, so it can be replaced with a constant, IF the value is a constant
                     errors.info("variable '${decl.name}' is never written to and was replaced by a constant", decl.position)
-                    val const = VarDecl(VarDeclType.CONST, decl.origin, decl.datatype, decl.zeropage, decl.arraysize, decl.name, decl.names, singleAssignment.value, decl.sharedWithAsm, decl.splitArray, decl.position)
+                    val const = VarDecl(VarDeclType.CONST, decl.origin, decl.datatype, decl.zeropage, decl.arraysize, decl.name, decl.names, singleAssignment.value, decl.sharedWithAsm, decl.splitArray, decl.alignment, decl.dirty, decl.position)
                     return listOf(
                         IAstModification.ReplaceNode(decl, const, parent),
                         IAstModification.Remove(singleAssignment, singleAssignment.parent as IStatementContainer)
@@ -268,11 +268,9 @@ class VarConstantValueTypeAdjuster(
 
 
 // Replace all constant identifiers with their actual value,
-// and the array var initializer values and sizes.
 // This is needed because further constant optimizations depend on those.
 internal class ConstantIdentifierReplacer(
     private val program: Program,
-    private val options: CompilationOptions,
     private val errors: IErrorReporter
 ) : AstWalker() {
 
@@ -287,7 +285,7 @@ internal class ConstantIdentifierReplacer(
     override fun after(identifier: IdentifierReference, parent: Node): Iterable<IAstModification> {
         // replace identifiers that refer to const value, with the value itself
         // if it's a simple type and if it's not a left hand side variable
-        if(identifier.parent is AssignTarget)
+        if(identifier.parent is AssignTarget || identifier.parent is Alias)
             return noModifications
         var forloop = identifier.parent as? ForLoop
         if(forloop==null)
@@ -392,110 +390,70 @@ internal class ConstantIdentifierReplacer(
         return noModifications
     }
 
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        // convert a range expression that is assigned to an array, to an array literal instead.
+        val range = assignment.value as? RangeExpression
+        if(range!=null) {
+            val targetDatatype = assignment.target.inferType(program)
+            if(targetDatatype.isArray) {
+                val decl = VarDecl(VarDeclType.VAR, VarDeclOrigin.ARRAYLITERAL, targetDatatype.getOr(DataType.UNDEFINED),
+                    ZeropageWish.DONTCARE, null, "dummy", emptyList(),
+                    assignment.value, false, false, 0u, false, Position.DUMMY)
+                val replaceValue = createConstArrayInitializerValue(decl)
+                if(replaceValue!=null) {
+                    return listOf(IAstModification.ReplaceNode(assignment.value, replaceValue, assignment))
+                }
+            }
+        }
+        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.
+        val rangeSize=(decl.value as? RangeExpression)?.size()
+        if(rangeSize!=null && rangeSize>65535) {
+            errors.err("range size overflow", decl.value!!.position)
+            return null
+        }
+
+        val rangeExpr = decl.value as? RangeExpression ?: return null
+
+        // convert the initializer range expression from a range, to an actual array literal.
+        val declArraySize = decl.arraysize?.constIndex()
+        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!!)
+        }
         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(),
+                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
-                val size = decl.arraysize?.constIndex() ?: return null
-                if(rangeExpr==null && numericLv!=null) {
-                    // arraysize initializer is a single int, and we know the array size.
-                    val fillvalue = numericLv.number
-                    if (fillvalue < options.compTarget.machine.FLOAT_MAX_NEGATIVE || fillvalue > options.compTarget.machine.FLOAT_MAX_POSITIVE)
-                        errors.err("float value overflow", numericLv.position)
-                    else {
-                        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 value, and we know the array size.
-                    if(numericLv.type!=DataType.BOOL) {
-                        if(options.strictBool || numericLv.type !in ByteDatatypes)
-                            errors.err("initializer value is not a boolean", numericLv.position)
-                        return null
-                    }
-                    val array = Array(size) {numericLv.number}.map { NumericLiteral(DataType.BOOL, it, numericLv.position) }.toTypedArray<Expression>()
-                    return ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_BOOL), array, position = numericLv.position)
+            DataType.ARRAY_F -> {
+                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) {
+                    return ArrayLiteral(InferredTypes.InferredType.known(DataType.ARRAY_F),
+                        constRange.map { NumericLiteral(DataType.FLOAT, it.toDouble(), decl.value!!.position) }.toTypedArray(),
+                        position = decl.value!!.position)
                 }
             }
             else -> return null

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -11,9 +11,9 @@ import prog8.ast.walk.IAstModification
 import prog8.code.core.*
 import kotlin.math.abs
 import kotlin.math.log2
-import kotlin.math.pow
 
-class ExpressionSimplifier(private val program: Program, private val options: CompilationOptions, private val errors: IErrorReporter) : AstWalker() {
+
+class ExpressionSimplifier(private val program: Program, private val errors: IErrorReporter) : AstWalker() {
     override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
         val mods = mutableListOf<IAstModification>()
 
@@ -238,57 +238,41 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
         if(expr.operator=="==") {
             if(rightDt==DataType.BOOL && leftDt==DataType.BOOL) {
                 val rightConstBool = rightVal?.asBooleanValue
-                if(rightConstBool!=null) {
-                    return if(rightConstBool)
-                        listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
-                    else
-                        listOf(IAstModification.ReplaceNode(expr, PrefixExpression("not", expr.left, expr.position), parent))
+                if(rightConstBool==true) {
+                    return listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
                 }
             }
             if (rightVal?.number == 1.0) {
-                if (options.strictBool) {
-                    if (rightDt != leftDt) {
-                        val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
-                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
-                    }
-                } else
-                    return listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
+                if (rightDt != leftDt) {
+                    val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
+                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                }
             }
             else if (rightVal?.number == 0.0) {
-                if (options.strictBool) {
-                    if (rightDt != leftDt) {
-                        val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
-                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
-                    }
+                if (rightDt != leftDt) {
+                    val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
+                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
                 }
             }
         }
         if (expr.operator=="!=") {
             if(rightDt==DataType.BOOL && leftDt==DataType.BOOL) {
                 val rightConstBool = rightVal?.asBooleanValue
-                if(rightConstBool!=null) {
-                    return if (rightConstBool)
-                        listOf(IAstModification.ReplaceNode(expr, PrefixExpression("not", expr.left, expr.position), parent))
-                    else
-                        listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
+                if(rightConstBool==false) {
+                    listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
                 }
             }
             if (rightVal?.number == 1.0) {
-                if(options.strictBool) {
-                    if(rightDt!=leftDt) {
-                        val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
-                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
-                    }
+                if(rightDt!=leftDt) {
+                    val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
+                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
                 }
             }
             else if (rightVal?.number == 0.0) {
-                if(options.strictBool) {
-                    if(rightDt!=leftDt) {
-                        val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
-                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
-                    }
-                } else
-                    return listOf(IAstModification.ReplaceNode(expr, expr.left, parent))
+                if(rightDt!=leftDt) {
+                    val right = NumericLiteral(leftDt, rightVal.number, rightVal.position)
+                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                }
             }
         }
 
@@ -426,21 +410,6 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
         return noModifications
     }
 
-    override fun after(expr: PrefixExpression, parent: Node): Iterable<IAstModification> {
-        if(expr.operator=="not") {
-            // not  X <compare> Y  ->   X <invertedcompare> Y
-            val binExpr = expr.expression as? BinaryExpression
-            if(binExpr!=null) {
-                val invertedOperator = invertedComparisonOperator(binExpr.operator)
-                if(invertedOperator!=null) {
-                    val inverted = BinaryExpression(binExpr.left, invertedOperator, binExpr.right, binExpr.position)
-                    return listOf(IAstModification.ReplaceNode(expr, inverted, parent))
-                }
-            }
-        }
-        return noModifications
-    }
-
     private fun applyAbsorptionLaws(expr: BinaryExpression): Expression? {
         // NOTE: only when the terms are not function calls!!!
         if(expr.left is IFunctionCall || expr.right is IFunctionCall)

codeOptimizers/src/prog8/optimizer/Extensions.kt

@@ -12,7 +12,7 @@ fun Program.constantFold(errors: IErrorReporter, options: CompilationOptions) {
     if(errors.noErrors()) {
         valuetypefixer.applyModifications()
 
-        val replacer = ConstantIdentifierReplacer(this, options, errors)
+        val replacer = ConstantIdentifierReplacer(this, errors)
         replacer.visit(this)
         if (errors.noErrors()) {
             replacer.applyModifications()
@@ -59,8 +59,8 @@ fun Program.inlineSubroutines(options: CompilationOptions): Int {
     return inliner.applyModifications()
 }
 
-fun Program.simplifyExpressions(errors: IErrorReporter, options: CompilationOptions) : Int {
-    val opti = ExpressionSimplifier(this, options, errors)
+fun Program.simplifyExpressions(errors: IErrorReporter) : Int {
+    val opti = ExpressionSimplifier(this, errors)
     opti.visit(this)
     return opti.applyModifications()
 }

codeOptimizers/src/prog8/optimizer/Inliner.kt

@@ -224,8 +224,11 @@ class Inliner(private val program: Program, private val options: CompilationOpti
                             noModifications
                     }
                     else -> {
-                        sub.hasBeenInlined=true
-                        listOf(IAstModification.ReplaceNode(origNode, toInline.copy(), parent))
+                        if(origNode !== toInline) {
+                            sub.hasBeenInlined = true
+                            listOf(IAstModification.ReplaceNode(origNode, toInline.copy(), parent))
+                        } else
+                            noModifications
                     }
                 }
             }
@@ -255,7 +258,7 @@ class Inliner(private val program: Program, private val options: CompilationOpti
                     is Return -> {
                         // is an expression, so we have to have a Return here in the inlined sub
                         // note that we don't have to process any args, because we online inline parameterless subroutines.
-                        if(toInline.value!=null) {
+                        if(toInline.value!=null && functionCallExpr!==toInline.value) {
                             sub.hasBeenInlined=true
                             listOf(IAstModification.ReplaceNode(functionCallExpr, toInline.value!!.copy(), parent))
                         }

codeOptimizers/src/prog8/optimizer/StatementOptimizer.kt

@@ -127,19 +127,6 @@ class StatementOptimizer(private val program: Program,
                     )
                 }
             }
-
-            // switch if/else around if the condition is a not
-            val prefixCond = ifElse.condition as? PrefixExpression
-            if(prefixCond?.operator=="not") {
-                errors.info("invert conditon and swap if/else blocks", ifElse.condition.position)
-                val newTruePart = AnonymousScope(ifElse.elsepart.statements, ifElse.elsepart.position)
-                val newElsePart = AnonymousScope(ifElse.truepart.statements, ifElse.truepart.position)
-                return listOf(
-                    IAstModification.ReplaceNode(ifElse.elsepart, newElsePart, ifElse),
-                    IAstModification.ReplaceNode(ifElse.truepart, newTruePart, ifElse),
-                    IAstModification.ReplaceNode(ifElse.condition, prefixCond.expression, ifElse)
-                )
-            }
         }
 
         return noModifications
@@ -200,35 +187,6 @@ class StatementOptimizer(private val program: Program,
             }
         }
 
-        val loopvarDt = forLoop.loopVarDt(program)
-        if(loopvarDt.istype(DataType.UWORD) || loopvarDt.istype(DataType.UBYTE)) {
-
-            fun incOrDec(inc: Boolean): Assignment {
-                val pos = forLoop.position
-                val loopVar = forLoop.loopVar
-                val addSubOne = BinaryExpression(loopVar.copy(), if(inc) "+" else "-", NumericLiteral.optimalInteger(1, pos), pos, false)
-                return Assignment(AssignTarget(loopVar.copy(), null, null, null, false, pos), addSubOne, AssignmentOrigin.USERCODE, pos)
-            }
-
-            if (range != null && range.to.constValue(program)?.number == 0.0 && range.step.constValue(program)?.number==-1.0) {
-                val fromExpr = range.from
-                if(fromExpr.constValue(program)==null) {
-                    // FOR X = something DOWNTO 0 {...} -->  X=something,  DO { ... , X-- } UNTIL X=255 (or 65535 if uword)
-                    val pos = forLoop.position
-                    val checkValue = NumericLiteral(loopvarDt.getOr(DataType.UNDEFINED), if(loopvarDt.istype(DataType.UBYTE)) 255.0 else 65535.0, pos)
-                    val condition = BinaryExpression(forLoop.loopVar.copy(), "==", checkValue, pos)
-                    val decOne = incOrDec(false)
-                    forLoop.body.statements.add(decOne)
-                    val replacement = AnonymousScope(mutableListOf(
-                        Assignment(AssignTarget(forLoop.loopVar.copy(), null, null, null, false, pos),
-                            fromExpr, AssignmentOrigin.OPTIMIZER, pos),
-                        UntilLoop(forLoop.body, condition, pos)
-                    ), pos)
-                    return listOf(IAstModification.ReplaceNode(forLoop, replacement, parent))
-                }
-            }
-        }
-
         return noModifications
     }
 

codeOptimizers/src/prog8/optimizer/UnusedCodeRemover.kt

@@ -21,6 +21,31 @@ class UnusedCodeRemover(private val program: Program,
 ): AstWalker() {
 
     private lateinit var callgraph: CallGraph
+    private val neverRemoveSubroutines = mutableListOf<Subroutine>()
+
+    init {
+        neverRemoveSubroutines.add(program.entrypoint)
+
+        program.allBlocks.singleOrNull { it.name=="sys" } ?.let {
+            val subroutines = it.statements.filterIsInstance<Subroutine>()
+            val push = subroutines.single { it.name == "push" }
+            val pushw = subroutines.single { it.name == "pushw" }
+            val pop = subroutines.single { it.name == "pop" }
+            val popw = subroutines.single { it.name == "popw" }
+            neverRemoveSubroutines.add(push)
+            neverRemoveSubroutines.add(pushw)
+            neverRemoveSubroutines.add(pop)
+            neverRemoveSubroutines.add(popw)
+        }
+
+        program.allBlocks.singleOrNull { it.name=="floats" } ?.let {
+            val subroutines = it.statements.filterIsInstance<Subroutine>()
+            val push = subroutines.single { it.name == "push" }
+            val pop = subroutines.single { it.name == "pop" }
+            neverRemoveSubroutines.add(push)
+            neverRemoveSubroutines.add(pop)
+        }
+    }
 
     override fun before(program: Program): Iterable<IAstModification> {
         callgraph = CallGraph(program)
@@ -92,7 +117,7 @@ class UnusedCodeRemover(private val program: Program,
 
     override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
         val forceOutput = "force_output" in subroutine.definingBlock.options()
-        if (subroutine !== program.entrypoint && !forceOutput && !subroutine.isAsmSubroutine) {
+        if (subroutine !in neverRemoveSubroutines && !forceOutput && !subroutine.isAsmSubroutine) {
             if(callgraph.unused(subroutine)) {
                 if(subroutine.containsNoCodeNorVars) {
                     if("ignore_unused" !in subroutine.definingBlock.options())

compiler/build.gradle

@@ -32,6 +32,7 @@ dependencies {
     implementation project(':codeGenIntermediate')
     implementation project(':codeGenExperimental')
     implementation project(':virtualmachine')
+    // implementation project(':beanshell')
     implementation "org.antlr:antlr4-runtime:4.13.2"
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
@@ -41,6 +42,7 @@ dependencies {
     testImplementation project(':codeCore')
     testImplementation project(':intermediate')
     testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.1'
+    testImplementation 'io.kotest:kotest-framework-datatest:5.9.1'
     testImplementation 'org.junit.jupiter:junit-jupiter:5.9.2'
     testRuntimeOnly 'org.junit.platform:junit-platform-launcher'
 
@@ -79,11 +81,11 @@ startScripts.enabled = true
 
 application {
     mainClass = 'prog8.CompilerMainKt'
-    applicationName = 'p8compile'
+    applicationName = 'prog8c'
 }
 
 shadowJar {
-    archiveBaseName = 'prog8compiler'
+    archiveBaseName = 'prog8c'
     archiveVersion = version
     // minimize()
 }

compiler/compiler.iml

@@ -15,7 +15,7 @@
     <orderEntry type="library" name="michael.bull.kotlin.result.jvm" level="project" />
     <orderEntry type="library" name="io.kotest.assertions.core.jvm" level="project" />
     <orderEntry type="library" name="io.kotest.runner.junit5.jvm" level="project" />
-    <orderEntry type="library" name="antlr.antlr4" level="project" />
+    <orderEntry type="library" name="io.kotest.framework.datatest" level="project" />
     <orderEntry type="module" module-name="codeCore" />
     <orderEntry type="module" module-name="compilerAst" />
     <orderEntry type="module" module-name="codeOptimizers" />
@@ -24,5 +24,6 @@
     <orderEntry type="module" module-name="codeGenIntermediate" />
     <orderEntry type="module" module-name="virtualmachine" />
     <orderEntry type="module" module-name="intermediate" scope="TEST" />
+    <orderEntry type="library" name="antlr.antlr4" level="project" />
   </component>
 </module>

compiler/res/prog8lib/atari/syslib.p8

@@ -12,37 +12,23 @@ atari {
         &uword COLCRS = 85
         &ubyte ROWCRS = 84
 
-    romsub $F24A = getchar() -> ubyte @A
-    romsub $F2B0 = outchar(ubyte character @ A)
-    romsub $F2FD = waitkey() -> ubyte @A
+    extsub $F24A = getchar() -> ubyte @A
+    extsub $F2B0 = outchar(ubyte character @ A)
+    extsub $F2FD = waitkey() -> ubyte @A
 
 }
 
 sys {
     ; ------- lowlevel system routines --------
 
-    const ubyte target = 8         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 8 = atari800XL
+    const ubyte target = 8         ;  compilation target specifier.  255=virtual, 128=C128, 64=C64, 32=PET, 16=CommanderX16, 8=atari800XL, 7=Neo6502
 
-    asmsub  init_system()  {
-        ; Initializes the machine to a sane starting state.
-        ; Called automatically by the loader program logic.
-        ; TODO
-        %asm {{
-            sei
-            cld
-            clc
-            ; TODO reset screen mode etc etc
-            clv
-            cli
-            rts
-        }}
-    }
-
-    asmsub  init_system_phase2()  {
-        %asm {{
-            rts     ; no phase 2 steps on the Atari
-        }}
-    }
+    const ubyte sizeof_bool = 1
+    const ubyte sizeof_byte = 1
+    const ubyte sizeof_ubyte = 1
+    const ubyte sizeof_word = 2
+    const ubyte sizeof_uword = 2
+    const ubyte sizeof_float = 0    ; undefined, no float support
 
     asmsub  reset_system()  {
         ; Soft-reset the system back to initial power-on Basic prompt.
@@ -154,6 +140,65 @@ _longcopy
         }}
     }
 
