avoid possible crash

In certain cases you will need to use string.copy() explicitly to overwrite strings with new strings.

.gitignore

@@ -1,5 +1,6 @@
 .idea/workspace.xml
 .idea/discord.xml
+.idea/developer-tools.xml
 build/
 dist/
 output/

.idea/libraries/KotlinJavaRuntime.xml

@@ -1,23 +1,23 @@
 <component name="libraryTable">
   <library name="KotlinJavaRuntime" type="repository">
-    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:1.9.24" />
+    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:2.0.20" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.9.24/kotlin-stdlib-jdk8-1.9.24.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.24/kotlin-stdlib-1.9.24.jar!/" />
+      <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/annotations/13.0/annotations-13.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.9.24/kotlin-stdlib-jdk7-1.9.24.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20.jar!/" />
     </CLASSES>
     <JAVADOC>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.9.24/kotlin-stdlib-jdk8-1.9.24-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.24/kotlin-stdlib-1.9.24-javadoc.jar!/" />
+      <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/annotations/13.0/annotations-13.0-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.9.24/kotlin-stdlib-jdk7-1.9.24-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20-javadoc.jar!/" />
     </JAVADOC>
     <SOURCES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/1.9.24/kotlin-stdlib-jdk8-1.9.24-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/1.9.24/kotlin-stdlib-1.9.24-sources.jar!/" />
+      <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/annotations/13.0/annotations-13.0-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/1.9.24/kotlin-stdlib-jdk7-1.9.24-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.0.20/kotlin-stdlib-jdk7-2.0.20-sources.jar!/" />
     </SOURCES>
   </library>
 </component>

.idea/libraries/antlr_antlr4.xml

@@ -1,13 +1,13 @@
 <component name="libraryTable">
   <library name="antlr.antlr4" type="repository">
-    <properties maven-id="org.antlr:antlr4:4.13.1">
+    <properties maven-id="org.antlr:antlr4:4.13.2">
       <exclude>
         <dependency maven-id="com.ibm.icu:icu4j" />
       </exclude>
     </properties>
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4/4.13.1/antlr4-4.13.1.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4-runtime/4.13.1/antlr4-runtime-4.13.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4/4.13.2/antlr4-4.13.2.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr4-runtime/4.13.2/antlr4-runtime-4.13.2.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/antlr/antlr-runtime/3.5.3/antlr-runtime-3.5.3.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/antlr/ST4/4.3.4/ST4-4.3.4.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/abego/treelayout/org.abego.treelayout.core/1.0.3/org.abego.treelayout.core-1.0.3.jar!/" />

.idea/libraries/io_kotest_assertions_core_jvm.xml

@@ -1,17 +1,17 @@
 <component name="libraryTable">
   <library name="io.kotest.assertions.core.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-assertions-core-jvm:5.9.0" />
+    <properties maven-id="io.kotest:kotest-assertions-core-jvm:5.9.1" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.9.0/kotest-assertions-core-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.9.0/kotest-assertions-shared-jvm-5.9.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.9.1/kotest-assertions-core-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/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/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-jdk8/1.8.0/kotlinx-coroutines-jdk8-1.8.0.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-common-jvm/5.9.0/kotest-common-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-api-jvm/5.9.0/kotest-assertions-api-jvm-5.9.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.9.1/kotest-common-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/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-core-jvm/1.8.0/kotlinx-coroutines-core-jvm-1.8.0.jar!/" />
     </CLASSES>
     <JAVADOC />

.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/io_kotest_runner_junit5_jvm.xml