+    asmsub memcmp(uword address1 @R0, uword address2 @R1, uword size @AY) -> byte @A {
+        ; Compares two blocks of memory
+        ; Returns -1 (255), 0 or 1, meaning: block 1 sorts before, equal or after block 2.
+        %asm {{
+            sta  P8ZP_SCRATCH_REG   ; lsb(size)
+            sty  P8ZP_SCRATCH_B1    ; msb(size)
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            sta  P8ZP_SCRATCH_W2
+            sty  P8ZP_SCRATCH_W2+1
+
+            ldx  P8ZP_SCRATCH_B1
+            beq  _no_msb_size
+
+_loop_msb_size
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dec  P8ZP_SCRATCH_B1        ; msb(size) -= 1
+            dex
+            bne  _loop_msb_size
+
+_no_msb_size
+            lda  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  +
+            rts
+
++           ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            cpy  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  -
+
+            lda #0
+            rts
+        }}
+    }
+
     inline asmsub read_flags() -> ubyte @A {
         %asm {{
         php
@@ -225,6 +270,7 @@ save_SCRATCH_ZPB1	.byte  0
 save_SCRATCH_ZPREG	.byte  0
 save_SCRATCH_ZPWORD1	.word  0
 save_SCRATCH_ZPWORD2	.word  0
+            ; !notreached!
         }}
     }
 
@@ -248,27 +294,38 @@ save_SCRATCH_ZPWORD2	.word  0
 
     asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
-        ;    TODO where to store A as exit code?
         %asm {{
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
             ldx  prog8_lib.orig_stackpointer
             txs
-            rts		; return to original caller
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit2(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y) {
         ; -- immediately exit the program with result values in the A, X and Y registers.
-        ;    TODO where to store A,X,Y as exit code?
         %asm {{
-            jmp  exit
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit3(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y, bool carry @Pc) {
         ; -- immediately exit the program with result values in the A, X and Y registers, and the Carry flag in the status register.
-        ;    TODO where to store A,X,Y,Carry as exit code?
         %asm {{
-            jmp  exit
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
+            lda  #0
+            rol  a
+            sta  p8_sys_startup.cleanup_at_exit._exitcodeCarry
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
@@ -279,6 +336,13 @@ save_SCRATCH_ZPWORD2	.word  0
         }}
     }
 
+    inline asmsub progstart() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_start
+            ldy  #>prog8_program_start
+        }}
+    }
+
     inline asmsub push(ubyte value @A) {
         %asm {{
             pha
@@ -427,6 +491,7 @@ cx16 {
     _cx16_vreg_storage
             .word 0,0,0,0,0,0,0,0
             .word 0,0,0,0,0,0,0,0
+            ; !notreached!
         }}
     }
 
@@ -447,3 +512,44 @@ cx16 {
     }
 
 }
+
+p8_sys_startup {
+    ; program startup and shutdown machinery. Needs to reside in normal system ram.
+
+    asmsub  init_system()  {
+        ; Initializes the machine to a sane starting state.
+        ; Called automatically by the loader program logic.
+        ; TODO
+        %asm {{
+            sei
+            ; TODO reset screen mode etc etc
+            cli
+            rts
+        }}
+    }
+
+    asmsub  init_system_phase2()  {
+        %asm {{
+            cld
+            clc
+            clv
+            rts
+        }}
+    }
+
+    asmsub  cleanup_at_exit() {
+        ; executed when the main subroutine does rts
+        %asm {{
+_exitcodeCarry = *+1
+            lda  #0
+            lsr  a
+_exitcode = *+1
+            lda  #0        ; exit code possibly modified in exit()
+_exitcodeX = *+1
+            ldx  #0
+_exitcodeY = *+1
+            ldy  #0
+            rts
+        }}
+    }
+}

compiler/res/prog8lib/buffers.p8

@@ -1,4 +1,4 @@
-; experimental buffer data structures
+; **experimental** buffer data structures, API subject to change!!
 
 %option no_symbol_prefixing, ignore_unused
 

compiler/res/prog8lib/c128/diskio.p8

@@ -0,0 +1,4 @@
+%import shared_cbm_diskio
+
+; No alterations here; everything is taken from the shared module.
+

compiler/res/prog8lib/c128/syslib.p8

@@ -56,51 +56,49 @@ cbm {
 ; CLEARSCR -> use txt.clear_screen
 ; HOMECRSR -> use txt.home or txt.plot
 
-romsub $FA65 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
-romsub $FF33 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
+extsub $FA65 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
+extsub $FF33 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
 
-; TODO c128 a bunch of kernal routines are missing here that are specific to the c128
-
-romsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
-romsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
-romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
-romsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
-romsub $FF8D = VECTOR(uword userptr @ XY, bool dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
-romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
-romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
-romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
-romsub $FF99 = MEMTOP(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
-romsub $FF9C = MEMBOT(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
-romsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
-romsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
-romsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
-romsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
-romsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
-romsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
-romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
-romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
-romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word  (use CLEARST to reset it to 0)
-romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
-romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
-romsub $FFC0 = OPEN() clobbers(X,Y) -> bool @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
-romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
-romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798 ($31E)) define an input channel
-romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
-romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
-romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-romsub $FFD2 = CHROUT(ubyte 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
-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)     also see STOP2
-romsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; (via 810 ($32A)) get a character       also see GETIN2
-romsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
-romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
-romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; get current window dimensions into X (columns) and Y (rows)  NOTE: changed behavior compared to VIC/C64/PET SCREEN() routine!
-romsub $FFED = SCRORG() -> ubyte @ X, ubyte @ Y                 ; get current window dimensions into X (columns) and Y (rows)  NOTE: changed behavior compared to VIC/C64/PET SCREEN() routine!
-romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, bool dir @ Pc) clobbers(A) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
-romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
+extsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
+extsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
+extsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
+extsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
+extsub $FF8D = VECTOR(uword userptr @ XY, bool dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
+extsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
+extsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
+extsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
+extsub $FF99 = MEMTOP(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
+extsub $FF9C = MEMBOT(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
+extsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
+extsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
+extsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
+extsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
+extsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
+extsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
+extsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
+extsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
+extsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word  (use CLEARST to reset it to 0)
+extsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
+extsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
+extsub $FFC0 = OPEN() clobbers(X,Y) -> bool @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
+extsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
+extsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798 ($31E)) define an input channel
+extsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
+extsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
+extsub $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.
+extsub $FFD2 = CHROUT(ubyte character @ A)                           ; (via 806 ($326)) output a character
+extsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+extsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
+extsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
+extsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock (A=lo,X=mid,Y=high)
+extsub $FFE1 = STOP() clobbers(X) -> bool @ Pz, ubyte @ A       ; (via 808 ($328)) check the STOP key (and some others in A)     also see STOP2
+extsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; (via 810 ($32A)) get a character       also see GETIN2
+extsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
+extsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+extsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; get current window dimensions into X (columns) and Y (rows)  NOTE: changed behavior compared to VIC/C64/PET SCREEN() routine!
+extsub $FFED = SCRORG() -> ubyte @ X, ubyte @ Y                 ; get current window dimensions into X (columns) and Y (rows)  NOTE: changed behavior compared to VIC/C64/PET SCREEN() routine!
+extsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, bool dir @ Pc) clobbers(A) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
+extsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
 
 ; ---- end of C64 compatible ROM kernal routines ----
 
@@ -319,8 +317,27 @@ c128 {
     &ubyte  VM4     = $0A2F         ; starting page for VDC attribute mem
 
 
+; TODO c128 a bunch of kernal routines are missing here that are specific to the c128
+
+extsub $FF6E = JSRFAR()
+
+
 ; ---- C128 specific system utility routines: ----
 
+inline asmsub banks(ubyte banks @A) {
+    ; -- set the memory bank configuration MMU register
+    %asm {{
+        sta  $FF00
+    }}
+}
+
+inline asmsub getbanks() -> ubyte @A {
+    ; -- get the current memory bank configuration from the MMU register
+    %asm {{
+        lda  $FF00
+    }}
+}
+
 asmsub  disable_basic() clobbers(A) {
     %asm {{
         lda $0a04   ; disable BASIC shadow registers
@@ -342,6 +359,56 @@ asmsub  disable_basic() clobbers(A) {
     }}
 }
 
+asmsub  x16jsrfar() {
+    %asm {{
+        ; setup a JSRFAR call (using X16 call convention)
+        ; see https://cx16.dk/c128-kernal-routines/jsrfar.html
+        sty  $08                ; save registers
+        stx  $07
+        sta  $06
+        php                    ; including PSR
+        pla
+        sta  $05
+
+        pla                    ; get original return address
+        sta  $fa                ; and store it in a temp ZP pointer
+        pla
+        sta  $fb
+
+        ldy  #$01
+        lda  ($fa),y            ; grab low byte of target address
+        sta  $04
+        iny
+        lda  ($fa),y            ; now the high byte
+        sta  $03
+        iny
+        lda  ($fa),y            ; then the target bank
+        sta  $02
+
+        ; replace the original return address by it + 3 to skip the data bytes
+        clc
+        lda  $fa
+        adc  #3
+        sta  $fa
+        lda  $fb
+        adc  #0
+        pha
+        lda  $fa
+        pha
+
+        jsr c128.JSRFAR        ; call kernal's jsrfar routine
+
+        lda  $05                ; populate registers
+        pha
+        lda  $06
+        ldx  $07
+        ldy  $08
+        plp
+
+        rts                    ; and return
+    }}
+}
+
 ; ---- end of C128 specific system utility routines ----
 
 }
@@ -349,82 +416,23 @@ asmsub  disable_basic() clobbers(A) {
 sys {
     ; ------- lowlevel system routines --------
 
-    const ubyte target = 128         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16.
+    const ubyte target = 128         ;  compilation target specifier.  255=virtual, 128=C128, 64=C64, 32=PET, 16=CommanderX16, 8=atari800XL, 7=Neo6502
 
-asmsub  init_system()  {
-    ; Initializes the machine to a sane starting state.
-    ; Called automatically by the loader program logic.
-    ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
-    ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
-    ; Also a different color scheme is chosen to identify ourselves a little.
-    ; Uppercase charset is activated.
-    %asm {{
-        sei
-        cld
-        lda  #0
-        sta  $ff00      ; select default bank 15
-        jsr  cbm.IOINIT
-        jsr  cbm.RESTOR
-        jsr  cbm.CINT
-        lda  #6
-        sta  c64.EXTCOL
-        lda  #7
-        sta  cbm.COLOR
-        lda  #0
-        sta  c64.BGCOL0
-        jsr  disable_runstop_and_charsetswitch
-        clc
-        clv
-        cli
-        rts
-    }}
-}
+    const ubyte sizeof_bool = 1
+    const ubyte sizeof_byte = 1
+    const ubyte sizeof_ubyte = 1
+    const ubyte sizeof_word = 2
+    const ubyte sizeof_uword = 2
+    const ubyte sizeof_float = 5
 
-asmsub  init_system_phase2()  {
-    %asm {{
-        rts     ; no phase 2 steps on the C128
-    }}
-}
 
-asmsub  cleanup_at_exit() {
-    ; executed when the main subroutine does rts
-    %asm {{
-        lda  #0
-        sta  $ff00          ; default bank 15
-        jsr  cbm.CLRCHN		; reset i/o channels
-        jsr  enable_runstop_and_charsetswitch
-_exitcodeCarry = *+1
-        lda  #0
-        lsr  a
-_exitcode = *+1
-        lda  #0        ; exit code possibly modified in exit()
-_exitcodeX = *+1
-        ldx  #0
-_exitcodeY = *+1
-        ldy  #0
-        rts
-    }}
-}
+    sub  disable_runstop_and_charsetswitch() {
+        p8_sys_startup.disable_runstop_and_charsetswitch()
+    }
 
-asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
-    %asm {{
-        lda  #$80
-        sta  247    ; disable charset switching
-        lda  #112
-        sta  808    ; disable run/stop key
-        rts
-    }}
-}
-
-asmsub  enable_runstop_and_charsetswitch() clobbers(A) {
-    %asm {{
-        lda  #0
-        sta  247    ; enable charset switching
-        lda  #110
-        sta  808    ; enable run/stop key
-        rts
-    }}
-}
+    sub  enable_runstop_and_charsetswitch() {
+        p8_sys_startup.enable_runstop_and_charsetswitch()
+    }
 
     asmsub save_prog8_internals() {
         %asm {{
@@ -445,6 +453,7 @@ save_SCRATCH_ZPB1	.byte  0
 save_SCRATCH_ZPREG	.byte  0
 save_SCRATCH_ZPWORD1	.word  0
 save_SCRATCH_ZPWORD2	.word  0
+            ; !notreached!
         }}
     }
 
@@ -496,8 +505,6 @@ _modified
 		tax
 		pla
 		rti
-
-_use_kernal     .byte  0
 		}}
 }
 
@@ -719,6 +726,65 @@ _longcopy
         }}
     }
 
+    asmsub memcmp(uword address1 @R0, uword address2 @R1, uword size @AY) -> byte @A {
+        ; Compares two blocks of memory
+        ; Returns -1 (255), 0 or 1, meaning: block 1 sorts before, equal or after block 2.
+        %asm {{
+            sta  P8ZP_SCRATCH_REG   ; lsb(size)
+            sty  P8ZP_SCRATCH_B1    ; msb(size)
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            sta  P8ZP_SCRATCH_W2
+            sty  P8ZP_SCRATCH_W2+1
+
+            ldx  P8ZP_SCRATCH_B1
+            beq  _no_msb_size
+
+_loop_msb_size
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dec  P8ZP_SCRATCH_B1        ; msb(size) -= 1
+            dex
+            bne  _loop_msb_size
+
+_no_msb_size
+            lda  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  +
+            rts
+
++           ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            cpy  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  -
+
+            lda #0
+            rts
+        }}
+    }
+
     inline asmsub read_flags() -> ubyte @A {
         %asm {{
         php
@@ -780,37 +846,37 @@ _longcopy
     asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         %asm {{
-            sta  cleanup_at_exit._exitcode
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit2(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y) {
         ; -- immediately exit the program with result values in the A, X and Y registers.
         %asm {{
-            sta  cleanup_at_exit._exitcode
-            stx  cleanup_at_exit._exitcodeX
-            sty  cleanup_at_exit._exitcodeY
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit3(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y, bool carry @Pc) {
         ; -- immediately exit the program with result values in the A, X and Y registers, and the Carry flag in the status register.
         %asm {{
-            sta  cleanup_at_exit._exitcode
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
             lda  #0
             rol  a
-            sta  cleanup_at_exit._exitcodeCarry
-            stx  cleanup_at_exit._exitcodeX
-            sty  cleanup_at_exit._exitcodeY
+            sta  p8_sys_startup.cleanup_at_exit._exitcodeCarry
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
@@ -821,6 +887,13 @@ _longcopy
         }}
     }
 
+    inline asmsub progstart() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_start
+            ldy  #>prog8_program_start
+        }}
+    }
+
     inline asmsub push(ubyte value @A) {
         %asm {{
             pha
@@ -970,6 +1043,7 @@ cx16 {
     _cx16_vreg_storage
             .word 0,0,0,0,0,0,0,0
             .word 0,0,0,0,0,0,0,0
+            ; !notreached!
         }}
     }
 
@@ -990,3 +1064,85 @@ cx16 {
     }
 
 }
+
+p8_sys_startup {
+    ; program startup and shutdown machinery. Needs to reside in normal system ram.
+
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
+    ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
+    ; Also a different color scheme is chosen to identify ourselves a little.
+    ; Uppercase charset is activated.
+    %asm {{
+        sei
+        lda  #0
+        sta  $ff00      ; select default bank 15
+        jsr  cbm.IOINIT
+        jsr  cbm.RESTOR
+        jsr  cbm.CINT
+        lda  #%00001110
+        sta  $ff00      ; bank out basic rom so we have ram from $1c00-$bfff
+        lda  #6
+        sta  c64.EXTCOL
+        lda  #7
+        sta  cbm.COLOR
+        lda  #0
+        sta  c64.BGCOL0
+        jsr  disable_runstop_and_charsetswitch
+        cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        cld
+        clc
+        clv
+        rts
+    }}
+}
+
+asmsub  cleanup_at_exit() {
+    ; executed when the main subroutine does rts
+    %asm {{
+        lda  #0
+        sta  $ff00          ; default bank 15
+        jsr  cbm.CLRCHN		; reset i/o channels
+        jsr  enable_runstop_and_charsetswitch
+_exitcodeCarry = *+1
+        lda  #0
+        lsr  a
+_exitcode = *+1
+        lda  #0        ; exit code possibly modified in exit()
+_exitcodeX = *+1
+        ldx  #0
+_exitcodeY = *+1
+        ldy  #0
+        rts
+    }}
+}
+
+asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #$80
+        sta  247    ; disable charset switching
+        lda  #112
+        sta  808    ; disable run/stop key
+        rts
+    }}
+}
+
+asmsub  enable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #0
+        sta  247    ; enable charset switching
+        lda  #110
+        sta  808    ; enable run/stop key
+        rts
+    }}
+}
+
+}

compiler/res/prog8lib/c128/textio.p8

@@ -13,7 +13,7 @@ txt {
 const ubyte DEFAULT_WIDTH = 40
 const ubyte DEFAULT_HEIGHT = 25
 
-romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
+extsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
 
 
 sub  clear_screen() {
@@ -314,6 +314,7 @@ _mod		sta  $ffff		; modified
 		rts
 
 _screenrows	.word  cbm.Screen + range(0, 1000, 40)
+        ; !notreached!
 	}}
 }
 
@@ -357,6 +358,7 @@ _mod		sta  $ffff		; modified
 		rts
 
 _colorrows	.word  $d800 + range(0, 1000, 40)
+        ; !notreached!
 	}}
 }
 

compiler/res/prog8lib/c64/diskio.p8

@@ -0,0 +1,4 @@
+%import shared_cbm_diskio
+
+; No alterations here; everything is taken from the shared module.
+

compiler/res/prog8lib/c64/floats.asm

@@ -120,12 +120,8 @@ cast_FAC1_as_uw_into_ya	.proc               ; also used for float 2 ub
 		.pend
 
 cast_FAC1_as_w_into_ay	.proc               ; also used for float 2 b
-		; -- cast fac1 to word into A/Y
-		;    clobbers X
-		jsr  AYINT
-		ldy  floats.AYINT_facmo
-		lda  floats.AYINT_facmo+1
-		rts
+		; -- cast fac1 to word into A/Y.  clobbers X
+		jmp  AYINT2
 		.pend
 
 

compiler/res/prog8lib/c64/floats.p8

@@ -15,71 +15,85 @@ floats {
 ; note: fac1/2 might get clobbered even if not mentioned in the function's name.
 ; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
 
-romsub $bba2 = MOVFM(uword mflpt @ AY) clobbers(A,Y)        ; load mflpt value from memory  in A/Y into fac1
-romsub $bba6 = FREADMEM() clobbers(A,Y)                     ; load mflpt value from memory  in $22/$23 into fac1
-romsub $ba8c = CONUPK(uword mflpt @ AY) clobbers(A,Y)       ; load mflpt value from memory  in A/Y into fac2
-romsub $ba90 = FAREADMEM() clobbers(A,Y)                    ; load mflpt value from memory  in $22/$23 into fac2
-romsub $bbfc = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
-romsub $bc0c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded the least significant bit)
-romsub $bc0f = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
-romsub $bbd4 = MOVMF(uword mflpt @ XY) clobbers(A,Y)        ; store fac1 to memory  X/Y as 5-byte mflpt
+extsub $bba2 = MOVFM(uword mflpt @ AY) clobbers(A,Y)        ; load mflpt value from memory  in A/Y into fac1
+extsub $bba6 = FREADMEM() clobbers(A,Y)                     ; load mflpt value from memory  in $22/$23 into fac1
+extsub $ba8c = CONUPK(uword mflpt @ AY) clobbers(A,Y)       ; load mflpt value from memory  in A/Y into fac2
+extsub $ba90 = FAREADMEM() clobbers(A,Y)                    ; load mflpt value from memory  in $22/$23 into fac2
+extsub $bbfc = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+extsub $bc0c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded the least significant bit)
+extsub $bc0f = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+extsub $bbd4 = MOVMF(uword mflpt @ XY) clobbers(A,Y)        ; store fac1 to memory  X/Y as 5-byte mflpt
 
 ; fac1 -> unsigned word in Y/A (might throw ILLEGAL QUANTITY) (result also in $14/15)
 ; (tip: use floats.GETADRAY to get A/Y output; lo/hi switched to normal little endian order)
-romsub $b7f7 = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
+extsub $b7f7 = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
 
-romsub $bc9b = QINT() clobbers(A,X,Y)           ; fac1 -> 4-byte signed integer in 98-101 ($62-$65), with the MSB FIRST.
-romsub $b1bf = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101 ($64-$65) MSB FIRST. (might throw ILLEGAL QUANTITY)
+extsub $bc9b = QINT() clobbers(A,X,Y)           ; fac1 -> 4-byte signed integer in 98-101 ($62-$65), with the MSB FIRST.
+extsub $b1bf = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101 ($64-$65) MSB FIRST. (might throw ILLEGAL QUANTITY) DON'T USE THIS, USE WRAPPER 'AYINT2' INSTEAD.
 
 ; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1
 ; (tip: use floats.GIVAYFAY to use A/Y input; lo/hi switched to normal order)
 ; there is also floats.GIVUAYFAY - unsigned word in A/Y (lo/hi) to fac1
 ; there is also floats.FREADS32  that reads from 98-101 ($62-$65) MSB FIRST
 ; there is also floats.FREADUS32  that reads from 98-101 ($62-$65) MSB FIRST
-; there is also floats.FREADS24AXY  that reads signed int24 into fac1 from A/X/Y (lo/mid/hi bytes)
-romsub $b391 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
+; there is also floats.FREAD{S,U}24AXY  that read (un)signed int24 into fac1 from A/X/Y (lo/mid/hi bytes)
+extsub $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.  Also see parse()
-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)
+extsub $b3a2 = FREADUY(ubyte value @ Y) clobbers(A,X,Y)     ; 8 bit unsigned Y -> float in fac1
+extsub $bc3c = FREADSA(byte value @ A) clobbers(A,X,Y)      ; 8 bit signed A -> float in fac1
+extsub $b7b5 = FREADSTR(ubyte length @ A) clobbers(A,X,Y)   ; str -> fac1, $22/23 must point to string, A=string length.  Also see parse()
+extsub $aabc = FPRINTLN() clobbers(A,X,Y)                   ; print string of fac1, on one line (= with newline) destroys fac1.  (consider FOUT + STROUT as well)
+extsub $bddd = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY ($0100)
 
-romsub $b849 = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for integer rounding- call this before INT
-romsub $bae2 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
-romsub $bafe = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive! You have to fix sign manually!
-romsub $bc5b = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
+extsub $b849 = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for integer rounding- call this before INT
+extsub $bae2 = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
+extsub $bafe = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive! You have to fix sign manually!
+extsub $bc5b = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
 
-romsub $b86a = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
-romsub $b867 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
-romsub $b853 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
-romsub $b850 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
-romsub $ba2b = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
-romsub $ba28 = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
-romsub $bb12 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  mind the order of the operands
-romsub $bb0f = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1
-romsub $bf7b = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
-romsub $bf78 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** mflpt from A/Y
-romsub $bd7e = FINLOG(byte value @A) clobbers (A, X, Y)     ; fac1 += signed byte in A
+extsub $b86a = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
+extsub $b867 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
+extsub $b853 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
+extsub $b850 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
+extsub $ba2b = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
+extsub $ba28 = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
+extsub $bb12 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  mind the order of the operands
+extsub $bb0f = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1
+extsub $bf7b = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+extsub $bf78 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** mflpt from A/Y
+extsub $bd7e = FINLOG(byte value @A) clobbers (A, X, Y)     ; fac1 += signed byte in A
 
-romsub $aed4 = NOTOP() clobbers(A,X,Y)                      ; fac1 = NOT(fac1)
-romsub $bccc = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to integer round instead of trunc
-romsub $b9ea = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
-romsub $bc39 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
-romsub $bc2b = SIGN() -> ubyte @ A                          ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
-romsub $bc58 = ABS()                                        ; fac1 = ABS(fac1)
-romsub $bf71 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
-romsub $bf74 = SQRA() clobbers(A,X,Y)                       ; fac1 = SQRT(fac2)
-romsub $bfed = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
-romsub $bfb4 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
-romsub $b8d7 = NORMAL() clobbers(A)                         ; normalize FAC1
-romsub $e097 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
-romsub $e264 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
-romsub $e26b = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
-romsub $e2b4 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
-romsub $e30e = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+extsub $aed4 = NOTOP() clobbers(A,X,Y)                      ; fac1 = NOT(fac1)
+extsub $bccc = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to integer round instead of trunc
+extsub $b9ea = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
+extsub $bc39 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
+extsub $bc2b = SIGN() -> ubyte @ A                          ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+extsub $bc58 = ABS()                                        ; fac1 = ABS(fac1)
+extsub $bf71 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+extsub $bf74 = SQRA() clobbers(A,X,Y)                       ; fac1 = SQRT(fac2)
+extsub $bfed = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+extsub $bfb4 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
+extsub $b8d7 = NORMAL() clobbers(A)                         ; normalize FAC1
+extsub $e097 = RND() clobbers(A,X,Y)                        ; fac1 = RND(fac1) float random number generator
+extsub $e264 = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+extsub $e26b = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+extsub $e2b4 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+extsub $e30e = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
 
 
+asmsub  AYINT2() clobbers(X) -> word @AY {
+    ; fac1-> signed word in AY. Safe wrapper around the AYINT kernal routine (not reading internal memory locations)
+    ; (might throw ILLEGAL QUANTITY)
+    %asm {{
+		jsr  AYINT
+		ldx  #<floats_temp_var
+		ldy  #>floats_temp_var
+		jsr  MOVMF
+		lda  floats_temp_var+4
+		ldy  floats_temp_var+3
+        rts
+    }}
+}
+
 asmsub  FREADS32() clobbers(A,X,Y)  {
 	; ---- fac1 = signed int32 from $62-$65 big endian (MSB FIRST)
 	%asm {{
@@ -104,7 +118,6 @@ asmsub  FREADUS32  () clobbers(A,X,Y)  {
 
 asmsub  FREADS24AXY  (ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) clobbers(A,X,Y)  {
 	; ---- fac1 = signed int24 (A/X/Y contain lo/mid/hi bytes)
-	;      note: there is no FREADU24AXY (unsigned), use FREADUS32 instead.
 	%asm {{
 		sty  $62
 		stx  $63
@@ -119,10 +132,43 @@ asmsub  FREADS24AXY  (ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) clobbers(A,X,Y)
 	}}
 }
 
+asmsub FREADU24AXY(ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) clobbers(A, X, Y) -> float @FAC1 {
+    ; ---- fac1 = unsigned int24 (A/X/Y contain lo/mid/hi bytes)
+        %asm{{
+                 FAC = $61
+                 sty FAC+1
+                 stx FAC+2
+                 sta FAC+3
+
+                 cpy #$00
+                 bne +
+                 cpx #$00
+                 bne +
+                 cmp #$00
+                 beq ++
+
+              +  ldx #$98
+                 bit FAC+1
+                 bmi +
+
+              -  dex
+                 asl FAC+3
+                 rol FAC+2
+                 rol FAC+1
+                 bpl -
+
+              +  stx FAC
+                 lda #$00
+                 sta FAC+4
+                 sta FAC+5
+                 rts
+       }}
+}
+
 asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
 	%asm {{
-		sty  $62
+		sty  $62    ; facmo
 		sta  $63
 		ldx  #$90
 		sec
@@ -151,7 +197,6 @@ asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
 	}}
 }
 
-&uword AYINT_facmo = $64      ; $64/$65 contain result of AYINT
 
 sub rnd() -> float {
     %asm {{
@@ -177,6 +222,13 @@ asmsub normalize(float value @FAC1) -> float @ FAC1 {
     }}
 }
 
+; get the jiffy clock as a float
+asmsub time() -> float @ FAC1 {
+    %asm {{
+        jsr cbm.RDTIM
+        jmp floats.FREADU24AXY
+    }}
+}
 
 %asminclude "library:c64/floats.asm"
 %asminclude "library:c64/floats_funcs.asm"

compiler/res/prog8lib/c64/graphics.p8

@@ -288,17 +288,30 @@ hline_zero2
 
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
         ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
-        ; Midpoint algorithm, filled
+        ; Midpoint algorithm, filled.
         if radius==0
             return
         ubyte @zp yy = 0
         word decisionOver2 = (1 as word)-radius
+        ubyte last_y3 = ycenter+radius
+        ubyte last_y4 = ycenter-radius
+        ubyte new_y3, new_y4
 
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
             horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
-            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
-            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
+
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                horizontal_line(xcenter-yy, last_y3, yy*2+1)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                horizontal_line(xcenter-yy, last_y4, yy*2+1)
+                last_y4 = new_y4
+            }
+
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -307,6 +320,10 @@ hline_zero2
             decisionOver2 += yy*$0002
             decisionOver2++
         }
+        ; draw the final two spans
+        yy--
+        horizontal_line(xcenter-yy, last_y3, yy*2+1)
+        horizontal_line(xcenter-yy, last_y4, yy*2+1)
     }
 
 
@@ -330,12 +347,11 @@ hline_zero2
 
     uword @zp internal_plotx     ; 0..319        ; separate 'parameter' for internal_plot()
 
-    asmsub  internal_plot(ubyte ploty @A) clobbers (A, X, Y) {      ; internal_plotx is 16 bits 0 to 319... doesn't fit in a register
+    asmsub  internal_plot(ubyte ploty @Y) clobbers (A, X, Y) {      ; internal_plotx is 16 bits 0 to 319... doesn't fit in a register
         %asm {{
-        tay
         lda  p8v_internal_plotx+1
         sta  P8ZP_SCRATCH_W2+1
-        lsr  a            ; 0
+        lsr  a      ; make a=0
         sta  P8ZP_SCRATCH_W2
         lda  p8v_internal_plotx
         pha
@@ -360,7 +376,7 @@ hline_zero2
 
 _ormask     .byte 128, 64, 32, 16, 8, 4, 2, 1
 
-; note: this can be even faster if we also have a 256 byte x-lookup table, but hey.
+; note: this can be even faster if we also have a 320 word x-lookup table, but hey, that's a lot of memory.
 ; see http://codebase64.org/doku.php?id=base:various_techniques_to_calculate_adresses_fast_common_screen_formats_for_pixel_graphics
 ; the y lookup tables encodes this formula:  BITMAP_ADDRESS + 320*(py>>3) + (py & 7)    (y from 0..199)
 ; We use the 64tass syntax for range expressions to calculate this table on assembly time.
@@ -369,6 +385,7 @@ _plot_y_values := p8c_BITMAP_ADDRESS + 320*(range(200)>>3) + (range(200) & 7)
 
 _y_lookup_lo    .byte  <_plot_y_values
 _y_lookup_hi    .byte  >_plot_y_values
+            ; !notreached!
 
         }}
     }

compiler/res/prog8lib/c64/syslib.p8

@@ -57,50 +57,50 @@ cbm {
 
 ; ---- CBM ROM kernal routines (C64 addresses) ----
 
-romsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use txt.print instead)
-romsub $E544 = CLEARSCR() clobbers(A,X,Y)                       ; clear the screen
-romsub $E566 = HOMECRSR() clobbers(A,X,Y)                       ; cursor to top left of screen
-romsub $EA31 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
-romsub $EA81 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
-romsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
-romsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
-romsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
-romsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
-romsub $FF8D = VECTOR(uword userptr @ XY, bool dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
-romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
-romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
-romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
-romsub $FF99 = MEMTOP(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
-romsub $FF9C = MEMBOT(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
-romsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
-romsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
-romsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
-romsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
-romsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
-romsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
-romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
-romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
-romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word  (use CLEARST to reset it to 0)
-romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
-romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
-romsub $FFC0 = OPEN() clobbers(X,Y) -> bool @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
-romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
-romsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798 ($31E)) define an input channel
-romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
-romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
-romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
-romsub $FFD2 = CHROUT(ubyte 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
-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)     also see STOP2
-romsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; (via 810 ($32A)) get a character       also see GETIN2
-romsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
-romsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
-romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; get size of text screen into X (columns) and Y (rows)
-romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, bool dir @ Pc) clobbers(A) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
-romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
+extsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use txt.print instead)
+extsub $E544 = CLEARSCR() clobbers(A,X,Y)                       ; clear the screen
+extsub $E566 = HOMECRSR() clobbers(A,X,Y)                       ; cursor to top left of screen
+extsub $EA31 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
+extsub $EA81 = IRQDFEND() clobbers(A,X,Y)                       ; default IRQ end/cleanup
+extsub $FF81 = CINT() clobbers(A,X,Y)                           ; (alias: SCINIT) initialize screen editor and video chip
+extsub $FF84 = IOINIT() clobbers(A, X)                          ; initialize I/O devices (CIA, SID, IRQ)
+extsub $FF87 = RAMTAS() clobbers(A,X,Y)                         ; initialize RAM, tape buffer, screen
+extsub $FF8A = RESTOR() clobbers(A,X,Y)                         ; restore default I/O vectors
+extsub $FF8D = VECTOR(uword userptr @ XY, bool dir @ Pc) clobbers(A,Y)     ; read/set I/O vector table
+extsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
+extsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
+extsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
+extsub $FF99 = MEMTOP(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
+extsub $FF9C = MEMBOT(uword address @ XY, bool dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
+extsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
+extsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
+extsub $FFA5 = ACPTR() -> ubyte @ A                             ; (alias: IECIN) input byte from serial bus
+extsub $FFA8 = CIOUT(ubyte databyte @ A)                        ; (alias: IECOUT) output byte to serial bus
+extsub $FFAB = UNTLK() clobbers(A)                              ; command serial bus device to UNTALK
+extsub $FFAE = UNLSN() clobbers(A)                              ; command serial bus device to UNLISTEN
+extsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
+extsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
+extsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word  (use CLEARST to reset it to 0)
+extsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
+extsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
+extsub $FFC0 = OPEN() clobbers(X,Y) -> bool @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
+extsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
+extsub $FFC6 = CHKIN(ubyte logical @ X) clobbers(A,X) -> bool @Pc    ; (via 798 ($31E)) define an input channel
+extsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320)) define an output channel
+extsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
+extsub $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.
+extsub $FFD2 = CHROUT(ubyte character @ A)                      ; (via 806 ($326)) output a character
+extsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> bool @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+extsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> bool @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
+extsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
+extsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock (A=lo,X=mid,Y=high)
+extsub $FFE1 = STOP() clobbers(X) -> bool @ Pz, ubyte @ A       ; (via 808 ($328)) check the STOP key (and some others in A)     also see STOP2
+extsub $FFE4 = GETIN() clobbers(X,Y) -> bool @Pc, ubyte @ A     ; (via 810 ($32A)) get a character       also see GETIN2
+extsub $FFE7 = CLALL() clobbers(A,X)                            ; (via 812 ($32C)) close all files
+extsub $FFEA = UDTIM() clobbers(A,X)                            ; update the software clock
+extsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; get size of text screen into X (columns) and Y (rows)
+extsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, bool dir @ Pc) clobbers(A) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
+extsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
 
 
 inline asmsub STOP2() clobbers(X,A) -> bool @Pz  {
@@ -307,6 +307,93 @@ c64 {
 
 ; ---- end of SID registers ----
 
+asmsub banks(ubyte banks @A) {
+    ; -- set the memory bank configuration
+    ;    see https://www.c64-wiki.com/wiki/Bank_Switching
+    %asm {{
+        and  #%00000111
+        sta  P8ZP_SCRATCH_REG
+        sei
+        lda  $01
+        and  #%11111000
+        ora  P8ZP_SCRATCH_REG
+        sta  $01
+        cli
+        rts
+    }}
+}
+
+inline asmsub getbanks() -> ubyte @A {
+    ; -- get the current memory bank configuration
+    ;    see https://www.c64-wiki.com/wiki/Bank_Switching
+    %asm {{
+        lda  $01
+        and  #%00000111
+    }}
+}
+
+    asmsub x16jsrfar() {
+        %asm {{
+            ; setup a JSRFAR call (using X16 call convention)
+            sta  P8ZP_SCRATCH_W2        ; save A
+            sty  P8ZP_SCRATCH_W2+1      ; save Y
+            php
+            pla
+            sta  P8ZP_SCRATCH_REG       ; save Status
+
+            pla
+            sta  P8ZP_SCRATCH_W1
+            pla
+            sta  P8ZP_SCRATCH_W1+1
+
+            ; retrieve arguments
+            ldy  #$01
+            lda  (P8ZP_SCRATCH_W1),y            ; grab low byte of target address
+            sta  _jmpfar+1
+            iny
+            lda  (P8ZP_SCRATCH_W1),y            ; now the high byte
+            sta  _jmpfar+2
+            iny
+            lda  (P8ZP_SCRATCH_W1),y            ; then the target bank
+            sta  P8ZP_SCRATCH_B1
+
+            ; adjust return address to skip over the arguments
+            clc
+            lda  P8ZP_SCRATCH_W1
+            adc  #3
+            sta  P8ZP_SCRATCH_W1
+            lda  P8ZP_SCRATCH_W1+1
+            adc  #0
+            pha
+            lda  P8ZP_SCRATCH_W1
+            pha
+            lda  $01        ; save old ram banks
+            pha
+            ; set target bank, restore A, Y and flags
+            lda  P8ZP_SCRATCH_REG
+            pha
+            lda  P8ZP_SCRATCH_B1
+            jsr  banks
+            lda  P8ZP_SCRATCH_W2
+            ldy  P8ZP_SCRATCH_W2+1
+            plp
+            jsr  _jmpfar        ; do the actual call
+            ; restore bank without clobbering status flags and A register
+            sta  P8ZP_SCRATCH_W1
+            php
+            pla
+            sta  P8ZP_SCRATCH_B1
+            pla
+            jsr  banks
+            lda  P8ZP_SCRATCH_B1
+            pha
+            lda  P8ZP_SCRATCH_W1
+            plp
+            rts
+
+_jmpfar     jmp  $0000          ; modified
+        }}
+    }
 
     sub get_vic_memory_base() -> uword {
         ; one of the 4 possible banks. $0000/$4000/$8000/$c000.
@@ -345,85 +432,22 @@ c64 {
 sys {
     ; ------- lowlevel system routines --------
 
-    const ubyte target = 64         ;  compilation target specifier.  64 = C64, 128 = C128, 16 = CommanderX16.
+    const ubyte target = 64         ;  compilation target specifier.  255=virtual, 128=C128, 64=C64, 32=PET, 16=CommanderX16, 8=atari800XL, 7=Neo6502
 