@@ -1,14 +1,14 @@
 <component name="libraryTable">
   <library name="io.kotest.runner.junit5.jvm" type="repository">
-    <properties maven-id="io.kotest:kotest-runner-junit5-jvm:5.9.0" />
+    <properties maven-id="io.kotest:kotest-runner-junit5-jvm:5.9.1" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-runner-junit5-jvm/5.9.0/kotest-runner-junit5-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-api-jvm/5.9.0/kotest-framework-api-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-shared-jvm/5.9.0/kotest-assertions-shared-jvm-5.9.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-runner-junit5-jvm/5.9.1/kotest-runner-junit5-jvm-5.9.1.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/github/java-diff-utils/java-diff-utils/4.12/java-diff-utils-4.12.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlinx/kotlinx-coroutines-test-jvm/1.8.0/kotlinx-coroutines-test-jvm-1.8.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.9.0/kotest-common-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-engine-jvm/5.9.0/kotest-framework-engine-jvm-5.9.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-common-jvm/5.9.1/kotest-common-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-engine-jvm/5.9.1/kotest-framework-engine-jvm-5.9.1.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/io/github/classgraph/classgraph/4.8.172/classgraph-4.8.172.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/com/github/ajalt/mordant/1.2.1/mordant-1.2.1.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/com/github/ajalt/colormath/1.2.0/colormath-1.2.0.jar!/" />
@@ -18,12 +18,12 @@
       <root url="jar://$MAVEN_REPOSITORY$/net/java/dev/jna/jna-platform/5.9.0/jna-platform-5.9.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy/1.10.9/byte-buddy-1.10.9.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/net/bytebuddy/byte-buddy-agent/1.10.9/byte-buddy-agent-1.10.9.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-discovery-jvm/5.9.0/kotest-framework-discovery-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.9.0/kotest-assertions-core-jvm-5.9.0.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-discovery-jvm/5.9.1/kotest-framework-discovery-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-assertions-core-jvm/5.9.1/kotest-assertions-core-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/kotest/kotest-assertions-api-jvm/5.9.0/kotest-assertions-api-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-extensions-jvm/5.9.0/kotest-extensions-jvm-5.9.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-concurrency-jvm/5.9.0/kotest-framework-concurrency-jvm-5.9.0.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://$MAVEN_REPOSITORY$/io/kotest/kotest-extensions-jvm/5.9.1/kotest-extensions-jvm-5.9.1.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/io/kotest/kotest-framework-concurrency-jvm/5.9.1/kotest-framework-concurrency-jvm-5.9.1.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://$MAVEN_REPOSITORY$/org/jetbrains/annotations/23.0.0/annotations-23.0.0.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/junit/platform/junit-platform-engine/1.8.2/junit-platform-engine-1.8.2.jar!/" />

.idea/libraries/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

@@ -12,9 +12,10 @@
       <module fileurl="file://$PROJECT_DIR$/compilerAst/compilerAst.iml" filepath="$PROJECT_DIR$/compilerAst/compilerAst.iml" />
       <module fileurl="file://$PROJECT_DIR$/docs/docs.iml" filepath="$PROJECT_DIR$/docs/docs.iml" />
       <module fileurl="file://$PROJECT_DIR$/examples/examples.iml" filepath="$PROJECT_DIR$/examples/examples.iml" />
+      <module fileurl="file://$PROJECT_DIR$/benchmark-program/benchmark-program.iml" filepath="$PROJECT_DIR$/benchmark-program/benchmark-program.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://$PROJECT_DIR$/virtualmachine/virtualmachine.iml" filepath="$PROJECT_DIR$/virtualmachine/virtualmachine.iml" />
     </modules>
   </component>
-</project>
+</project>

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)
 
@@ -52,26 +54,23 @@ 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
-- modularity, symbol scoping, subroutines
+- 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. 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 
-- many built-in functions such as ``sin``, ``cos``, ``rnd``, ``abs``, ``min``, ``max``, ``sqrt``, ``msb``, ``rol``, ``ror``, ``swap``, ``sort`` and ``reverse``
+- 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
 
 *Rapid edit-compile-run-debug cycle:*
 

benchmark-program/Makefile

@@ -0,0 +1,12 @@
+.PHONY:  all clean emu
+
+all:  benchmark.prg
+
+clean:
+	rm -f *.prg *.PRG *.asm *.vice-* *.BIN *.PAL *.zip *.7z
+
+emu:  benchmark.prg
+	x16emu -run -prg $< -warp
+
+benchmark.prg: benchmark.p8 b_3d.p8 b_adpcm.p8 b_circles.p8 b_life.p8 b_mandelbrot.p8 b_maze.p8 b_queens.p8 b_textelite.p8
+	p8compile $< -target cx16 

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 gfx2
+%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
+            gfx2.screen_mode(1)
+
+        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
+                gfx2.clear_screen(0)
+            total_num_circles += draw_circles(use_kernal, max_time)
+        }
+
+        if use_kernal
+            void cx16.set_screen_mode(3)
+        else {
+            gfx2.screen_mode(0)
+        }
+
+        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
+                        gfx2.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-program.iml

@@ -0,0 +1,10 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<module type="WEB_MODULE" version="4">
+  <component name="NewModuleRootManager" inherit-compiler-output="true">
+    <exclude-output />
+    <content url="file://$MODULE_DIR$">
+      <excludeFolder url="file://$MODULE_DIR$/compiled" />
+    </content>
+    <orderEntry type="sourceFolder" forTests="false" />
+  </component>
+</module>

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 gfx2")
+        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

@@ -222,7 +222,7 @@ class StMemVar(name: String,
     init{
         require(dt!=DataType.BOOL && dt!=DataType.ARRAY_BOOL)
         if(dt in ArrayDatatypes || dt == DataType.STR)
-            require(length!=null) { "memory mapped array or string must have known length" }
+            requireNotNull(length)
     }
 }
 

codeCore/src/prog8/code/SymbolTableMaker.kt

@@ -3,7 +3,8 @@ package prog8.code
 import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.code.target.VMTarget
-import java.util.*
+import kotlin.collections.ArrayDeque
+
 
 class SymbolTableMaker(private val program: PtProgram, private val options: CompilationOptions) {
     fun make(): SymbolTable {
@@ -13,8 +14,8 @@ 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)
         }
@@ -35,7 +36,7 @@ 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) }
@@ -104,7 +105,7 @@ 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 +113,14 @@ 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> {