+    const ubyte sizeof_bool = 1
+    const ubyte sizeof_byte = 1
+    const ubyte sizeof_ubyte = 1
+    const ubyte sizeof_word = 2
+    const ubyte sizeof_uword = 2
+    const ubyte sizeof_float = 5
 
-asmsub  init_system()  {
-    ; Initializes the machine to a sane starting state.
-    ; Called automatically by the loader program logic.
-    ; This means that the BASIC, KERNAL and CHARGEN ROMs are banked in,
-    ; the VIC, SID and CIA chips are reset, screen is cleared, and the default IRQ is set.
-    ; Also a different color scheme is chosen to identify ourselves a little.
-    ; Uppercase charset is activated.
-    %asm {{
-        sei
-        cld
-        lda  #%00101111
-        sta  $00
-        lda  #%00100111
-        sta  $01
-        jsr  cbm.IOINIT
-        jsr  cbm.RESTOR
-        jsr  cbm.CINT
-        lda  #6
-        sta  c64.EXTCOL
-        lda  #7
-        sta  cbm.COLOR
-        lda  #0
-        sta  c64.BGCOL0
-        jsr  disable_runstop_and_charsetswitch
-        clc
-        clv
-        cli
-        rts
-    }}
-}
+    sub  disable_runstop_and_charsetswitch() {
+        p8_sys_startup.disable_runstop_and_charsetswitch()
+    }
 
-asmsub  init_system_phase2()  {
-    %asm {{
-        rts     ; no phase 2 steps on the C64
-    }}
-}
-
-asmsub  cleanup_at_exit() {
-    ; executed when the main subroutine does rts
-    %asm {{
-        lda  #31
-        sta  $01        ; bank the kernal in
-        jsr  cbm.CLRCHN		; reset i/o channels
-        jsr  enable_runstop_and_charsetswitch
-_exitcodeCarry = *+1
-        lda  #0
-        lsr  a
-_exitcode = *+1
-        lda  #0        ; exit code possibly modified in exit()
-_exitcodeX = *+1
-        ldx  #0
-_exitcodeY = *+1
-        ldy  #0
-        rts
-    }}
-}
-
-asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
-    %asm {{
-        lda  #$80
-        sta  657    ; disable charset switching
-        lda  #239
-        sta  808    ; disable run/stop key
-        rts
-    }}
-}
-
-asmsub  enable_runstop_and_charsetswitch() clobbers(A) {
-    %asm {{
-        lda  #0
-        sta  657    ; enable charset switching
-        lda  #237
-        sta  808    ; enable run/stop key
-        rts
-    }}
-}
+    sub  enable_runstop_and_charsetswitch() {
+        p8_sys_startup.enable_runstop_and_charsetswitch()
+    }
 
     asmsub save_prog8_internals() {
         %asm {{
@@ -444,6 +468,7 @@ save_SCRATCH_ZPB1	.byte  0
 save_SCRATCH_ZPREG	.byte  0
 save_SCRATCH_ZPWORD1	.word  0
 save_SCRATCH_ZPWORD2	.word  0
+            ; !notreached!
         }}
     }
 
@@ -717,6 +742,65 @@ _longcopy
         }}
     }
 
+    asmsub memcmp(uword address1 @R0, uword address2 @R1, uword size @AY) -> byte @A {
+        ; Compares two blocks of memory
+        ; Returns -1 (255), 0 or 1, meaning: block 1 sorts before, equal or after block 2.
+        %asm {{
+            sta  P8ZP_SCRATCH_REG   ; lsb(size)
+            sty  P8ZP_SCRATCH_B1    ; msb(size)
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            sta  P8ZP_SCRATCH_W2
+            sty  P8ZP_SCRATCH_W2+1
+
+            ldx  P8ZP_SCRATCH_B1
+            beq  _no_msb_size
+
+_loop_msb_size
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dec  P8ZP_SCRATCH_B1        ; msb(size) -= 1
+            dex
+            bne  _loop_msb_size
+
+_no_msb_size
+            lda  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  +
+            rts
+
++           ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  (P8ZP_SCRATCH_W2),y
+            bcs  +
+            lda  #-1
+            rts
++           beq  +
+            lda  #1
+            rts
++           iny
+            cpy  P8ZP_SCRATCH_REG       ; lsb(size)
+            bne  -
+
+            lda #0
+            rts
+        }}
+    }
+
     inline asmsub read_flags() -> ubyte @A {
         %asm {{
         php
@@ -778,37 +862,37 @@ _longcopy
     asmsub exit(ubyte returnvalue @A) {
         ; -- immediately exit the program with a return code in the A register
         %asm {{
-            sta  cleanup_at_exit._exitcode
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit2(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y) {
         ; -- immediately exit the program with result values in the A, X and Y registers.
         %asm {{
-            sta  cleanup_at_exit._exitcode
-            stx  cleanup_at_exit._exitcodeX
-            sty  cleanup_at_exit._exitcodeY
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
     asmsub exit3(ubyte resulta @A, ubyte resultx @X, ubyte resulty @Y, bool carry @Pc) {
         ; -- immediately exit the program with result values in the A, X and Y registers, and the Carry flag in the status register.
         %asm {{
-            sta  cleanup_at_exit._exitcode
+            sta  p8_sys_startup.cleanup_at_exit._exitcode
             lda  #0
             rol  a
-            sta  cleanup_at_exit._exitcodeCarry
-            stx  cleanup_at_exit._exitcodeX
-            sty  cleanup_at_exit._exitcodeY
+            sta  p8_sys_startup.cleanup_at_exit._exitcodeCarry
+            stx  p8_sys_startup.cleanup_at_exit._exitcodeX
+            sty  p8_sys_startup.cleanup_at_exit._exitcodeY
             ldx  prog8_lib.orig_stackpointer
             txs
-            jmp  cleanup_at_exit
+            jmp  p8_sys_startup.cleanup_at_exit
         }}
     }
 
@@ -819,6 +903,13 @@ _longcopy
         }}
     }
 
+    inline asmsub progstart() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_start
+            ldy  #>prog8_program_start
+        }}
+    }
+
     inline asmsub push(ubyte value @A) {
         %asm {{
             pha
@@ -968,6 +1059,7 @@ cx16 {
     _cx16_vreg_storage
             .word 0,0,0,0,0,0,0,0
             .word 0,0,0,0,0,0,0,0
+            ; !notreached!
         }}
     }
 
@@ -988,3 +1080,101 @@ cx16 {
     }
 
 }
+
+p8_sys_startup {
+    ; program startup and shutdown machinery. Needs to reside in normal system ram.
+
+asmsub  init_system()  {
+    ; Initializes the machine to a sane starting state.
+    ; Called automatically by the loader program logic.
+    ; This means that the KERNAL and CHARGEN ROMs are banked in,
+    ; BASIC ROM is NOT banked in (so we have another 8Kb of RAM at our disposal),
+    ; 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.
+    %asm {{
+        sei
+        lda  #%00101111
+        sta  $00
+        lda  #%00100110   ; kernal and i/o banked in, basic off
+        sta  $01
+        jsr  cbm.IOINIT
+        jsr  cbm.RESTOR
+        jsr  cbm.CINT
+        lda  #6
+        sta  c64.EXTCOL
+        lda  #7
+        sta  cbm.COLOR
+        lda  #0
+        sta  c64.BGCOL0
+        jsr  disable_runstop_and_charsetswitch
+        lda  #PROG8_C64_BANK_CONFIG     ; apply bank config
+        sta  $01
+        and  #1
+        bne  +
+        ; basic is not banked in, adjust MEMTOP
+        ldx  #<$d000
+        ldy  #>$d000
+        clc
+        jsr  cbm.MEMTOP
++       cli
+        rts
+    }}
+}
+
+asmsub  init_system_phase2()  {
+    %asm {{
+        cld
+        clc
+        clv
+        rts
+    }}
+}
+
+asmsub  cleanup_at_exit() {
+    ; executed when the main subroutine does rts
+    %asm {{
+        lda  #%00101111
+        sta  $00
+        lda  #31
+        sta  $01            ; bank the kernal and basic in
+        ldx  #<$a000
+        ldy  #>$a000
+        clc
+        jsr  cbm.MEMTOP     ; adjust MEMTOP down again
+        jsr  cbm.CLRCHN		; reset i/o channels
+        jsr  enable_runstop_and_charsetswitch
+_exitcodeCarry = *+1
+        lda  #0
+        lsr  a
+_exitcode = *+1
+        lda  #0        ; exit code possibly modified in exit()
+_exitcodeX = *+1
+        ldx  #0
+_exitcodeY = *+1
+        ldy  #0
+        rts
+    }}
+}
+
+asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #$80
+        sta  657    ; disable charset switching
+        lda  #239
+        sta  808    ; disable run/stop key
+        rts
+    }}
+}
+
+asmsub  enable_runstop_and_charsetswitch() clobbers(A) {
+    %asm {{
+        lda  #0
+        sta  657    ; enable charset switching
+        lda  #237
+        sta  808    ; enable run/stop key
+        rts
+    }}
+}
+
+}

compiler/res/prog8lib/c64/textio.p8

@@ -13,7 +13,7 @@ txt {
 const ubyte DEFAULT_WIDTH = 40
 const ubyte DEFAULT_HEIGHT = 25
 
-romsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
+extsub $FFD2 = chrout(ubyte character @ A)    ; for consistency. You can also use cbm.CHROUT directly ofcourse. Note: takes a PETSCII encoded character.
 
 sub  clear_screen() {
     chrout(147)
@@ -317,6 +317,7 @@ _mod		sta  $ffff		; modified
 		rts
 
 _screenrows	.word  cbm.Screen + range(0, 1000, 40)
+        ; !notreached!
 	}}
 }
 
@@ -360,6 +361,7 @@ _mod		sta  $ffff		; modified
 		rts
 
 _colorrows	.word  $d800 + range(0, 1000, 40)
+        ; !notreached!
 	}}
 }
 

compiler/res/prog8lib/compression.p8

@@ -0,0 +1,409 @@
+; **experimental** data compression/decompression routines, API subject to change!!
+
+compression {
+
+    %option no_symbol_prefixing, ignore_unused
+
+    sub encode_rle_outfunc(uword data, uword size, uword output_function, bool is_last_block) {
+        ; -- Compress the given data block using ByteRun1 aka PackBits RLE encoding.
+        ;    output_function = address of a routine that gets a byte arg in A,
+        ;                      this is the next RLE byte to write to the compressed output buffer or file.
+        ;    is_last_block = usually true, but you can set it to false if you want to concatenate multiple
+        ;                    compressed blocks (for instance if the source data is >64Kb)
+        ;    Worst case result storage size needed = (size + (size+126) / 127) + 1
+        ;    This routine is not optimized for speed but for readability and ease of use.
+        uword idx = 0
+        uword literals_start_idx = 0
+        ubyte literals_length = 0
+
+        asmsub call_output_function(ubyte arg @A) {
+            %asm {{
+                jmp  (p8v_output_function)
+            }}
+        }
+
+        sub next_same_span() {
+            ; returns length in cx16.r1L, and the byte value in cx16.r1H
+            cx16.r1H = data[idx]
+            cx16.r1L = 0
+            while data[idx]==cx16.r1H and cx16.r1L<128 and idx<size {
+                idx++
+                cx16.r1L++
+            }
+        }
+
+        sub output_literals() {
+            call_output_function(literals_length-1)
+            uword dataptr = data + literals_start_idx
+            ubyte i
+            for i in 0 to literals_length-1 {
+                call_output_function(@(dataptr))
+                dataptr++
+            }
+            literals_length = 0
+        }
+
+        while idx<size {
+            next_same_span()     ; count in r1L, value in r1H
+            if cx16.r1L>1 {
+                ; a replicate run
+                if literals_length>0
+                    output_literals()
+                call_output_function((cx16.r1L^255)+2)        ;  257-cx16.r1L
+                call_output_function(cx16.r1H)
+            }
+            else {
+                ; add more to the literals run
+                if literals_length==128
+                    output_literals()
+                if literals_length==0
+                    literals_start_idx = idx-1
+                literals_length++
+            }
+        }
+
+        if literals_length>0
+            output_literals()
+
+        if is_last_block
+            call_output_function(128)
+    }
+
+    sub encode_rle(uword data, uword size, uword target, bool is_last_block) -> uword {
+        ; -- Compress the given data block using ByteRun1 aka PackBits RLE encoding.
+        ;    Returns the size of the compressed RLE data. Worst case result storage size needed = (size + (size+126) / 127) + 1.
+        ;    is_last_block = usually true, but you can set it to false if you want to concatenate multiple
+        ;                    compressed blocks (for instance if the source data is >64Kb)
+        ;    This routine is not optimized for speed but for readability and ease of use.
+        uword idx = 0
+        uword literals_start_idx = 0
+        ubyte literals_length = 0
+        uword orig_target = target
+
+        sub next_same_span() {
+            ; returns length in cx16.r1L, and the byte value in cx16.r1H
+            cx16.r1H = data[idx]
+            cx16.r1L = 0
+            while data[idx]==cx16.r1H and cx16.r1L<128 and idx<size {
+                idx++
+                cx16.r1L++
+            }
+        }
+
+        sub output_literals() {
+            @(target) = literals_length-1
+            target++
+            uword dataptr = data + literals_start_idx
+            ubyte i
+            for i in 0 to literals_length-1 {
+                @(target) = @(dataptr)
+                target++
+                dataptr++
+            }
+            literals_length = 0
+        }
+
+        while idx<size {
+            next_same_span()     ; count in r1L, value in r1H
+            if cx16.r1L>1 {
+                ; a replicate run
+                if literals_length>0
+                    output_literals()
+                @(target) = (cx16.r1L^255)+2        ;  257-cx16.r1L
+                target++
+                @(target) = cx16.r1H
+                target++
+            }
+            else {
+                ; add more to the literals run
+                if literals_length==128
+                    output_literals()
+                if literals_length==0
+                    literals_start_idx = idx-1
+                literals_length++
+            }
+        }
+
+        if literals_length>0
+            output_literals()
+
+        if is_last_block {
+            @(target) = 128
+            target ++
+        }
+
+        return target-orig_target
+    }
+
+    asmsub decode_rle_srcfunc(uword source_function @AY, uword target @R0, uword maxsize @R1) clobbers(X) -> uword @AY {
+        ; -- decodes "ByteRun1" (aka PackBits) RLE compressed data. Control byte value 128 ends the decoding.
+        ;    instead of a source buffer, you provide a callback function that must return the next byte to compress in A.
+        ;    Stops decompressing when the maxsize has been reached.
+        %asm {{
+            sta  _cb_mod1+1
+            sty  _cb_mod1+2
+            sta  _cb_mod2+1
+            sty  _cb_mod2+2
+            sta  _cb_mod3+1
+            sty  _cb_mod3+2
+            lda  cx16.r0L
+            ldy  cx16.r0H
+            sta  P8ZP_SCRATCH_W2        ; target ptr
+            sta  _orig_target
+            sty  P8ZP_SCRATCH_W2+1
+            sty  _orig_target+1
+            lda  cx16.r0L
+            clc
+            adc  cx16.r1L
+            sta  cx16.r1L
+            lda  cx16.r0H
+            adc  cx16.r1H
+            sta  cx16.r1H        ; decompression limit
+
+_loop
+	        ; while target (W2) < limit (r1)
+        	lda  P8ZP_SCRATCH_W2
+	        ldy  P8ZP_SCRATCH_W2+1
+	        cmp  cx16.r1L
+	        tya
+	        sbc  cx16.r1H
+	        bcs  _end
+
+_cb_mod1    jsr  $ffff      ; modified
+            bpl  _literals
+            cmp  #128
+            beq  _end
+
+            ; replicate next byte -n+1 times
+            eor  #255
+            clc
+            adc  #2
+            sta  P8ZP_SCRATCH_REG
+_cb_mod2    jsr  $ffff      ; modified
+            ldx  P8ZP_SCRATCH_REG
+            ldy  #0
+-           sta  (P8ZP_SCRATCH_W2),y
+            iny
+            dex
+            bne  -
+            ; add A to target
+            lda  P8ZP_SCRATCH_REG
+            clc
+            adc  P8ZP_SCRATCH_W2
+            sta  P8ZP_SCRATCH_W2
+            lda  #0
+            adc  P8ZP_SCRATCH_W2+1
+            sta  P8ZP_SCRATCH_W2+1
+            jmp  _loop
+_literals
+            ; copy the next n+1 bytes
+            pha
+            sta  P8ZP_SCRATCH_B1
+            ldy  #0
+            sty  P8ZP_SCRATCH_REG
+_cb_mod3    jsr  $ffff      ; modified
+            ldy  P8ZP_SCRATCH_REG
+            sta  (P8ZP_SCRATCH_W2),y
+            inc  P8ZP_SCRATCH_REG
+            dec  P8ZP_SCRATCH_B1
+            bpl  _cb_mod3
+            ; add N+1 to target
+            pla
+            sec
+            adc  P8ZP_SCRATCH_W2
+            sta  P8ZP_SCRATCH_W2
+            lda  #0
+            adc  P8ZP_SCRATCH_W2+1
+            sta  P8ZP_SCRATCH_W2+1
+            jmp  _loop
+
+_orig_target    .word  0
+_end
+            ; return w2-orig_target, the size of the decompressed data
+            lda  P8ZP_SCRATCH_W2
+            ldy  P8ZP_SCRATCH_W2+1
+            sec
+            sbc  _orig_target
+            tax
+            tya
+            sbc  _orig_target+1
+            tay
+            txa
+            rts
+        }}
+    }
+
+    asmsub decode_rle(uword compressed @AY, uword target @R0, uword maxsize @R1) clobbers(X) -> uword @AY {
+        ; -- decodes "ByteRun1" (aka PackBits) RLE compressed data. Control byte value 128 ends the decoding.
+        ;    Stops decompressing when the maxsize has been reached.
+        %asm {{
+            sta  P8ZP_SCRATCH_W1        ; compressed data ptr
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r0L
+            ldy  cx16.r0H
+            sta  P8ZP_SCRATCH_W2        ; target ptr
+            sta  _orig_target
+            sty  P8ZP_SCRATCH_W2+1
+            sty  _orig_target+1
+            lda  cx16.r0L
+            clc
+            adc  cx16.r1L
+            sta  cx16.r1L
+            lda  cx16.r0H
+            adc  cx16.r1H
+            sta  cx16.r1H        ; decompression limit
+
+_loop       ; while target (W2) < limit (r1)
+        	lda  P8ZP_SCRATCH_W2
+	        ldy  P8ZP_SCRATCH_W2+1
+	        cmp  cx16.r1L
+	        tya
+	        sbc  cx16.r1H
+	        bcs  _end
+
+            ldy  #0
+            lda  (P8ZP_SCRATCH_W1),y
+            bpl  _literals
+            cmp  #128
+            beq  _end
+
+            ; replicate next byte -n+1 times
+            eor  #255
+            clc
+            adc  #2
+            pha
+            tax
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            ldy  #0
+-           sta  (P8ZP_SCRATCH_W2),y
+            iny
+            dex
+            bne  -
+            ; add A to target
+            pla
+            clc
+            adc  P8ZP_SCRATCH_W2
+            sta  P8ZP_SCRATCH_W2
+            lda  #0
+            adc  P8ZP_SCRATCH_W2+1
+            sta  P8ZP_SCRATCH_W2+1
+            ; increase source by 2
+            clc
+            lda  P8ZP_SCRATCH_W1
+            adc  #2
+            sta  P8ZP_SCRATCH_W1
+            lda  #0
+            adc  P8ZP_SCRATCH_W1+1
+            sta  P8ZP_SCRATCH_W1+1
+            jmp  _loop
+_literals
+            ; copy the next n+1 bytes
+            pha
+            tax
+            inc  P8ZP_SCRATCH_W1
+            bne  +
+            inc  P8ZP_SCRATCH_W1+1
++           ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            iny
+            dex
+            bpl  -
+            ; add N+1 to source
+            pla
+            tax
+            sec
+            adc  P8ZP_SCRATCH_W1
+            sta  P8ZP_SCRATCH_W1
+            lda  #0
+            adc  P8ZP_SCRATCH_W1+1
+            sta  P8ZP_SCRATCH_W1+1
+            ; add N+1 to target as well
+            txa
+            sec
+            adc  P8ZP_SCRATCH_W2
+            sta  P8ZP_SCRATCH_W2
+            lda  #0
+            adc  P8ZP_SCRATCH_W2+1
+            sta  P8ZP_SCRATCH_W2+1
+            jmp  _loop
+
+_orig_target    .word  0
+_end
+            ; return w2-orig_target, the size of the decompressed data
+            lda  P8ZP_SCRATCH_W2
+            ldy  P8ZP_SCRATCH_W2+1
+            sec
+            sbc  _orig_target
+            tax
+            tya
+            sbc  _orig_target+1
+            tay
+            txa
+            rts
+        }}
+    }
+
+
+/***
+    ; prog8 source code for the asm routine above:
+
+    sub decode_rle_prog8(uword @zp compressed, uword @zp target, uword maxsize) -> uword {
+        cx16.r0 = target    ; original target
+        cx16.r1 = target+maxsize     ; decompression limit
+
+        while target<cx16.r1 {
+            cx16.r2L = @(compressed)
+            if_neg {
+                if cx16.r2L==128
+                    break
+                ; replicate the next byte -n+1 times
+                compressed++
+                cx16.r3L = @(compressed)
+                repeat 2+(cx16.r2L^255) {
+                    @(target) = cx16.r3L
+                    target++
+                }
+                compressed++
+            } else {
+                ; copy the next n+1 bytes
+                compressed++
+                repeat cx16.r2L+1 {
+                    @(target) = @(compressed)
+                    compressed++
+                    target++
+                }
+            }
+        }
+        return target-cx16.r0
+    }
+
+    sub decode_rle_callback_prog8(uword producer_callback, uword @zp target, uword maxsize) -> uword {
+        cx16.r0 = target   ; original target
+        cx16.r1 = target+maxsize     ; decompression limit
+
+        while target<cx16.r1 {
+            cx16.r2L = lsb(call(producer_callback))
+            if_neg {
+                if cx16.r2L==128
+                    break
+                ; replicate the next byte -n+1 times
+                cx16.r3L = lsb(call(producer_callback))
+                repeat 2+(cx16.r2L^255) {
+                    @(target) = cx16.r3L
+                    target++
+                }
+            } else {
+                ; copy the next n+1 bytes
+                repeat cx16.r2L+1 {
+                    @(target) = lsb(call(producer_callback))
+                    target++
+                }
+            }
+        }
+        return target-cx16.r0
+    }
+***/
+
+}