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
@@ -195,10 +206,17 @@ class PtBinaryExpression(val operator: String, type: DataType, position: Positio
 
 
 class PtContainmentCheck(position: Position): PtExpression(DataType.BOOL, position) {
-    val element: PtExpression
+    val needle: PtExpression
         get() = children[0] as PtExpression
-    val iterable: PtIdentifier
-        get() = children[1] as PtIdentifier
+    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
+    }
 }
 
 
@@ -227,12 +245,6 @@ 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 {
@@ -267,12 +279,12 @@ class PtNumber(type: DataType, val number: Double, position: Position) : PtExpre
     override fun hashCode(): Int = Objects.hash(type, number)
 
     override fun equals(other: Any?): Boolean {
-        if(other==null || other !is PtNumber)
-            return false
+        return if(other==null || other !is PtNumber)
+            false
         else if(type!=DataType.BOOL && other.type!=DataType.BOOL)
-            return number==other.number
+            number==other.number
         else
-            return type==other.type && number==other.number
+            type==other.type && number==other.number
     }
 
     operator fun compareTo(other: PtNumber): Int = number.compareTo(other.number)
@@ -302,6 +314,7 @@ 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

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

@@ -21,7 +21,16 @@ fun printAst(root: PtNode, skipLibraries: Boolean, output: (text: String) -> Uni
                 else
                     "&"
             }
-            is PtArray -> "array len=${node.children.size} ${type(node.type)}"
+            is PtArray -> {
+                val valuelist = node.children.map {
+                    when (it) {
+                        is PtNumber -> it.number.toString()
+                        is PtIdentifier -> it.name
+                        else -> "?"
+                    }
+                }.joinToString(", ")
+                "array len=${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 -> {

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

@@ -5,6 +5,7 @@ import prog8.code.core.*
 
 sealed interface IPtSubroutine {
     val name: String
+    val scopedName: String
 }
 
 class PtAsmSub(

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

@@ -76,14 +76,14 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "ror"       to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
     "rol2"      to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
     "ror2"      to FSignature(false, listOf(FParam("item", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
-    "sort"      to FSignature(false, listOf(FParam("array", ArrayDatatypes)), null),
-    "reverse"   to FSignature(false, listOf(FParam("array", ArrayDatatypes)), null),
     // cmp returns a status in the carry flag, but not a proper return value
     "cmp"       to FSignature(false, listOf(FParam("value1", 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),
+    "prog8_ifelse_bittest_notset" to FSignature(true, listOf(FParam("variable", ByteDatatypes), FParam("bitnumber", arrayOf(DataType.UBYTE))), DataType.BOOL),
     "abs"       to FSignature(true, listOf(FParam("value", NumericDatatypes)), null),
     "abs__byte"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.BYTE))), DataType.BYTE),
     "abs__word"  to FSignature(true, listOf(FParam("value", arrayOf(DataType.WORD))), DataType.WORD),
@@ -99,8 +99,6 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "divmod"           to FSignature(false, listOf(FParam("dividend", arrayOf(DataType.UBYTE, DataType.UWORD)), FParam("divisor", arrayOf(DataType.UBYTE, DataType.UWORD)), FParam("quotient", arrayOf(DataType.UBYTE, DataType.UWORD)), FParam("remainder", arrayOf(DataType.UBYTE, DataType.UWORD))), null),
     "divmod__ubyte"    to FSignature(false, listOf(FParam("dividend", arrayOf(DataType.UBYTE)), FParam("divisor", arrayOf(DataType.UBYTE)), FParam("quotient", arrayOf(DataType.UBYTE)), FParam("remainder", arrayOf(DataType.UBYTE))), null),
     "divmod__uword"    to FSignature(false, listOf(FParam("dividend", arrayOf(DataType.UWORD)), FParam("divisor", arrayOf(DataType.UWORD)), FParam("quotient", arrayOf(DataType.UWORD)), FParam("remainder", arrayOf(DataType.UWORD))), null),
-    "any"       to FSignature(true, listOf(FParam("values", ArrayDatatypes)), DataType.BOOL),
-    "all"       to FSignature(true, listOf(FParam("values", ArrayDatatypes)), DataType.BOOL),
     "lsb"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE),
     "msb"       to FSignature(true, listOf(FParam("value", arrayOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE),
     "mkword"    to FSignature(true, listOf(FParam("msb", arrayOf(DataType.UBYTE)), FParam("lsb", arrayOf(DataType.UBYTE))), DataType.UWORD),
@@ -130,12 +128,12 @@ 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),
 )
 
 val InplaceModifyingBuiltinFunctions = setOf(
     "setlsb", "setmsb",
     "rol", "ror", "rol2", "ror2",
-    "sort", "reverse",
     "divmod", "divmod__ubyte", "divmod__uword"
 )

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

@@ -27,7 +27,6 @@ class CompilationOptions(val output: OutputType,
                          var slabsHighBank: Int? = null,
                          var slabsGolden: Boolean = false,
                          var splitWordArrays: Boolean = false,
-                         var strictBool: Boolean = true,
                          var breakpointCpuInstruction: String? = null,
                          var outputDir: Path = Path(""),
                          var symbolDefs: Map<String, String> = emptyMap()

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

@@ -1,6 +1,11 @@
 package prog8.code.core
 
 import kotlin.math.abs
+import kotlin.math.pow
+
+val powersOfTwoFloat = (0..16).map { (2.0).pow(it) }.toTypedArray()
+val negativePowersOfTwoFloat = powersOfTwoFloat.map { -it }.toTypedArray()
+val powersOfTwoInt = (0..16).map { 2.0.pow(it).toInt() }.toTypedArray()
 
 fun Number.toHex(): String {
     //  0..15 -> "0".."15"

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

@@ -29,7 +29,6 @@ interface IMachineDefinition {
     fun convertFloatToBytes(num: Double): List<UByte>
     fun convertBytesToFloat(bytes: List<UByte>): Double
 
-    fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String>
     fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path)
     fun isIOAddress(address: UInt): Boolean
 }

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

@@ -8,7 +8,8 @@ enum class Encoding(val prefix: String) {
     ISO("iso"),                 // cx16  (iso-8859-15)
     ISO5("iso5"),               // cx16  (iso-8859-5, cyrillic)
     ISO16("iso16"),             // cx16  (iso-8859-16, eastern european)
-    CP437("cp437")              // cx16  (ibm pc, codepage 437)
+    CP437("cp437"),             // cx16  (ibm pc, codepage 437)
+    KATAKANA("kata")                 // cx16  (katakana)
 }
 
 interface IStringEncoding {

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

@@ -10,8 +10,8 @@ fun optimizeIntermediateAst(program: PtProgram, options: CompilationOptions, st:
     if (!options.optimize)
         return
     while (errors.noErrors() &&
-        (optimizeCommonSubExpressions(program, errors)
-                + optimizeAssignTargets(program, st, errors)) > 0
+            (optimizeBitTest(program, options)
+            + optimizeAssignTargets(program, st, errors)) > 0
     ) {
         // keep rolling
     }
@@ -27,97 +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 ->
@@ -164,6 +73,51 @@ private fun optimizeAssignTargets(program: PtProgram, st: SymbolTable, errors: I
     return changes
 }
 
+
+private fun optimizeBitTest(program: PtProgram, options: CompilationOptions): Int {
+    if(options.compTarget.machine.cpu == CpuType.VIRTUAL)
+        return 0        // the special bittest optimization is not yet valid for the IR
+
+    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
+                        }
+                    }
+                }
+            }
+        }
+        recurse
+    }
+    return changes
+}
+
+
 internal fun isSame(identifier: PtIdentifier, type: DataType, returnedRegister: RegisterOrPair): Boolean {
     if(returnedRegister in Cx16VirtualRegisters) {
         val regname = returnedRegister.name.lowercase()
@@ -189,24 +143,3 @@ 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/Encoder.kt

@@ -19,6 +19,7 @@ object Encoder: IStringEncoding {
             Encoding.ISO5 -> IsoCyrillicEncoding.encode(str)
             Encoding.ISO16 -> IsoEasternEncoding.encode(str)
             Encoding.CP437 -> Cp437Encoding.encode(str)
+            Encoding.KATAKANA -> KatakanaEncoding.encode(str)
             else -> throw InternalCompilerException("unsupported encoding $encoding")
         }
         return coded.fold(
@@ -35,6 +36,7 @@ object Encoder: IStringEncoding {
             Encoding.ISO5 -> IsoCyrillicEncoding.decode(bytes)
             Encoding.ISO16 -> IsoEasternEncoding.decode(bytes)
             Encoding.CP437 -> Cp437Encoding.decode(bytes)
+            Encoding.KATAKANA -> KatakanaEncoding.decode(bytes)
             else -> throw InternalCompilerException("unsupported encoding $encoding")
         }
         return decoded.fold(

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

@@ -25,13 +25,6 @@ class AtariMachineDefinition: IMachineDefinition {
     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 importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return if (compilerOptions.output == OutputType.XEX)
-            listOf("syslib")
-        else
-            emptyList()
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         val emulatorName: String
         val cmdline: List<String>

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

@@ -36,13 +36,6 @@ class C128MachineDefinition: IMachineDefinition {
         return m5.toDouble()
     }
 
-    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            listOf("syslib")
-        else
-            emptyList()
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         if(selectedEmulator!=1) {
             System.err.println("The c128 target only supports the main emulator (Vice).")