compiler/res/prog8lib/conv.p8

@@ -200,6 +200,7 @@ _allzero    lda  #'0'
         sta  string_out,x
         inx
         bne  _end
+        ; !notreached!
 	}}
 }
 
@@ -402,6 +403,7 @@ _digit
 		bne  _parse
 		; never reached
 _negative	.byte  0
+        ; !notreached!
 	}}
 }
 
@@ -472,6 +474,7 @@ _try_iso
         bcs  _stop
         and  #63
         bne  _add_letter
+		; !notreached!
 	}}
 }
 
@@ -702,7 +705,7 @@ decHundreds		.byte  0
 decTens			.byte  0
 decOnes   		.byte  0
 			.byte  0		; zero-terminate the decimal output string
-
+		; !notreached!
     }}
 }
 
@@ -736,6 +739,7 @@ asmsub  internal_ubyte2hex  (ubyte value @A) clobbers(X) -> ubyte @A, ubyte @Y
 		rts
 
 _hex_digits	.text "0123456789abcdef"	; can probably be reused for other stuff as well
+		; !notreached!
 	}}
 }
 
@@ -753,6 +757,7 @@ asmsub  internal_uword2hex  (uword value @AY) clobbers(A,Y)  {
 		sty  output+3
 		rts
 output		.text  "0000", $00      ; 0-terminated output buffer (to make printing easier)
+		; !notreached!
 	}}
 }
 

compiler/res/prog8lib/cx16/bmx.p8

@@ -1,4 +1,5 @@
 ; Routines to load and save "BMX" files (commander X16 bitmap format) Version 1.
+; Bitmap data is loaded directly into VRAM without intermediary buffering.
 ; Only uncompressed images are supported for now.
 ; BMX Specification: https://cx16forum.com/forum/viewtopic.php?t=6945
 

compiler/res/prog8lib/cx16/diskio.p8

@@ -145,6 +145,7 @@ io_error:
     uword list_pattern
     uword list_blocks
     bool iteration_in_progress = false
+    bool write_iteration_in_progress = false
     str list_filetype = "???"       ; prg, seq, dir
     str list_filename = "?" * 50
 
@@ -423,6 +424,7 @@ _end        jsr  cbm.READST
 
     sub f_close() {
         ; -- end an iterative file loading session (close channels).
+        ;    it is safe to call this multiple times, or when no file is open for reading.
         if iteration_in_progress {
             cbm.CLRCHN()
             cbm.CLOSE(READ_IO_CHANNEL)
@@ -460,7 +462,10 @@ _end        jsr  cbm.READST
         cbm.SETLFS(WRITE_IO_CHANNEL, drivenumber, WRITE_IO_CHANNEL)
         void cbm.OPEN()             ; open 13,8,13,"filename"
         if_cc {
-            return cbm.READST()==0
+            if cbm.READST()==0 {
+                write_iteration_in_progress = true
+                return true
+            }
         }
         cbm.CLOSE(WRITE_IO_CHANNEL)
         f_close_w()
@@ -498,8 +503,12 @@ no_mciout:
 
     sub f_close_w() {
         ; -- end an iterative file writing session (close channels).
-        cbm.CLRCHN()
-        cbm.CLOSE(WRITE_IO_CHANNEL)
+        ;    it is safe to call this multiple times, or when no file is open for reading.
+        if write_iteration_in_progress {
+            cbm.CLRCHN()
+            cbm.CLOSE(WRITE_IO_CHANNEL)
+            write_iteration_in_progress = false
+        }
     }
 
 
@@ -703,7 +712,7 @@ io_error:
     ; CommanderX16 extensions over the basic C64/C128 diskio routines:
 
     ; For use directly after a load or load_raw call (don't mess with the ram bank yet):
-    ; Calculates the number of bytes loaded (files > 64Kb ar truncated to 16 bits)
+    ; Calculates the number of bytes loaded (files > 64Kb are truncated to 16 bits)
     sub load_size(ubyte startbank, uword startaddress, uword endaddress) -> uword {
         return $2000 * (cx16.getrambank() - startbank) + endaddress - startaddress
     }
@@ -757,6 +766,11 @@ internal_vload:
         }}
     }
 
+    ; note: There is no vsave_raw() routine because the Kernal doesn't have a VSAVE routine.
+    ;       You'll have to write your own loop that reads vram data and use
+    ;       cbm.CHROUT or cx16.MCIOUT to write it to an open output file.
+
+
     sub chdir(str path) {
         ; -- change current directory.
         list_filename[0] = 'c'
@@ -864,6 +878,15 @@ io_error:
         send_command(list_filename)
     }
 
+    sub exists(str filename) -> bool {
+        ; -- returns true if the given file exists on the disk, otherwise false
+        if f_open(filename) {
+            f_close()
+            return true
+        }
+        return false
+    }
+
     sub f_seek(uword pos_hiword, uword pos_loword) {
         ; -- seek in the reading file opened with f_open, to the given 32-bits position
         ;    Note: this will not work if you have already read the last byte of the file! Then you must close and reopen the file first.
@@ -935,15 +958,14 @@ io_error:
             return
 
         cx16.r0 = cx16.r1 = cx16.r2 = cx16.r3 = 0
-
-        sub read4hex() -> uword {
-            str hex = "0000"
-            hex[0] = cbm.CHRIN()
-            hex[1] = cbm.CHRIN()
-            hex[2] = cbm.CHRIN()
-            hex[3] = cbm.CHRIN()
-            return conv.hex2uword(hex)
-        }
     }
 
+    sub read4hex() -> uword {
+        str hex = "0000"
+        hex[0] = cbm.CHRIN()
+        hex[1] = cbm.CHRIN()
+        hex[2] = cbm.CHRIN()
+        hex[3] = cbm.CHRIN()
+        return conv.hex2uword(hex)
+    }
 }

compiler/res/prog8lib/cx16/emudbg.p8

@@ -62,4 +62,25 @@ emudbg {
         if is_emulator()
             EMU_DBGOUT2 = value
     }
+
+    sub reset_cpu_cycles() {
+        if is_emulator()
+            EMU_CPUCLK_L = 0
+    }
+
+    asmsub cpu_cycles() -> uword @R0, uword @R1 {
+        ; -- returns the 32 bits cpu clock counter in R1:R0,
+        ;    returns lower 16 bits in R0, upper 16 bits in R1.
+        %asm {{
+            lda  p8v_EMU_CPUCLK_L
+            sta  cx16.r0L
+            lda  p8v_EMU_CPUCLK_M
+            sta  cx16.r0H
+            lda  p8v_EMU_CPUCLK_H
+            sta  cx16.r1L
+            lda  p8v_EMU_CPUCLK_U
+            sta  cx16.r1H
+            rts
+        }}
+    }
 }

compiler/res/prog8lib/cx16/floats.p8

@@ -14,69 +14,82 @@ floats {
 ; note: fac1/2 might get clobbered even if not mentioned in the function's name.
 ; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
 
-romsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 'facmo' and 'faclo', MSB FIRST. (might throw ILLEGAL QUANTITY)  See "basic.sym" kernal symbol file for their memory locations.
+extsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 'facmo' and 'faclo', MSB FIRST. DON'T USE THIS, USE WRAPPER 'AYINT2' INSTEAD.  (might throw ILLEGAL QUANTITY)
 
 ; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1
 ; there is also floats.GIVUAYFAY - unsigned word in A/Y (lo/hi) to fac1
 ; (tip: use GIVAYFAY to use A/Y input; lo/hi switched to normal order)
-romsub $fe03 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
+extsub $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(uword string @XY, ubyte length @A) clobbers(A,X,Y) -> float @FAC1      ; convert ASCII string in XY and length in A, to floating point in FAC1. WARNING: only implemented in ROM 47+. Safer to use floats.parse() instead.
+extsub $fe06 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
+extsub $fe09 = VAL_1(uword string @XY, ubyte length @A) clobbers(A,X,Y) -> float @FAC1      ; convert ASCII string in XY and length in A, to floating point in FAC1. WARNING: only implemented in ROM 47+. Safer to use floats.parse() 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)
-romsub $fe0c = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
-romsub $fe0f = FLOATC() clobbers(A,X,Y)                     ; convert address to floating point
+extsub $fe0c = GETADR() clobbers(X) -> ubyte @ Y, ubyte @ A
+extsub $fe0f = FLOATC() clobbers(A,X,Y)                     ; convert address to floating point
 
-romsub $fe12 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
-romsub $fe15 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
-romsub $fe18 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
-romsub $fe1b = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
-romsub $fe1e = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
-romsub $fe21 = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
-romsub $fe24 = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1
-romsub $fe27 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  mind the order of the operands
-romsub $fe2a = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
-romsub $fe2d = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to integer round instead of trunc
-romsub $fe30 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
-romsub $fe33 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
-romsub $fe36 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** float in A/Y
-romsub $fe39 = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
-romsub $fe3c = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
-romsub $fe3f = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
-romsub $fe42 = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
-romsub $fe45 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
-romsub $fe48 = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
-romsub $fe4b = ROUND() clobbers(A,X,Y)                      ; round least significant bit of fac1
-romsub $fe4e = ABS() clobbers(A,X,Y)                        ; fac1 = ABS(fac1)
-romsub $fe51 = SIGN() clobbers(X,Y) -> ubyte @ A            ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
-romsub $fe54 = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
-romsub $fe57 = RND_0() clobbers(A,X,Y)                      ; fac1 = RND(fac1) float random number generator  NOTE: incompatible with C64's RND routine
-romsub $fe57 = RND() clobbers(A,X,Y)                        ; alias for RND_0
-romsub $fe5a = CONUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in A/Y into fac2
-romsub $fe5d = ROMUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in current bank in A/Y into fac2
-romsub $fe60 = MOVFRM(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in A/Y into fac1  (use MOVFM instead)
-romsub $fe63 = MOVFM(uword mflpt @ AY) clobbers(A,X,Y)      ; load mflpt value from memory in A/Y into fac1
-romsub $fe66 = MOVMF(uword mflpt @ XY) clobbers(A,X,Y)      ; store fac1 to memory  X/Y as 5-byte mflpt
-romsub $fe69 = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
-romsub $fe6c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded the least significant bit)
+extsub $fe12 = FSUB(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt from A/Y - fac1
+extsub $fe15 = FSUBT() clobbers(A,X,Y)                      ; fac1 = fac2-fac1   mind the order of the operands
+extsub $fe18 = FADD(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 += mflpt value from A/Y
+extsub $fe1b = FADDT() clobbers(A,X,Y)                      ; fac1 += fac2
+extsub $fe1e = FMULT(uword mflpt @ AY) clobbers(A,X,Y)      ; fac1 *= mflpt value from A/Y
+extsub $fe21 = FMULTT() clobbers(A,X,Y)                     ; fac1 *= fac2
+extsub $fe24 = FDIV(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = mflpt in A/Y / fac1
+extsub $fe27 = FDIVT() clobbers(A,X,Y)                      ; fac1 = fac2/fac1  mind the order of the operands
+extsub $fe2a = LOG() clobbers(A,X,Y)                        ; fac1 = LN(fac1)  (natural log)
+extsub $fe2d = INT() clobbers(A,X,Y)                        ; INT() truncates, use FADDH first to integer round instead of trunc
+extsub $fe30 = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
+extsub $fe33 = NEGOP() clobbers(A)                          ; switch the sign of fac1 (fac1 = -fac1)
+extsub $fe36 = FPWR(uword mflpt @ AY) clobbers(A,X,Y)       ; fac1 = fac2 ** float in A/Y
+extsub $fe39 = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+extsub $fe3c = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
+extsub $fe3f = COS() clobbers(A,X,Y)                        ; fac1 = COS(fac1)
+extsub $fe42 = SIN() clobbers(A,X,Y)                        ; fac1 = SIN(fac1)
+extsub $fe45 = TAN() clobbers(A,X,Y)                        ; fac1 = TAN(fac1)
+extsub $fe48 = ATN() clobbers(A,X,Y)                        ; fac1 = ATN(fac1)
+extsub $fe4b = ROUND() clobbers(A,X,Y)                      ; round least significant bit of fac1
+extsub $fe4e = ABS() clobbers(A,X,Y)                        ; fac1 = ABS(fac1)
+extsub $fe51 = SIGN() clobbers(X,Y) -> ubyte @ A            ; SIGN(fac1) to A, $ff, $0, $1 for negative, zero, positive
+extsub $fe54 = FCOMP(uword mflpt @ AY) clobbers(X,Y) -> ubyte @ A   ; A = compare fac1 to mflpt in A/Y, 0=equal 1=fac1 is greater, 255=fac1 is less than
+extsub $fe57 = RND_0() clobbers(A,X,Y)                      ; fac1 = RND(fac1) float random number generator  NOTE: incompatible with C64's RND routine
+extsub $fe57 = RND() clobbers(A,X,Y)                        ; alias for RND_0
+extsub $fe5a = CONUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in A/Y into fac2
+extsub $fe5d = ROMUPK(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in current bank in A/Y into fac2
+extsub $fe60 = MOVFRM(uword mflpt @ AY) clobbers(A,X,Y)     ; load mflpt value from memory in A/Y into fac1  (use MOVFM instead)
+extsub $fe63 = MOVFM(uword mflpt @ AY) clobbers(A,X,Y)      ; load mflpt value from memory in A/Y into fac1
+extsub $fe66 = MOVMF(uword mflpt @ XY) clobbers(A,X,Y)      ; store fac1 to memory  X/Y as 5-byte mflpt
+extsub $fe69 = MOVFA() clobbers(A,X)                        ; copy fac2 to fac1
+extsub $fe6c = MOVAF() clobbers(A,X)                        ; copy fac1 to fac2  (rounded the least significant bit)
 
 ; X16 additions
-romsub $fe6f = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for integer rounding- call this before INT
-romsub $fe72 = ZEROFC() clobbers(A,X,Y)                     ; fac1 = 0
-romsub $fe75 = NORMAL() clobbers(A,X,Y)                     ; normalize fac1
-romsub $fe78 = NEGFAC() clobbers(A)                         ; switch the sign of fac1 (fac1 = -fac1) (doesn't work, juse use NEGOP() instead!)
-romsub $fe7b = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
-romsub $fe7e = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive! Have to restore sign manually!
-romsub $fe81 = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
-romsub $fe84 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
-romsub $fe87 = FLOAT() clobbers(A,X,Y)                      ; FAC = (s8).A
-romsub $fe8a = FLOATS() clobbers(A,X,Y)                     ; FAC = (s16)facho+1:facho
-romsub $fe8d = QINT() clobbers(A,X,Y)                       ; facho:facho+1:facho+2:facho+3 = u32(FAC)
-romsub $fe90 = FINLOG(byte value @A) clobbers (A, X, Y)     ; fac1 += signed byte in A
+extsub $fe6f = FADDH() clobbers(A,X,Y)                      ; fac1 += 0.5, for integer rounding- call this before INT
+extsub $fe72 = ZEROFC() clobbers(A,X,Y)                     ; fac1 = 0
+extsub $fe75 = NORMAL() clobbers(A,X,Y)                     ; normalize fac1
+extsub $fe78 = NEGFAC() clobbers(A)                         ; switch the sign of fac1 (fac1 = -fac1) (doesn't work, juse use NEGOP() instead!)
+extsub $fe7b = MUL10() clobbers(A,X,Y)                      ; fac1 *= 10
+extsub $fe7e = DIV10() clobbers(A,X,Y)                      ; fac1 /= 10 , CAUTION: result is always positive! Have to restore sign manually!
+extsub $fe81 = MOVEF() clobbers(A,X)                        ; copy fac1 to fac2
+extsub $fe84 = SGN() clobbers(A,X,Y)                        ; fac1 = SGN(fac1), result of SIGN (-1, 0 or 1)
+extsub $fe87 = FLOAT() clobbers(A,X,Y)                      ; FAC = (s8).A
+extsub $fe8a = FLOATS() clobbers(A,X,Y)                     ; FAC = (s16)facho+1:facho
+extsub $fe8d = QINT() clobbers(A,X,Y)                       ; facho:facho+1:facho+2:facho+3 = u32(FAC)
+extsub $fe90 = FINLOG(byte value @A) clobbers (A, X, Y)     ; fac1 += signed byte in A
 