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

@@ -37,13 +37,6 @@ class C64MachineDefinition: IMachineDefinition {
         return m5.toDouble()
     }
 
-    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            listOf("syslib")
-        else
-            emptyList()
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         if(selectedEmulator!=1) {
             System.err.println("The c64 target only supports the main emulator (Vice).")

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

@@ -36,13 +36,6 @@ class CX16MachineDefinition: IMachineDefinition {
         return m5.toDouble()
     }
 
-    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            listOf("syslib")
-        else
-            emptyList()
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         val emulator: String
         val extraArgs: List<String>

codeCore/src/prog8/code/target/encodings/KatakanaEncoding.kt

@@ -0,0 +1,122 @@
+package prog8.code.target.encodings
+
+import com.github.michaelbull.result.Err
+import com.github.michaelbull.result.Ok
+import com.github.michaelbull.result.Result
+import java.io.CharConversionException
+import java.nio.charset.Charset
+
+
+object JapaneseCharacterConverter {
+    // adapted from https://github.com/raminduw/Japanese-Character-Converter
+
+    private val ZENKAKU_KATAKANA = charArrayOf(
+        'ァ', 'ア', 'ィ', 'イ', 'ゥ',
+        'ウ', 'ェ', 'エ', 'ォ', 'オ', 'カ', 'ガ', 'キ', 'ギ', 'ク', 'グ', 'ケ', 'ゲ',
+        'コ', 'ゴ', 'サ', 'ザ', 'シ', 'ジ', 'ス', 'ズ', 'セ', 'ゼ', 'ソ', 'ゾ', 'タ',
+        'ダ', 'チ', 'ヂ', 'ッ', 'ツ', 'ヅ', 'テ', 'デ', 'ト', 'ド', 'ナ', 'ニ', 'ヌ',
+        'ネ', 'ノ', 'ハ', 'バ', 'パ', 'ヒ', 'ビ', 'ピ', 'フ', 'ブ', 'プ', 'ヘ', 'ベ',
+        'ペ', 'ホ', 'ボ', 'ポ', 'マ', 'ミ', 'ム', 'メ', 'モ', 'ャ', 'ヤ', 'ュ', 'ユ',
+        'ョ', 'ヨ', 'ラ', 'リ', 'ル', 'レ', 'ロ', 'ヮ', 'ワ', 'ヰ', 'ヱ', 'ヲ', 'ン',
+        'ヴ', 'ヵ', 'ヶ'
+    )
+
+    private val HANKAKU_HIRAGANA = charArrayOf(
+        'ぁ', 'あ', 'ぃ', 'い', 'ぅ', 'う', 'ぇ', 'え',
+        'ぉ', 'お', 'か', 'が', 'き', 'ぎ', 'く', 'ぐ',
+        'け', 'げ', 'こ', 'ご', 'さ', 'ざ', 'し', 'じ',
+        'す', 'ず', 'せ', 'ぜ', 'そ', 'ぞ', 'た', 'だ',
+        'ち', 'ぢ', 'っ', 'つ', 'づ', 'て', 'で', 'と',
+        'ど', 'な', 'に', 'ぬ', 'ね', 'の', 'は', 'ば',
+        'ぱ', 'ひ', 'び', 'ぴ', 'ふ', 'ぶ', 'ぷ', 'へ',
+        'べ', 'ぺ', 'ほ', 'ぼ', 'ぽ', 'ま', 'み', 'む',
+        'め', 'も', 'ゃ', 'や', 'ゅ', 'ゆ', 'ょ', 'よ',
+        'ら', 'り', 'る', 'れ', 'ろ', 'ゎ', 'わ', 'ゐ',
+        'ゑ', 'を', 'ん', 'ゔ', 'ゕ', 'ゖ'
+    )
+
+    private val HANKAKU_KATAKANA = arrayOf(
+        "ｧ", "ｱ", "ｨ", "ｲ", "ｩ",
+        "ｳ", "ｪ", "ｴ", "ｫ", "ｵ", "ｶ", "ｶﾞ", "ｷ", "ｷﾞ", "ｸ", "ｸﾞ", "ｹ",
+        "ｹﾞ", "ｺ", "ｺﾞ", "ｻ", "ｻﾞ", "ｼ", "ｼﾞ", "ｽ", "ｽﾞ", "ｾ", "ｾﾞ", "ｿ",
+        "ｿﾞ", "ﾀ", "ﾀﾞ", "ﾁ", "ﾁﾞ", "ｯ", "ﾂ", "ﾂﾞ", "ﾃ", "ﾃﾞ", "ﾄ", "ﾄﾞ",
+        "ﾅ", "ﾆ", "ﾇ", "ﾈ", "ﾉ", "ﾊ", "ﾊﾞ", "ﾊﾟ", "ﾋ", "ﾋﾞ", "ﾋﾟ", "ﾌ",
+        "ﾌﾞ", "ﾌﾟ", "ﾍ", "ﾍﾞ", "ﾍﾟ", "ﾎ", "ﾎﾞ", "ﾎﾟ", "ﾏ", "ﾐ", "ﾑ", "ﾒ",
+        "ﾓ", "ｬ", "ﾔ", "ｭ", "ﾕ", "ｮ", "ﾖ", "ﾗ", "ﾘ", "ﾙ", "ﾚ", "ﾛ", "ﾜ",
+        "ﾜ", "ｲ", "ｴ", "ｦ", "ﾝ", "ｳﾞ", "ｶ", "ｹ"
+    )
+
+    private val ZENKAKU_KATAKANA_FIRST_CHAR_CODE = ZENKAKU_KATAKANA.first().code
+    private val HANKAKU_HIRAGANA_FIRST_CHAR_CODE = HANKAKU_HIRAGANA.first().code
+
+    private fun zenkakuKatakanaToHankakuKatakana(c: Char): String = if (c in ZENKAKU_KATAKANA) HANKAKU_KATAKANA[c.code - ZENKAKU_KATAKANA_FIRST_CHAR_CODE] else c.toString()
+    private fun hankakuKatakanaToZenkakuKatakana(c: Char): Char = if (c in HANKAKU_HIRAGANA) ZENKAKU_KATAKANA[c.code - HANKAKU_HIRAGANA_FIRST_CHAR_CODE] else c
+
+    fun zenkakuKatakanaToHankakuKatakana(s: String): String = buildString {
+        for (element in s) {
+            val converted = hankakuKatakanaToZenkakuKatakana(element)
+            val convertedChar = zenkakuKatakanaToHankakuKatakana(converted)
+            append(convertedChar)
+        }
+    }
+}
+
+object KatakanaEncoding {
+    val charset: Charset = Charset.forName("JIS_X0201")
+
+    fun encode(str: String): Result<List<UByte>, CharConversionException> {
+        return try {
+            val mapped = str.map { chr ->
+                when (chr) {
+
+                    '\u0000' -> 0u
+                    '\u00a0' -> 0xa0u // $a0 isn't technically a part of JIS X 0201 spec, and so we need to handle this ourselves
+
+                    '♥' -> 0xe3u
+                    '♦' -> 0xe4u
+                    '♣' -> 0xe5u
+                    '♠' -> 0xe6u
+
+                    '大' -> 0xeau
+                    '中' -> 0xebu
+                    '小' -> 0xecu
+                    '百' -> 0xedu
+                    '千' -> 0xeeu
+                    '万' -> 0xefu
+                    '♪' -> 0xf0u
+                    '土' -> 0xf1u
+                    '金' -> 0xf2u
+                    '木' -> 0xf3u
+                    '水' -> 0xf4u
+                    '火' -> 0xf5u
+                    '月' -> 0xf6u
+                    '日' -> 0xf7u
+                    '時' -> 0xf8u
+                    '分' -> 0xf9u
+                    '秒' -> 0xfau
+                    '年' -> 0xfbu
+                    '円' -> 0xfcu
+                    '人' -> 0xfdu
+                    '生' -> 0xfeu
+                    '〒' -> 0xffu
+                    in '\u8000'..'\u80ff' -> {
+                        // special case: take the lower 8 bit hex value directly
+                        (chr.code - 0x8000).toUByte()
+                    }
+                    else -> charset.encode(chr.toString())[0].toUByte()
+                }
+            }
+            Ok(mapped)
+        } catch (ce: CharConversionException) {
+            Err(ce)
+        }
+    }
+
+    fun decode(bytes: Iterable<UByte>): Result<String, CharConversionException> {
+        return try {
+            Ok(String(bytes.map { it.toByte() }.toByteArray(), charset))
+        } catch (ce: CharConversionException) {
+            Err(ce)
+        }
+    }
+}