 
+asmsub  AYINT2() clobbers(X) -> word @AY {
+    ; fac1-> signed word in AY. Safe wrapper around the AYINT kernal routine (not reading internal memory locations)
+    ; (might throw ILLEGAL QUANTITY)
+    %asm {{
+		jsr  AYINT
+		ldx  #<floats_temp_var
+		ldy  #>floats_temp_var
+		jsr  MOVMF
+		lda  floats_temp_var+4
+		ldy  floats_temp_var+3
+        rts
+    }}
+}
 
 asmsub  FREADSA  (byte value @A) clobbers(A,X,Y) {
     ; ---- 8 bit signed A -> float in fac1
@@ -90,8 +103,42 @@ asmsub  FREADSA  (byte value @A) clobbers(A,X,Y) {
     }}
 }
 
+asmsub FREADU24AXY(ubyte lo @ A, ubyte mid @ X, ubyte hi @ Y) clobbers(A, X, Y) -> float @FAC1 {
+    ; ---- fac1 = unsigned int24 (A/X/Y contain lo/mid/hi bytes)
+        %asm{{
+                 FAC = $C3
+                 sty FAC+1
+                 stx FAC+2
+                 sta FAC+3
+
+                 cpy #$00
+                 bne +
+                 cpx #$00
+                 bne +
+                 cmp #$00
+                 beq ++
+
+              +  ldx #$98
+                 bit FAC+1
+                 bmi +
+
+              -  dex
+                 asl FAC+3
+                 rol FAC+2
+                 rol FAC+1
+                 bpl -
+
+               + stx FAC
+                 stz FAC+4
+                 stz FAC+5
+                 rts
+       }}
+}
+
 asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
+	; See "basic.sym" kernal symbol file for the facmo memory location.
+	; TODO find a way to not depend on that internal fac memory location. MOVFM doesn't work.
 	%asm {{
 	sty  $c4        ; facmo
 	sta  $c5        ; facmo+1
@@ -154,12 +201,11 @@ _borked
 +       jmp  sys.exit
 
 _msg    .text 13,"?rom 47+ required for val1",13,0
+        ; !notreached!
     }}
 }
 
 
-&uword AYINT_facmo = $c6      ; $c6/$c7 contain result of AYINT   (See "basic.sym" kernal symbol file)
-
 sub rnd() -> float {
     %asm {{
         lda  #1
@@ -173,6 +219,14 @@ asmsub normalize(float value @FAC1) -> float @ FAC1 {
     }}
 }
 
+; get the jiffy clock as a float
+asmsub time() -> float @ FAC1 {
+    %asm {{
+        jsr cbm.RDTIM
+        jmp floats.FREADU24AXY
+    }}
+}
+
 %asminclude "library:c64/floats.asm"
 %asminclude "library:c64/floats_funcs.asm"
 

compiler/res/prog8lib/cx16/gfx_hires.p8

@@ -1,114 +1,49 @@
 ; Bitmap pixel graphics routines for the CommanderX16
-; Custom routines to use the full-screen 640x480 and 320x240 screen modes.
+; Custom routines to use the full-screen 640x480 in 4 color mode.
+; (monochrome mode can be used via the monogfx module)
 ; (These modes are not supported by the documented GRAPH_xxxx kernal routines)
 ;
 ; No text layer is currently shown, text can be drawn as part of the bitmap itself.
 ; Note: for similar graphics routines that also work on the C-64, use the "graphics" module instead.
 ; Note: for identical routines for a monochrome 1 bpp screen, use the "monogfx" module instead.
 ; Note: for color palette manipulation, use the "palette" module or write Vera registers yourself.
-; Note: this library implements code for various resolutions and color depths. This takes up memory.
-;       If you're memory constrained you should probably not use this built-in library,
-;       but make a copy in your project only containing the code for the required resolution.
 ;
-; 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!
+; NOTE: the bitmap screen data is positioned in vram at $00000
+; This is compatible with the CX16's screen mode 128.  (void cx16.set_screen_mode(128))
 ;
-; NOTE: the bitmap screen data is positioned in vram at $0:0000
-;
-; SCREEN MODE LIST:
-;   mode 0 = reset back to default text mode
-;   mode 1 = bitmap 320 x 240 x 256c (8 bpp)
-;   mode 2 = bitmap 640 x 480 x 4c (2 bpp)
-;   mode 3 = bitmap 320 x 240 x 4c (not yet implemented: just use 256c, there's enough vram for that)
-;   mode 4 = bitmap 320 x 240 x 16c (not yet implemented: just use 256c, there's enough vram for that)
-;   mode 5 = bitmap 640 x 400 x 16c (not yet implemented)
 
-gfx2 {
+gfx_hires {
 
     %option ignore_unused
 
-    ; read-only control variables:
-    ubyte active_mode = 0
-    uword width = 0
-    uword height = 0
-    ubyte bpp = 0
+    const uword WIDTH = 640
+    const uword HEIGHT = 480
 
-    sub screen_mode(ubyte mode) {
+    sub graphics_mode() {
+        ; set hires 4c mode
         cx16.VERA_CTRL=0
-        when mode {
-            1 -> {
-                ; lores 256c
-                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
-                cx16.VERA_DC_HSCALE = 64
-                cx16.VERA_DC_VSCALE = 64
-                cx16.VERA_L1_CONFIG = %00000111
-                cx16.VERA_L1_MAPBASE = 0
-                cx16.VERA_L1_TILEBASE = 0
-            }
-            2 -> {
-                ; highres 4c
-                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
-                cx16.VERA_DC_HSCALE = 128
-                cx16.VERA_DC_VSCALE = 128
-                cx16.VERA_L1_CONFIG = %00000101
-                cx16.VERA_L1_MAPBASE = 0
-                cx16.VERA_L1_TILEBASE = %00000001
-            }
-            else -> {
-                ; back to default text mode
-                if active_mode!=0 {
-                    cx16.r15L = cx16.VERA_DC_VIDEO & %00000111 ; retain chroma + output mode
-                    cbm.CINT()
-                    cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11111000) | cx16.r15L
-                }
-            }
-        }
-
-        init_mode(mode)
-        if active_mode!=0
-            clear_screen(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 = %00000101
+        cx16.VERA_L1_MAPBASE = 0
+        cx16.VERA_L1_TILEBASE = %00000001
+        clear_screen(0)
     }
 
-    sub init_mode(ubyte mode) {
-        ; set the internal configuration variables corresponding to the given screenmode
-        ; doesn't manipulate Vera / the actual display mode
-        active_mode = mode
-        when mode {
-            1 -> {
-                width = 320
-                height = 240
-                bpp = 8
-            }
-            2 -> {
-                width = 640
-                height = 480
-                bpp = 2
-            }
-            else -> {
-                width = 0
-                height = 0
-                bpp = 0
-                active_mode = 0
-            }
-        }
+    sub text_mode() {
+        ; 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) {
         position(0, 0)
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                repeat 240/2
-                    cs_innerloop640(color)
-            }
-            2 -> {
-                ; highres 4c
-                ubyte[] colors = [%00000000, %01010101, %10101010, %11111111]
-                color = colors[color&3]
-                repeat 480/4
-                    cs_innerloop640(color)
-            }
-        }
+        ubyte[] colors = [%00000000, %01010101, %10101010, %11111111]
+        color = colors[color&3]
+        repeat 480/4
+            cs_innerloop640(color)
         position(0, 0)
     }
 
@@ -125,6 +60,18 @@ gfx2 {
         vertical_line(xx+rwidth-1, yy+1, rheight-2, color)
     }
 
+    sub safe_rect(uword xx, uword yy, uword rwidth, uword rheight, ubyte color) {
+        ; does bounds checking and clipping
+        safe_horizontal_line(xx, yy, rwidth, color)
+        if rheight==1
+            return
+        safe_horizontal_line(xx, yy+rheight-1, rwidth, color)
+        safe_vertical_line(xx, yy+1, rheight-2, color)
+        if rwidth==1
+            return
+        safe_vertical_line(xx+rwidth-1, yy+1, rheight-2, color)
+    }
+
     sub fillrect(uword xx, uword yy, uword rwidth, uword rheight, ubyte color) {
         ; Draw a filled rectangle of the given size and color.
         ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
@@ -136,140 +83,124 @@ gfx2 {
         }
     }
 
+    sub safe_fillrect(uword xx, uword yy, uword rwidth, uword rheight, ubyte color) {
+        ; Draw a filled rectangle of the given size and color.
+        ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
+        ; This safe version does bounds checking and clipping.
+        if xx>=WIDTH or yy>=WIDTH
+            return
+        if msb(xx)&$80!=0 {
+            rwidth += xx
+            xx = 0
+        }
+        if xx>=WIDTH
+            return
+        if xx+rwidth>WIDTH
+            rwidth = WIDTH-xx
+        if rwidth>WIDTH
+            return
+
+        if yy + rheight > HEIGHT
+            rheight = HEIGHT-yy
+        if rheight>HEIGHT
+            return
+
+        repeat rheight {
+            horizontal_line(xx, yy, rwidth, color)
+            yy++
+        }
+    }
+
     sub horizontal_line(uword xx, uword yy, uword length, ubyte color) {
         if length==0
             return
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                position(xx, yy)
-                %asm {{
-                    lda  p8v_color
-                    ldx  p8v_length+1
-                    beq  +
-                    ldy  #0
--                   sta  cx16.VERA_DATA0
-                    iny
-                    bne  -
-                    dex
-                    bne  -
-+                   ldy  p8v_length     ; remaining
-                    beq  +
--                   sta  cx16.VERA_DATA0
-                    dey
-                    bne  -
-+
-                }}
-            }
-            2 -> {
-                ; highres 4c ....also mostly usable for lores 4c?
-                color &= 3
-                ubyte[4] colorbits
-                ubyte ii
-                for ii in 3 downto 0 {
-                    colorbits[ii] = color
-                    color <<= 2
-                }
-                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-                %asm {{
-                    lda  cx16.VERA_ADDR_H
-                    and  #%00000111         ; no auto advance
-                    sta  cx16.VERA_ADDR_H
-                    stz  cx16.VERA_CTRL     ; setup vera addr 0
-                    lda  cx16.r1
-                    and  #1
-                    sta  cx16.VERA_ADDR_H
-                    lda  cx16.r0
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.r0+1
-                    sta  cx16.VERA_ADDR_M
-                    ldx  p8v_xx
-                }}
+        ; highres 4c ....also mostly usable for lores 4c?
+        color &= 3
+        ubyte[4] colorbits
+        ubyte ii
+        for ii in 3 downto 0 {
+            colorbits[ii] = color
+            color <<= 2
+        }
+        void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+        %asm {{
+            lda  cx16.VERA_ADDR_H
+            and  #%00000111         ; no auto advance
+            sta  cx16.VERA_ADDR_H
+            stz  cx16.VERA_CTRL     ; setup vera addr 0
+            lda  cx16.r1
+            and  #1
+            sta  cx16.VERA_ADDR_H
+            lda  cx16.r0
+            sta  cx16.VERA_ADDR_L
+            lda  cx16.r0+1
+            sta  cx16.VERA_ADDR_M
+            ldx  p8v_xx
+        }}
 
-                repeat length {
-                    %asm {{
-                        txa
-                        and  #3
-                        tay
-                        lda  cx16.VERA_DATA0
-                        and  p8b_gfx2.p8s_plot.p8v_mask4c,y
-                        ora  p8v_colorbits,y
-                        sta  cx16.VERA_DATA0
-                        cpy  #%00000011         ; next vera byte?
-                        bne  ++
-                        inc  cx16.VERA_ADDR_L
-                        bne  ++
-                        inc  cx16.VERA_ADDR_M
+        repeat length {
+            %asm {{
+                txa
+                and  #3
+                tay
+                lda  cx16.VERA_DATA0
+                and  p8b_gfx_hires.p8s_plot.p8v_mask4c,y
+                ora  p8v_colorbits,y
+                sta  cx16.VERA_DATA0
+                cpy  #%00000011         ; next vera byte?
+                bne  ++
+                inc  cx16.VERA_ADDR_L
+                bne  ++
+                inc  cx16.VERA_ADDR_M
 +                       bne  +
-                        inc  cx16.VERA_ADDR_H
+                inc  cx16.VERA_ADDR_H
 +                       inx                     ; next pixel
-                    }}
-                }
-            }
+            }}
         }
     }
 
     sub safe_horizontal_line(uword xx, uword yy, uword length, ubyte color) {
         ; does bounds checking and clipping
-        if msb(yy)&$80!=0 or yy>=height
+        if msb(yy)&$80!=0 or yy>=HEIGHT
             return
         if msb(xx)&$80!=0 {
             length += xx
             xx = 0
         }
-        if xx>=width
+        if xx>=WIDTH
             return
-        if xx+length>width
-            length = width-xx
-        if length>width
+        if xx+length>WIDTH
+            length = WIDTH-xx
+        if length>WIDTH
             return
 
         horizontal_line(xx, yy, length, color)
     }
 
     sub vertical_line(uword xx, uword yy, uword lheight, ubyte color) {
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                ; set vera auto-increment to 320 pixel increment (=next line)
-                position(xx,yy)
-                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
-                %asm {{
-                    ldy  p8v_lheight
-                    beq  +
-                    lda  p8v_color
--                   sta  cx16.VERA_DATA0
-                    dey
-                    bne  -
-+
-                }}
-            }
-            2 -> {
-                ; highres 4c
-                ; use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment is possible
-                if lheight==0
-                    return
-                position2(xx,yy,true)
-                set_both_strides(13)    ; 160 increment = 1 line in 640 px 4c mode
-                ;; color &= 3
-                ;; color <<= gfx2.plot.shift4c[lsb(xx) & 3]
-                cx16.r2L = lsb(xx) & 3
-                when color & 3 {
-                    0 -> color = 0
-                    1 -> color = gfx2.plot.shiftedleft_4c_1[cx16.r2L]
-                    2 -> color = gfx2.plot.shiftedleft_4c_2[cx16.r2L]
-                    3 -> color = gfx2.plot.shiftedleft_4c_3[cx16.r2L]
-                }
-                ubyte @shared mask = gfx2.plot.mask4c[lsb(xx) & 3]
-                repeat lheight {
-                    %asm {{
-                        lda  cx16.VERA_DATA0
-                        and  p8v_mask
-                        ora  p8v_color
-                        sta  cx16.VERA_DATA1
-                    }}
-                }
-            }
+        ; highres 4c
+        ; use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment is possible
+        if lheight==0
+            return
+        position2(xx,yy,true)
+        set_both_strides(13)    ; 160 increment = 1 line in 640 px 4c mode
+        ;; color &= 3
+        ;; color <<= plot.shift4c[lsb(xx) & 3]
+        cx16.r2L = lsb(xx) & 3
+        when color & 3 {
+            0 -> color = 0
+            1 -> color = gfx_hires.plot.shiftedleft_4c_1[cx16.r2L]
+            2 -> color = gfx_hires.plot.shiftedleft_4c_2[cx16.r2L]
+            3 -> color = gfx_hires.plot.shiftedleft_4c_3[cx16.r2L]
+        }
+        ubyte @shared mask = gfx_hires.plot.mask4c[lsb(xx) & 3]
+        repeat lheight {
+            %asm {{
+                lda  cx16.VERA_DATA0
+                and  p8v_mask
+                ora  p8v_color
+                sta  cx16.VERA_DATA1
+            }}
         }
 
         sub set_both_strides(ubyte stride) {
@@ -282,9 +213,30 @@ gfx2 {
 
     }
 
+    sub safe_vertical_line(uword xx, uword yy, uword lheight, ubyte color) {
+        ; does bounds checking and clipping
+        if msb(yy)&$80!=0 or yy>=HEIGHT
+            return
+        if msb(xx)&$80!=0 or xx>=WIDTH
+            return
+        if msb(yy)&$80!=0 {
+            lheight += yy
+            yy = 0
+        }
+        if yy+lheight>HEIGHT
+            lheight = HEIGHT-yy
+        if lheight>HEIGHT
+            return
+
+        vertical_line(xx, yy, lheight, color)
+    }
+
     sub line(uword @zp x1, uword @zp y1, uword @zp x2, uword @zp y2, ubyte color) {
         ; Bresenham algorithm.
         ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+
+        ; NOTE: if you are using lores 256c (mode 1), the line routine in the fastline module can be used and is MUCH faster than this one.
+
         if y1>y2 {
             ; make sure dy is always positive to have only 4 instead of 8 special cases
             cx16.r0 = x1
@@ -479,12 +431,25 @@ gfx2 {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
+        uword last_y3 = ycenter+radius
+        uword last_y4 = ycenter-radius
+        uword new_y3, new_y4
 
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
-            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
-            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
+
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
+                last_y4 = new_y4
+            }
+
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -493,6 +458,11 @@ gfx2 {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
+
+        ; draw the final two spans
+        yy--
+        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
@@ -502,12 +472,23 @@ gfx2 {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
+        uword last_y3 = ycenter+radius
+        uword last_y4 = ycenter-radius
+        uword new_y3, new_y4
 
         while radius>=yy {
             safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
             safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
-            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
-            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
+                last_y4 = new_y4
+            }
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -516,6 +497,10 @@ gfx2 {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
+        ; draw the final two spans
+        yy--
+        safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+        safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
     }
 
     sub plot(uword @zp xx, uword @zp yy, ubyte @zp color) {
@@ -525,107 +510,68 @@ gfx2 {
         ubyte[4] shiftedleft_4c_2 = [2<<6, 2<<4, 2<<2, 2<<0]
         ubyte[4] shiftedleft_4c_3 = [3<<6, 3<<4, 3<<2, 3<<0]
 
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-                %asm {{
-                    stz  cx16.VERA_CTRL
-                    lda  cx16.r1
-                    ora  #%00010000     ; enable auto-increment so next_pixel() can be used after this
-                    sta  cx16.VERA_ADDR_H
-                    lda  cx16.r0+1
-                    sta  cx16.VERA_ADDR_M
-                    lda  cx16.r0
-                    sta  cx16.VERA_ADDR_L
-                    lda  p8v_color
-                    sta  cx16.VERA_DATA0
-                }}
-            }
-            2 -> {
-                ; highres 4c   ....also mostly usable for lores 4c?
-                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-                cx16.r2L = lsb(xx) & 3       ; xbits
-                ; color &= 3
-                ; color <<= shift4c[cx16.r2L]
-                when color & 3 {
-                    0 -> color = 0
-                    1 -> color = shiftedleft_4c_1[cx16.r2L]
-                    2 -> color = shiftedleft_4c_2[cx16.r2L]
-                    3 -> color = shiftedleft_4c_3[cx16.r2L]
-                }
-                %asm {{
-                    stz  cx16.VERA_CTRL
-                    lda  cx16.r1L
-                    sta  cx16.VERA_ADDR_H
-                    lda  cx16.r0H
-                    sta  cx16.VERA_ADDR_M
-                    lda  cx16.r0L
-                    sta  cx16.VERA_ADDR_L
-                    ldy  cx16.r2L           ; xbits
-                    lda  p8v_mask4c,y
-                    and  cx16.VERA_DATA0
-                    ora  p8v_color
-                    sta  cx16.VERA_DATA0
-                }}
-            }
+        ; highres 4c   ....also mostly usable for lores 4c?
+        void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+        cx16.r2L = lsb(xx) & 3       ; xbits
+        ; color &= 3
+        ; color <<= shift4c[cx16.r2L]
+        when color & 3 {
+            0 -> color = 0
+            1 -> color = shiftedleft_4c_1[cx16.r2L]
+            2 -> color = shiftedleft_4c_2[cx16.r2L]
+            3 -> color = shiftedleft_4c_3[cx16.r2L]
         }
+        %asm {{
+            stz  cx16.VERA_CTRL
+            lda  cx16.r1L
+            sta  cx16.VERA_ADDR_H
+            lda  cx16.r0H
+            sta  cx16.VERA_ADDR_M
+            lda  cx16.r0L
+            sta  cx16.VERA_ADDR_L
+            ldy  cx16.r2L           ; xbits
+            lda  p8v_mask4c,y
+            and  cx16.VERA_DATA0
+            ora  p8v_color
+            sta  cx16.VERA_DATA0
+        }}
     }
 
     sub safe_plot(uword xx, uword yy, ubyte color) {
         ; 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
+        if xx >= WIDTH or yy >= HEIGHT
             return
         plot(xx, yy, color)
     }
 
     sub pget(uword @zp xx, uword yy) -> ubyte {
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-                %asm {{
-                    stz  cx16.VERA_CTRL
-                    lda  cx16.r1
-                    sta  cx16.VERA_ADDR_H
-                    lda  cx16.r0+1
-                    sta  cx16.VERA_ADDR_M
-                    lda  cx16.r0
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.VERA_DATA0
-                }}
-            }
-            2 -> {
-                ; hires 4c
-                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-                %asm {{
-                    stz  cx16.VERA_CTRL
-                    lda  cx16.r1L
-                    sta  cx16.VERA_ADDR_H
-                    lda  cx16.r0H
-                    sta  cx16.VERA_ADDR_M
-                    lda  cx16.r0L
-                    sta  cx16.VERA_ADDR_L
-                    lda  cx16.VERA_DATA0
-                    pha
-                    lda  p8v_xx
-                    and  #3
-                    tay
-                    lda  p8b_gfx2.p8s_plot.p8v_shift4c,y
-                    tay
-                    pla
-                    cpy  #0
-                    beq  +
--                   lsr  a
-                    dey
-                    bne  -
-+                   and #3
-                }}
-            }
-            else -> return 0
-        }
+        void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
+        %asm {{
+            stz  cx16.VERA_CTRL
+            lda  cx16.r1L
+            sta  cx16.VERA_ADDR_H
+            lda  cx16.r0H
+            sta  cx16.VERA_ADDR_M
+            lda  cx16.r0L
+            sta  cx16.VERA_ADDR_L
+            lda  cx16.VERA_DATA0
+            pha
+            lda  p8v_xx
+            and  #3
+            tay
+            lda  p8b_gfx_hires.p8s_plot.p8v_shift4c,y
+            tay
+            pla
+            cpy  #0
+            beq  +
+-           lsr  a
+            dey
+            bne  -
++           and #3
+            rts
+        }}
     }
 
     sub fill(uword x, uword y, ubyte new_color) {
@@ -657,7 +603,7 @@ gfx2 {
             if cx16.r12L==MAXDEPTH
                 return
             cx16.r0s = sy+sdy
-            if cx16.r0s>=0 and cx16.r0s<=height-1 {
+            if cx16.r0s>=0 and cx16.r0s<=HEIGHT-1 {
 ;;                stack_xl[cx16.r12L] = sxl
 ;;                stack_xr[cx16.r12L] = sxr
 ;;                stack_y[cx16.r12L] = sy
@@ -714,29 +660,23 @@ gfx2 {
         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
+        if xx<0 or xx>WIDTH-1 or yy<0 or yy>HEIGHT-1
             return
-        if gfx2.active_mode == 2 set_color_masks()
+        set_color_masks()
         push_stack(xx, xx, yy, 1)
         push_stack(xx, xx, yy + 1, -1)
         word left = 0
         while cx16.r12L!=0 {
             pop_stack()
             xx = x1
-            when active_mode {
-                1 -> if fill_scanline_left_8bpp() goto skip
-                2 -> if fill_scanline_left_2bpp() goto skip
-            }
+            if fill_scanline_left_2bpp() goto skip
             left = xx + 1
             if left < x1
                 push_stack(left, x1 - 1, yy, -dy)
             xx = x1 + 1
 
             do {
-                when active_mode {
-                    1 -> fill_scanline_right_8bpp()
-                    2 -> fill_scanline_right_2bpp()
-                }
+                fill_scanline_right_2bpp()
                 push_stack(left, xx - 1, yy, dy)
                 if xx > x2 + 1
                     push_stack(x2 + 1, xx - 1, yy, -dy)
@@ -759,37 +699,12 @@ skip:
             cx16.VERA_CTRL = 1
             cx16.VERA_ADDR_H = cx16.r1L
             cx16.VERA_ADDR = cx16.r0
-        }
-
-        sub fill_scanline_left_8bpp() -> bool {
-            void addr_mul_24_for_lores_256c(yy as uword, xx as uword)      ; 24 bits result is in r0 and r1L (highest byte)
-            cx16.r1L |= %00011000     ; auto decrement enabled
-            set_vera_address()
-            cx16.r9s = xx
-            while xx >= 0 {
-                if cx16.VERA_DATA0 != cx16.r11L
-                    break
-                cx16.VERA_DATA1 = cx16.r10L
-                xx--
-            }
-            return xx==cx16.r9s
-        }
-
-        sub fill_scanline_right_8bpp() {
-            void addr_mul_24_for_lores_256c(yy as uword, xx as uword)      ; 24 bits result is in r0 and r1L (highest byte)
-            cx16.r1L |= %00010000    ; auto increment enabled
-            set_vera_address()
-            while xx <= width-1 {
-                if cx16.VERA_DATA0 != cx16.r11L
-                    break
-                cx16.VERA_DATA1 = cx16.r10L
-                xx++
-            }
+            cx16.VERA_CTRL = 0
         }
 
         sub fill_scanline_left_2bpp() -> bool {
             uword vx = xx as uword
-            void gfx2.addr_mul_24_for_highres_4c(yy as uword,vx)
+            void gfx_hires.addr_mul_24_for_highres_4c(yy as uword,vx)
             cx16.r1L |= %0001_1000  ; auto decrement
             set_vera_address()
             cmask = amask[lsb(vx) & 3]  ; set the color mask for the first color pel
@@ -833,7 +748,7 @@ set_byte:
         }
 
         sub fill_scanline_right_2bpp() {
-            void gfx2.addr_mul_24_for_highres_4c(yy as uword,xx as uword)
+            void gfx_hires.addr_mul_24_for_highres_4c(yy as uword,xx as uword)
             cx16.r1L |= %00010000    ; auto increment
             set_vera_address()
             cmask = amask[lsb(xx) & 3]  ; set the color mask for the first color pel
@@ -867,7 +782,7 @@ set_byte:
                 }
 set_byte:
                 cx16.VERA_DATA1 = nvub
-                if xx >= gfx2.width-1 break
+                if xx >= WIDTH-1 break
                 cmask = $C0
             }
         }
@@ -879,16 +794,7 @@ set_byte:
     }
 
     sub position(uword @zp xx, uword yy) {
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                void addr_mul_24_for_lores_256c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-            }
-            2 -> {
-                ; highres 4c
-                void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
-            }
-        }
+        void addr_mul_24_for_highres_4c(yy, xx)      ; 24 bits result is in r0 and r1L (highest byte)
         cx16.r2L = cx16.r1L
         cx16.vaddr(cx16.r2L, cx16.r0, 0, 1)
     }
@@ -970,83 +876,36 @@ set_byte:
         ubyte[8] @shared char_bitmap_bytes_left
         ubyte[8] @shared char_bitmap_bytes_right
 
-        when active_mode {
-            1 -> {
-                ; lores 256c
-                while @(sctextptr)!=0 {
-                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
-                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
-                    repeat 8 {
-                        position(xx,yy)
-                        yy++
-                        %asm {{
-                            ldx  p8v_color
-                            lda  cx16.VERA_DATA1
-                            sta  P8ZP_SCRATCH_B1
-                            ldy  #8
--                           asl  P8ZP_SCRATCH_B1
-                            bcc  +
-                            stx  cx16.VERA_DATA0    ; write a pixel
-                            bra  ++
-+                           lda  cx16.VERA_DATA0    ; don't write a pixel, but do advance to the next address
-+                           dey
-                            bne  -
-                        }}
-                    }
-                    xx+=8
-                    yy-=8
-                    sctextptr++
-                }
-            }
-            2 -> {
-                ; hires 4c
-                ; we're going to use a few cx16 registers to make sure every variable is in zeropage in the inner loop.
-                cx16.r11L = color
-                while @(sctextptr)!=0 {
-                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
-                    cx16.vaddr(charset_bank, chardataptr, 1, 1)  ; for reading the chardata from Vera data channel 1
-                    position(xx, yy)              ; only calculated once, we update vera address in the loop instead
-                    cx16.VERA_ADDR_H &= $0f     ; no auto increment
-                    repeat 8 {
-                        cx16.r10L = cx16.VERA_DATA1  ; get the next 8 horizontal character bits
-                        cx16.r7 = xx
-                        repeat 8 {
-                            cx16.r10L <<= 1
-                            if_cs {
-                                cx16.r2L = cx16.r7L & 3       ; xbits
-                                when cx16.r11L & 3 {
-                                    1 -> cx16.r12L = gfx2.plot.shiftedleft_4c_1[cx16.r2L]
-                                    2 -> cx16.r12L = gfx2.plot.shiftedleft_4c_2[cx16.r2L]
-                                    3 -> cx16.r12L = gfx2.plot.shiftedleft_4c_3[cx16.r2L]
-                                    else -> cx16.r12L = 0
-                                }
-                                cx16.VERA_DATA0 = cx16.VERA_DATA0 & gfx2.plot.mask4c[cx16.r2L] | cx16.r12L
-                            }
-                            cx16.r7++
-                            if (cx16.r7 & 3) == 0 {
-                                ; increment the pixel address by one
-                                %asm {{
-                                    stz  cx16.VERA_CTRL
-                                    clc
-                                    lda  cx16.VERA_ADDR_L
-                                    adc  #1
-                                    sta  cx16.VERA_ADDR_L
-                                    lda  cx16.VERA_ADDR_M
-                                    adc  #0
-                                    sta  cx16.VERA_ADDR_M
-                                    lda  cx16.VERA_ADDR_H
-                                    adc  #0
-                                    sta  cx16.VERA_ADDR_H
-                                }}
-                            }
+        ; we're going to use a few cx16 registers to make sure every variable is in zeropage in the inner loop.
+        cx16.r11L = color
+        while @(sctextptr)!=0 {
+            chardataptr = charset_addr + (@(sctextptr) as uword)*8
+            cx16.vaddr(charset_bank, chardataptr, 1, 1)  ; for reading the chardata from Vera data channel 1
+            position(xx, yy)              ; only calculated once, we update vera address in the loop instead
+            cx16.VERA_ADDR_H &= $0f     ; no auto increment
+            repeat 8 {
+                cx16.r10L = cx16.VERA_DATA1  ; get the next 8 horizontal character bits
+                cx16.r7 = xx
+                repeat 8 {
+                    cx16.r10L <<= 1
+                    if_cs {
+                        cx16.r2L = cx16.r7L & 3       ; xbits
+                        when cx16.r11L & 3 {
+                            1 -> cx16.r12L = gfx_hires.plot.shiftedleft_4c_1[cx16.r2L]
+                            2 -> cx16.r12L = gfx_hires.plot.shiftedleft_4c_2[cx16.r2L]
+                            3 -> cx16.r12L = gfx_hires.plot.shiftedleft_4c_3[cx16.r2L]
+                            else -> cx16.r12L = 0
                         }
-
-                        ; increment pixel address to the next line
+                        cx16.VERA_DATA0 = cx16.VERA_DATA0 & gfx_hires.plot.mask4c[cx16.r2L] | cx16.r12L
+                    }
+                    cx16.r7++
+                    if (cx16.r7 & 3) == 0 {
+                        ; increment the pixel address by one
                         %asm {{
                             stz  cx16.VERA_CTRL
                             clc
                             lda  cx16.VERA_ADDR_L
-                            adc  #(640-8)/4
+                            adc  #1
                             sta  cx16.VERA_ADDR_L
                             lda  cx16.VERA_ADDR_M
                             adc  #0
@@ -1056,10 +915,25 @@ set_byte:
                             sta  cx16.VERA_ADDR_H
                         }}
                     }
-                    xx+=8
-                    sctextptr++
                 }
+
+                ; increment pixel address to the next line
+                %asm {{
+                    stz  cx16.VERA_CTRL
+                    clc
+                    lda  cx16.VERA_ADDR_L
+                    adc  #(640-8)/4
+                    sta  cx16.VERA_ADDR_L
+                    lda  cx16.VERA_ADDR_M
+                    adc  #0
+                    sta  cx16.VERA_ADDR_M
+                    lda  cx16.VERA_ADDR_H
+                    adc  #0
+                    sta  cx16.VERA_ADDR_H
+                }}
             }
+            xx+=8
+            sctextptr++
         }
     }
 
@@ -1130,50 +1004,4 @@ set_byte:
             rts
         }}
     }
-
-    asmsub addr_mul_24_for_lores_256c(uword yy @R0, uword xx @AY) clobbers(A) -> uword @R0, ubyte @R1  {
-        ; yy * 320 + xx (24 bits calculation)
-        %asm {{
-            sta  P8ZP_SCRATCH_W1
-            sty  P8ZP_SCRATCH_W1+1
-            lda  cx16.r0
-            sta  P8ZP_SCRATCH_B1
-            lda  cx16.r0+1
-            sta  cx16.r1
-            sta  P8ZP_SCRATCH_REG
-            lda  cx16.r0
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            asl  a
-            rol  P8ZP_SCRATCH_REG
-            sta  cx16.r0
-            lda  P8ZP_SCRATCH_B1
-            clc
-            adc  P8ZP_SCRATCH_REG
-            sta  cx16.r0+1
-            bcc  +
-            inc  cx16.r1
-+           ; now add the value to this 24-bits number
-            lda  cx16.r0
-            clc
-            adc  P8ZP_SCRATCH_W1
-            sta  cx16.r0
-            lda  cx16.r0+1
-            adc  P8ZP_SCRATCH_W1+1
-            sta  cx16.r0+1
-            bcc  +
-            inc  cx16.r1
-+           lda  cx16.r1
-            rts
-        }}
-    }
-
 }