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

@@ -36,13 +36,6 @@ class PETMachineDefinition: IMachineDefinition {
         return m5.toDouble()
     }
 
-    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return if (compilerOptions.launcher == CbmPrgLauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            listOf("syslib")
-        else
-            emptyList()
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         if(selectedEmulator!=1) {
             System.err.println("The pet target only supports the main emulator (Vice).")

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

@@ -50,10 +50,6 @@ class VirtualMachineDefinition: IMachineDefinition {
         return Double.fromBits(b0 or b1 or b2 or b3 or b4 or b5 or b6 or b7)
     }
 
-    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
-        return listOf("syslib")
-    }
-
     override fun launchEmulator(selectedEmulator: Int, programNameWithPath: Path) {
         println("\nStarting Virtual Machine...")
         // to not have external module dependencies in our own module, we launch the virtual machine via reflection

codeGenCpu6502/build.gradle

@@ -27,7 +27,8 @@ dependencies {
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:2.0.0"
 
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.0'
+    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 +9,7 @@ import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.code.target.Cx16Target
 import prog8.codegen.cpu6502.assignment.*
-import java.util.*
+import kotlin.collections.ArrayDeque
 import kotlin.io.path.Path
 import kotlin.io.path.writeLines
 
@@ -104,7 +104,7 @@ 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")
             }
         }
@@ -135,7 +135,7 @@ private fun prefixScopedName(name: String, type: Char): String {
     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
@@ -163,7 +163,9 @@ 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, newValue, arraySize, position)
+        result.parent = parent
+        result
     }
     else this
 }
@@ -232,7 +234,7 @@ 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 +376,7 @@ 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 +387,14 @@ 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
@@ -772,7 +774,7 @@ class AsmGen6502Internal (
 
     private fun translate(stmt: PtRepeatLoop) {
         val endLabel = makeLabel("repeatend")
-        loopEndLabels.push(endLabel)
+        loopEndLabels.add(endLabel)
 
         when (stmt.count) {
             is PtNumber -> {
@@ -816,7 +818,7 @@ class AsmGen6502Internal (
             }
         }
 
-        loopEndLabels.pop()
+        loopEndLabels.removeLast()
     }
 
     private fun repeatWordCount(iterations: Int, stmt: PtRepeatLoop) {

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

@@ -50,6 +50,13 @@ internal fun optimizeAssembly(lines: MutableList<String>, machine: IMachineDefin
         numberOfOptimizations++
     }
 
+    mods = optimizeTSBtoRegularOr(linesByFour)
+    if(mods.isNotEmpty()) {
+        apply(mods, lines)
+        linesByFour = getLinesBy(lines, 4)
+        numberOfOptimizations++
+    }
+
     var linesByFourteen = getLinesBy(lines, 14)
     mods = optimizeSameAssignments(linesByFourteen, machine, symbolTable)
     if(mods.isNotEmpty()) {
@@ -384,6 +391,7 @@ private fun optimizeStoreLoadSame(
     for (lines in linesByFour) {
         val first = lines[1].value.trimStart()
         val second = lines[2].value.trimStart()
+        val third = lines[3].value.trimStart()
 
         // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can OFTEN be eliminated
         if ((first.startsWith("sta ") && second.startsWith("lda ")) ||
@@ -393,7 +401,6 @@ private fun optimizeStoreLoadSame(
                 (first.startsWith("ldy ") && second.startsWith("ldy ")) ||
                 (first.startsWith("ldx ") && second.startsWith("ldx "))
         ) {
-            val third = lines[3].value.trimStart()
             val attemptRemove =
                 if(third.isBranch()) {
                     // a branch instruction follows, we can only remove the load instruction if
@@ -508,34 +515,36 @@ private fun optimizeJsrRtsAndOtherCombinations(linesByFour: Sequence<List<Indexe
     for (lines in linesByFour) {
         val first = lines[0].value
         val second = lines[1].value
-        if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {
-            mods += Modification(lines[0].index, false, lines[0].value.replace("jsr", "jmp"))
-            mods += Modification(lines[1].index, true, null)
-        }
-        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)
-        }
+        val third = lines[2].value
+
+        if(!haslabel(second)) {
+            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 ((" 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)
@@ -577,6 +586,58 @@ private fun optimizeJsrRtsAndOtherCombinations(linesByFour: Sequence<List<Indexe
                 mods += Modification(lines[3].index, true, null)
             }
         }
+
+
+        fun sameLabel(branchInstr: String, jumpInstr: String, labelInstr: String): Boolean {
+            if('(' in jumpInstr) return false       // indirect jump cannot be replaced
+            val label = labelInstr.trimEnd().substringBefore(':').substringBefore(' ').substringBefore('\t')
+            val branchLabel = branchInstr.trimStart().substring(3).trim()
+            return label==branchLabel
+        }
+
+        // beq Label + jmp Addr + Label  -> bne Addr
+        if((" jmp" in second || "\tjmp " in second) && haslabel(third)) {
+            if((" beq " in first || "\tbeq " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bne")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bne " in first || "\tbne " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "beq")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bcc " in first || "\tbcc " in first) && sameLabel(first, second, third)){
+                val branch = second.replace("jmp", "bcs")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bcs " in first || "\tbcs " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bcc")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bpl " in first || "\tbpl " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bmi")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bmi " in first || "\tbmi " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bpl")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bvc " in first || "\tbvc " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bvs")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+            else if((" bvs " in first || "\tbvs " in first) && sameLabel(first, second, third)) {
+                val branch = second.replace("jmp", "bvc")
+                mods.add(Modification(lines[0].index, true, null))
+                mods.add(Modification(lines[1].index, false, branch))
+            }
+        }
     }
     return mods
 }
@@ -614,6 +675,52 @@ private fun optimizeUselessPushPopStack(linesByFour: Sequence<List<IndexedValue<
         optimize('a', lines)
         optimize('x', lines)
         optimize('y', lines)
+
+        val first = lines[1].value.trimStart()
+        val second = lines[2].value.trimStart()
+        val third = lines[3].value.trimStart()
+
+        // phy + ldy + pla -> tya + ldy
+        // phx + ldx + pla -> txa + ldx
+        // pha + lda + pla -> nop
+        if(first=="phy" && second.startsWith("ldy ") && third=="pla") {
+            mods.add(Modification(lines[3].index, true, null))
+            mods.add(Modification(lines[1].index, false, "  tya"))
+        }
+        else if(first=="phx" && second.startsWith("ldx ") && third=="pla") {
+            mods.add(Modification(lines[3].index, true, null))
+            mods.add(Modification(lines[1].index, false, "  txa"))
+        }
+        else if(first=="pha" && second.startsWith("lda ") && third=="pla") {
+            mods.add(Modification(lines[1].index, true, null))
+            mods.add(Modification(lines[2].index, true, null))
+            mods.add(Modification(lines[3].index, true, null))
+        }
+    }
+
+
+    return mods
+}
+
+
+private fun optimizeTSBtoRegularOr(linesByFour: Sequence<List<IndexedValue<String>>>): List<Modification> {
+    // Asm peephole:   lda var2 / tsb var1 / lda var1  Replace this with this to save 1 cycle:   lda var1 / ora var2 / sta var1
+    val mods = mutableListOf<Modification>()
+
+    for(lines in linesByFour) {
+        val first = lines[0].value.trimStart()
+        val second = lines[1].value.trimStart()
+        val third = lines[2].value.trimStart()
+        if(first.startsWith("lda") && second.startsWith("tsb") && third.startsWith("lda")) {
+            val operand1 = first.substring(3)
+            val operand2 = second.substring(3)
+            val operand3 = third.substring(3)
+            if(operand1!=operand2 && operand2==operand3) {
+                mods.add(Modification(lines[0].index, false, "  lda  $operand2"))
+                mods.add(Modification(lines[1].index, false, "  ora  $operand1"))
+                mods.add(Modification(lines[2].index, false, "  sta  $operand2"))
+            }
+        }
     }
     return mods
 }