compiler/res/prog8lib/cx16/gfx_lores.p8

@@ -0,0 +1,901 @@
+; optimized graphics routines for just the single screen mode: lores 320*240, 256c  (8bpp)
+; bitmap image needs to start at VRAM addres $00000.
+; This is compatible with the CX16's screen mode 128.  (void cx16.set_screen_mode(128))
+
+
+%import syslib
+%import verafx
+
+gfx_lores {
+    %option ignore_unused
+
+    const uword WIDTH = 320
+    const ubyte HEIGHT = 240
+
+    sub graphics_mode() {
+        ; enable 320x240 256c bitmap graphics 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 = %00000111
+        cx16.VERA_L1_MAPBASE = 0
+        cx16.VERA_L1_TILEBASE = 0
+        clear_screen(0)
+        drawmode_eor(false)
+    }
+
+    sub text_mode() {
+        ; 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 drawmode_eor(bool enabled) {
+        ; with EOR drawing mode you can have non destructive drawing (2*EOR=restore original)
+        eor_mode = enabled
+    }
+
+    bool eor_mode
+
+    sub clear_screen(ubyte color) {
+        if verafx.available() {
+            ; use verafx cache writes to quicly clear the screen
+            const ubyte vbank = 0
+            const uword vaddr = 0
+            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 = color
+            cx16.VERA_FX_CACHE_M = color
+            cx16.VERA_FX_CACHE_H = color
+            cx16.VERA_FX_CACHE_U = color
+            cx16.VERA_CTRL = 2<<1       ; dcsel = 2
+            cx16.VERA_FX_MULT = 0
+            cx16.VERA_FX_CTRL = %01000000    ; cache write enable
+            repeat 320/4/4 {
+                %asm {{
+                    ldy  #240
+-                   stz  cx16.VERA_DATA0
+                    stz  cx16.VERA_DATA0
+                    stz  cx16.VERA_DATA0
+                    stz  cx16.VERA_DATA0
+                    dey
+                    bne  -
+                }}
+            }
+            cx16.VERA_FX_CTRL = 0       ; cache write disable
+            cx16.VERA_CTRL = 0
+            return
+        }
+        ; fallback to cpu clear
+        cx16.VERA_CTRL=0
+        cx16.VERA_ADDR=0
+        cx16.VERA_ADDR_H = 1<<4    ; 1 pixel auto increment
+        repeat HEIGHT {
+            %asm {{
+                lda  p8v_color
+                ldy  #p8c_WIDTH/8
+-               .rept 8
+                sta  cx16.VERA_DATA0
+                .endrept
+                dey
+                bne  -
+            }}
+        }
+        cx16.VERA_ADDR=0
+        cx16.VERA_ADDR_H = 0
+    }
+
+    sub rect(uword xx, ubyte yy, uword rwidth, ubyte rheight, ubyte color) {
+        if rwidth==0 or rheight==0
+            return
+        horizontal_line(xx, yy, rwidth, color)
+        if rheight==1
+            return
+        horizontal_line(xx, yy+rheight-1, rwidth, color)
+        vertical_line(xx, yy+1, rheight-2, color)
+        if rwidth==1
+            return
+        vertical_line(xx+rwidth-1, yy+1, rheight-2, color)
+    }
+
+    sub safe_rect(uword xx, ubyte yy, uword rwidth, ubyte rheight, ubyte color) {
+        ; does bounds checking and clipping
+        safe_horizontal_line(xx, yy, rwidth, color)
+        if rheight==1
+            return
+        uword bottomyy = yy as uword + rheight -1
+        if bottomyy<HEIGHT
+            safe_horizontal_line(xx, lsb(bottomyy), rwidth, color)
+        safe_vertical_line(xx, yy+1, rheight-2, color)
+        if rwidth==1
+            return
+        safe_vertical_line(xx+rwidth-1, yy+1, rheight-2, color)
+    }
+
+    sub fillrect(uword xx, ubyte yy, uword rwidth, ubyte rheight, ubyte color) {
+        ; 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, color)
+            yy++
+        }
+    }
+
+    sub safe_fillrect(uword xx, ubyte yy, uword rwidth, ubyte rheight, ubyte color) {
+        ; Draw a filled rectangle of the given size and color.
+        ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
+        ; This safe version does bounds checking and clipping.
+        if xx>=WIDTH or yy>=HEIGHT
+            return
+        if msb(xx)&$80!=0 {
+            rwidth += xx
+            xx = 0
+        }
+        if xx>=WIDTH
+            return
+        if xx+rwidth>WIDTH
+            rwidth = WIDTH-xx
+        if rwidth>WIDTH
+            return
+
+        if yy as uword + rheight > HEIGHT
+            rheight = HEIGHT-yy
+        if rheight>HEIGHT
+            return
+
+        repeat rheight {
+            horizontal_line(xx, yy, rwidth, color)
+            yy++
+        }
+    }
+
+    sub horizontal_line(uword xx, ubyte yy, uword length, ubyte color) {
+        if length==0
+            return
+        position(xx, yy)
+        ; set vera auto-increment to 1 pixel
+        cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (1<<4)
+        if eor_mode {
+            cx16.vaddr_clone(0)      ; also setup port 1, for reading
+            %asm {{
+                ldx  p8v_length+1
+                beq  +
+                ldy  #0
+-               lda  p8v_color
+                eor  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                iny
+                bne  -
+                dex
+                bne  -
++               ldy  p8v_length     ; remaining
+                beq  +
+-               lda  p8v_color
+                eor  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                dey
+                bne  -
++
+            }}
+        } else {
+            %asm {{
+                lda  p8v_color
+                ldx  p8v_length+1
+                beq  +
+                ldy  #0
+-               sta  cx16.VERA_DATA0
+                iny
+                bne  -
+                dex
+                bne  -
++               ldy  p8v_length     ; remaining
+                beq  +
+-               sta  cx16.VERA_DATA0
+                dey
+                bne  -
++
+            }}
+        }
+    }
+
+    sub safe_horizontal_line(uword xx, ubyte yy, uword length, ubyte color) {
+        ; does bounds checking and clipping
+        if 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, color)
+    }
+
+    sub vertical_line(uword xx, ubyte yy, ubyte lheight, ubyte color) {
+        if lheight==0
+            return
+        position(xx, yy)
+        ; set vera auto-increment to 320 pixel increment (=next line)
+        cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
+        if eor_mode {
+            cx16.vaddr_clone(0)      ; also setup port 1, for reading
+            %asm {{
+                ldy  p8v_lheight
+                beq  +
+-               lda  p8v_color
+                eor  cx16.VERA_DATA1
+                sta  cx16.VERA_DATA0
+                dey
+                bne  -
++
+            }}
+        } else {
+            %asm {{
+                ldy  p8v_lheight
+                lda  p8v_color
+-               sta  cx16.VERA_DATA0
+                dey
+                bne  -
+            }}
+        }
+    }
+
+    sub safe_vertical_line(uword xx, ubyte yy, ubyte lheight, ubyte color) {
+        ; does bounds checking and clipping
+        if yy>=HEIGHT
+            return
+        if msb(xx)&$80!=0 or xx>=WIDTH
+            return
+        if yy as uword + lheight > HEIGHT
+            lheight = HEIGHT-yy
+        if lheight>HEIGHT
+            return
+
+        vertical_line(xx, yy, lheight, color)
+    }
+
+    sub line(uword x1, ubyte y1, uword x2, ubyte y2, ubyte color) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        ; NOTE:  this is about twice as fast as the kernal routine GRAPH_draw_line
+        ;        it trades memory for speed (uses inline plot routine and multiplication lookup tables)
+        ;
+        ; NOTE:  is currently still a regular 6502 routine, could likely be made much faster with the VeraFX line helper.
+
+        cx16.r3L = y2    ; ensure zeropage
+        cx16.r1L = y1    ; ensure zeropage
+
+        if cx16.r1L > cx16.r3L {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            cx16.r0 = x1
+            x1 = x2
+            x2 = cx16.r0
+            cx16.r0L = cx16.r1L
+            cx16.r1L = cx16.r3L
+            cx16.r3L = cx16.r0L
+        }
+        word @zp dx = x2 as word
+        word @zp dy = cx16.r3L
+        dx -= x1
+        dy -= cx16.r1L
+
+        if dx==0 {
+            vertical_line(x1, cx16.r1L, lsb(dy)+1, color)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, cx16.r1L, abs(dx) as uword +1, color)
+            return
+        }
+
+        word @zp d = 0
+        bool positive_ix = true
+        if dx < 0 {
+            dx = -dx
+            positive_ix = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+
+        cx16.r0  = x1    ; ensure zeropage
+        cx16.r2  = x2    ; ensure zeropage
+
+        cx16.VERA_CTRL = 0
+        if dx >= dy {
+            if positive_ix {
+                repeat {
+                    plot()
+                    if cx16.r0==cx16.r2
+                        return
+                    cx16.r0++
+                    d += dy2
+                    if d > dx {
+                        cx16.r1L++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    plot()
+                    if cx16.r0==cx16.r2
+                        return
+                    cx16.r0--
+                    d += dy2
+                    if d > dx {
+                        cx16.r1L++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if positive_ix {
+                repeat {
+                    plot()
+                    if cx16.r1L == cx16.r3L
+                        return
+                    cx16.r1L++
+                    d += dx2
+                    if d > dy {
+                        cx16.r0++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    plot()
+                    if cx16.r1L == cx16.r3L
+                        return
+                    cx16.r1L++
+                    d += dx2
+                    if d > dy {
+                        cx16.r0--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+
+        asmsub plot() {
+            ; internal plot routine for the line algorithm: x in r0,  y in r1,  color in variable.
+            %asm {{
+                ldy  cx16.r1L
+                clc
+                lda  times320_lo,y
+                adc  cx16.r0L
+                sta  cx16.VERA_ADDR_L
+                lda  times320_mid,y
+                adc  cx16.r0H
+                sta  cx16.VERA_ADDR_M
+                lda  #0
+                adc  times320_hi,y
+                sta  cx16.VERA_ADDR_H
+
+                lda  p8v_eor_mode
+                bne  +
+                lda  p8v_color
+                sta  cx16.VERA_DATA0
+                rts
++               lda  p8v_color
+                eor  cx16.VERA_DATA0
+                sta  cx16.VERA_DATA0
+                rts
+            }}
+        }
+    }
+
+    sub circle(uword @zp xcenter, ubyte @zp ycenter, ubyte radius, ubyte color) {
+        ; 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, color=color.
+            plot(cx16.r14, cx16.r15L, color)
+        }
+    }
+
+    sub safe_circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, ubyte color) {
+        ; This version does bounds checks and clipping, but is a lot slower.
+        ; 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, color=color.
+            if cx16.r15 < HEIGHT
+                safe_plot(cx16.r14, cx16.r15L, color)
+        }
+    }
+
+    sub disc(uword @zp xcenter, ubyte @zp ycenter, ubyte @zp radius, ubyte color) {
+        ; 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
+        ubyte last_y3 = ycenter+radius
+        ubyte last_y4 = ycenter-radius
+        ubyte new_y3, new_y4
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
+
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
+                last_y4 = new_y4
+            }
+
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+
+        ; draw the final two spans
+        yy--
+        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
+        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
+    }
+
+    sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
+        ; This version does bounds checks and clipping, but is a lot slower.
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+        uword last_y3 = ycenter+radius
+        uword last_y4 = ycenter-radius
+        uword new_y3, new_y4
+
+        while radius>=yy {
+            uword liney = ycenter+yy
+            if msb(liney)==0
+                safe_horizontal_line(xcenter-radius, lsb(ycenter+yy), radius*$0002+1, color)
+            liney = ycenter-yy
+            if msb(liney)==0
+                safe_horizontal_line(xcenter-radius, lsb(ycenter-yy), radius*$0002+1, color)
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                if msb(last_y3)==0
+                    safe_horizontal_line(xcenter-yy, lsb(last_y3), yy*$0002+1, color)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                if msb(last_y4)==0
+                    safe_horizontal_line(xcenter-yy, lsb(last_y4), yy*$0002+1, color)
+                last_y4 = new_y4
+            }
+            yy++
+            if decisionOver2>=0 {
+                radius--
+                decisionOver2 -= radius*$0002
+            }
+            decisionOver2 += yy*$0002
+            decisionOver2++
+        }
+        ; draw the final two spans
+        yy--
+        if msb(last_y3)==0
+            safe_horizontal_line(xcenter-yy, lsb(last_y3), yy*$0002+1, color)
+        if msb(last_y4)==0
+            safe_horizontal_line(xcenter-yy, lsb(last_y4), yy*$0002+1, color)
+    }
+
+    asmsub plot(uword x @AX, ubyte y @Y, ubyte color @R0) {
+        ; x in r0,  y in r1,   color.
+        %asm {{
+            clc
+            adc  times320_lo,y
+            sta  cx16.VERA_ADDR_L
+            txa
+            adc  times320_mid,y
+            sta  cx16.VERA_ADDR_M
+            lda  #0
+            adc  times320_hi,y
+            sta  cx16.VERA_ADDR_H
+
+            lda  p8v_eor_mode
+            bne  +
+            lda  cx16.r0L
+            sta  cx16.VERA_DATA0
+            rts
++           lda  cx16.r0L
+            eor  cx16.VERA_DATA0
+            sta  cx16.VERA_DATA0
+            rts
+        }}
+    }
+
+    sub safe_plot(uword xx, ubyte yy, ubyte color) {
+        ; A plot that does bounds checks to see if the pixel is inside the screen.
+        if msb(xx)&$80!=0
+            return
+        if xx >= WIDTH or yy >= HEIGHT
+            return
+        plot(xx, yy, color)
+    }
+
+    asmsub pget(uword x @AX, ubyte y @Y) -> ubyte @A {
+        ; returns the color of the pixel
+        %asm {{
+            jsr  p8s_position
+            lda  cx16.VERA_DATA0
+            rts
+        }}
+    }
+
+    sub fill(uword x, ubyte y, ubyte new_color) {
+        ; reuse a few virtual registers in ZP for variables
+        &ubyte fillm = &cx16.r7L
+        &ubyte seedm = &cx16.r8L
+        &ubyte cmask = &cx16.r8H
+        &ubyte vub   = &cx16.r13L
+        &ubyte nvub  = &cx16.r13H
+        ubyte[4] amask = [$c0,$30,$0c,$03] ; array of cmask bytes
+
+        ; 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 = 100
+        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 = new_color
+
+        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  p8v_sxl
+                    sta  p8v_stack_xl_lsb,y
+                    lda  p8v_sxl+1
+                    sta  p8v_stack_xl_msb,y
+                    lda  p8v_sxr
+                    sta  p8v_stack_xr_lsb,y
+                    lda  p8v_sxr+1
+                    sta  p8v_stack_xr_msb,y
+                    lda  p8v_sy
+                    sta  p8v_stack_y_lsb,y
+                    lda  p8v_sy+1
+                    sta  p8v_stack_y_msb,y
+                    ldy  cx16.r12L
+                    lda  p8v_sdy
+                    sta  p8v_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  p8v_stack_xl_lsb,y
+                sta  p8v_x1
+                lda  p8v_stack_xl_msb,y
+                sta  p8v_x1+1
+                lda  p8v_stack_xr_lsb,y
+                sta  p8v_x2
+                lda  p8v_stack_xr_msb,y
+                sta  p8v_x2+1
+                lda  p8v_stack_y_lsb,y
+                sta  p8v_yy
+                lda  p8v_stack_y_msb,y
+                sta  p8v_yy+1
+                ldy  cx16.r12L
+                lda  p8v_stack_dy,y
+                sta  p8v_dy
+            }}
+            yy+=dy
+        }
+        cx16.r11L = pget(xx as uword, lsb(yy))        ; 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!=0 {
+            pop_stack()
+            xx = x1
+            if fill_scanline_left_8bpp() goto skip
+            left = xx + 1
+            if left < x1
+                push_stack(left, x1 - 1, yy, -dy)
+            xx = x1 + 1
+
+            do {
+                fill_scanline_right_8bpp()
+                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, lsb(yy)) == cx16.r11L
+                        break
+                    xx++
+                }
+                left = xx
+            } until xx>x2
+        }
+
+        sub set_vera_address(bool decr) {
+            ; set both data0 and data1 addresses
+            position(xx as uword, lsb(yy))
+            cx16.r0 = cx16.VERA_ADDR
+            cx16.r1L = cx16.VERA_ADDR_H & 1 | if decr %00011000 else %00010000
+            cx16.VERA_ADDR_H = cx16.r1L
+            cx16.VERA_CTRL = 1
+            cx16.VERA_ADDR = cx16.r0
+            cx16.VERA_ADDR_H = cx16.r1L
+            cx16.VERA_CTRL = 0
+        }
+
+        sub fill_scanline_left_8bpp() -> bool {
+            set_vera_address(true)
+            cx16.r9s = xx
+            while xx >= 0 {
+                if cx16.VERA_DATA0 != cx16.r11L
+                    break
+                cx16.VERA_DATA1 = cx16.r10L
+                xx--
+            }
+            return xx==cx16.r9s
+        }
+
+        sub fill_scanline_right_8bpp() {
+            set_vera_address(false)
+            while xx <= WIDTH-1 {
+                if cx16.VERA_DATA0 != cx16.r11L
+                    break
+                cx16.VERA_DATA1 = cx16.r10L
+                xx++
+            }
+        }
+    }
+
+    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 etc. etc.
+        cx16.screen_set_charset(charset, 0)
+    }
+
+    const ubyte charset_bank = $1
+    const uword charset_addr = $f000       ; in bank 1, so $1f000
+
+    sub text(uword @zp xx, uword yy, ubyte color, uword textptr) {
+        ; -- Write some text at the given pixel position. The text string must be in an encoding approprite for the charset.
+        ;    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
+
+        while @(textptr)!=0 {
+            chardataptr = charset_addr + (@(textptr) as uword)*8
+            cx16.vaddr(charset_bank, chardataptr, 1, 1)
+            repeat 8 {
+                position(xx,lsb(yy))
+                yy++
+                %asm {{
+                    ldx  p8v_color
+                    lda  cx16.VERA_DATA1
+                    sta  P8ZP_SCRATCH_B1
+                    ldy  #8
+-                   asl  P8ZP_SCRATCH_B1
+                    bcc  +
+                    stx  cx16.VERA_DATA0    ; write a pixel
+                    bra  ++
++                   lda  cx16.VERA_DATA0    ; don't write a pixel, but do advance to the next address
++                   dey
+                    bne  -
+                }}
+            }
+            xx+=8
+            yy-=8
+            textptr++
+        }
+    }
+
+    asmsub position(uword x @AX, ubyte y @Y) {
+        %asm {{
+            clc
+            adc  times320_lo,y
+            sta  cx16.VERA_ADDR_L
+            txa
+            adc  times320_mid,y
+            sta  cx16.VERA_ADDR_M
+            lda  #%00010000         ; auto increment on
+            adc  times320_hi,y
+            sta  cx16.VERA_ADDR_H
+            rts
+        }}
+    }
+
+    inline asmsub next_pixel(ubyte color @A) {
+        ; -- sets the next pixel byte to the graphics chip.
+        ;    for 8 bpp screens this will plot 1 pixel.
+        ;    for 2 bpp screens it will plot 4 pixels at once (color = bit pattern).
+        %asm {{
+            sta  cx16.VERA_DATA0
+        }}
+    }
+
+    asmsub next_pixels(uword pixels @AY, uword amount @R0) clobbers(A, X, Y)  {
+        ; -- sets the next bunch of pixels from a prepared array of bytes.
+        ;    for 8 bpp screens this will plot 1 pixel per byte.
+        ;    for 2 bpp screens it will plot 4 pixels at once (colors are the bit patterns per byte).
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r0+1
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1       ; next page of 256 pixels
+            dex
+            bne  -
+
++           ldx  cx16.r0           ; remaining pixels
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            dex
+            bne  -
++           rts
+        }}
+    }
+
+    asmsub set_8_pixels_from_bits(ubyte bits @R0, ubyte oncolor @A, ubyte offcolor @Y) clobbers(X) {
+        ; this is only useful in 256 color mode where one pixel equals one byte value.
+        %asm {{
+            ldx  #8
+-           asl  cx16.r0
+            bcc  +
+            sta  cx16.VERA_DATA0
+            bra  ++
++           sty  cx16.VERA_DATA0
++           dex
+            bne  -
+            rts
+        }}
+    }
+
+    %asm {{
+; multiplication by 320 lookup table
+times320 := 320*range(240)
+
+times320_lo     .byte <times320
+times320_mid    .byte >times320
+times320_hi     .byte `times320
+    }}
+}

compiler/res/prog8lib/cx16/graphics.p8

@@ -1,11 +1,10 @@
 %import syslib
-%import textio
 
 ; Bitmap pixel graphics module for the CommanderX16
 ; 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" or "monogfx" module instead. (which is Cx16-specific)
+; For other resolutions or other color modes, use the "gfx_lores", "gfx_hires", or "monogfx" module instead.
 ; 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!
@@ -16,7 +15,7 @@
 graphics {
     %option ignore_unused
 
-    romsub $feff = FB_cursor_position2()  clobbers(A,X,Y)     ; alias for the normal FB_cursor_position() call but reuses existing r0 and r1
+    extsub $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
@@ -35,7 +34,7 @@ graphics {
     sub disable_bitmap_mode() {
         ; enables text mode, erase the text screen, color white
         void cx16.screen_mode(0, false)
-        txt.clear_screen()
+        cbm.CHROUT(147)
     }
 
 
@@ -76,83 +75,30 @@ graphics {
 
     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
-        ;cx16.r3 = radius*2
-        ;cx16.GRAPH_draw_oval(false)          ; currently this call is not implemented on cx16, does a BRK
-
-        ; Midpoint algorithm
         if radius==0
             return
-        ubyte @zp xx = radius
-        ubyte @zp yy = 0
-        word @zp decisionOver2 = (1 as word)-xx
-
-        while xx>=yy {
-            cx16.r0 = xcenter + xx
-            cx16.r1 = ycenter + yy
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter - xx
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter + xx
-            cx16.r1 = ycenter - yy
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter - xx
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter + yy
-            cx16.r1 = ycenter + xx
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter - yy
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter + yy
-            cx16.r1 = ycenter - xx
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-            cx16.r0 = xcenter - yy
-            FB_cursor_position2()
-            cx16.FB_set_pixel(stroke_color)
-
-            yy++
-            if decisionOver2>=0 {
-                xx--
-                decisionOver2 -= xx*$0002
-            }
-            decisionOver2 += yy*$0002
-            decisionOver2++
-        }
+        cx16.GRAPH_draw_oval(xcenter - radius, ycenter - radius, radius*2, radius*2, false)
     }
 
     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
-        word decisionOver2 = (1 as word)-radius
+        cx16.GRAPH_draw_oval(xcenter - radius, ycenter - radius, radius*2, radius*2, true)
+    }
 
-        while radius>=yy {
-            horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
-            horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
-            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
-            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
+    sub oval(uword xcenter, ubyte ycenter, uword h_radius, ubyte v_radius) {
+        ; specific for the X16 which has a kernal routine that can draw ovals
+        if h_radius==0 or v_radius==0
+            return
+        cx16.GRAPH_draw_oval(xcenter - h_radius, ycenter - v_radius, h_radius*2, v_radius*2, false)
+    }
 
-            yy++
-            if decisionOver2>=0 {
-                radius--
-                decisionOver2 -= radius*$0002
-            }
-            decisionOver2 += yy*$0002
-            decisionOver2++
-        }
+    sub filled_oval(uword xcenter, ubyte ycenter, uword h_radius, ubyte v_radius) {
+        ; specific for the X16 which has a kernal routine that can draw ovals
+        if h_radius==0 or v_radius==0
+            return
+        cx16.GRAPH_draw_oval(xcenter - h_radius, ycenter - v_radius, h_radius*2, v_radius*2, true)
     }
 
     inline asmsub  plot(uword plotx @R0, uword ploty @R1) clobbers(A, X, Y) {