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

@@ -33,7 +33,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(fcall, resultRegister)
             "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(fcall, resultRegister)
             "abs__byte", "abs__word", "abs__float" -> funcAbs(fcall, resultRegister, sscope)
-            "any", "all" -> funcAnyAll(fcall, resultRegister, sscope)
             "sgn" -> funcSgn(fcall, resultRegister, sscope)
             "sqrt__ubyte", "sqrt__uword", "sqrt__float" -> funcSqrt(fcall, resultRegister, sscope)
             "divmod__ubyte" -> funcDivmod(fcall)
@@ -44,8 +43,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "ror2" -> funcRor2(fcall)
             "setlsb" -> funcSetLsbMsb(fcall, false)
             "setmsb" -> funcSetLsbMsb(fcall, true)
-            "sort" -> funcSort(fcall)
-            "reverse" -> funcReverse(fcall)
             "memory" -> funcMemory(fcall, discardResult, resultRegister)
             "peekw" -> funcPeekW(fcall, resultRegister)
             "peekf" -> funcPeekF(fcall, resultRegister)
@@ -67,7 +64,10 @@ 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)
@@ -267,33 +267,110 @@ 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")
 
-        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""")
+        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  cx16.JSRFAR
+                .word  ${constAddress.toHex()}
+                .byte  $constBank""")
+        } else {
+            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // bank
+            asmgen.out("  sta  (++)+0")
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)     // jump address
+            asmgen.out("  sta  (+)+0 |  sty  (+)+1")
+            asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.AY)     // uword argument
+            asmgen.out("""
+                jsr  cx16.JSRFAR
++               .word  0
++               .byte  0""")
+        }
+
+        // note that by convention the values in A+Y registers are now the return value of the call.
+        if(resultRegister!=null)  {
+            assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister, false, fcall.position, null, asmgen), RegisterOrPair.AY)
+        }
+    }
+
+    private fun funcCallFar2(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        if(asmgen.options.compTarget.name != "cx16")
+            throw AssemblyError("callfar2 only works on cx16 target at this time")
+
+        fun assignArgs() {
+            fun assign(value: PtExpression, register: Char) {
+                when(value) {
+                    is PtBool -> asmgen.out("  ld$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 constBank = fcall.args[0].asConstInteger()
+        val constAddress = fcall.args[1].asConstInteger()
+        if(constBank!=null && constAddress!=null) {
+            assignArgs()
+            asmgen.out("""
+                jsr  cx16.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  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)
@@ -403,115 +480,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         }
     }
 
-    private fun funcReverse(fcall: PtBuiltinFunctionCall) {
-        val variable = fcall.args.single() as PtIdentifier
-        val symbol = asmgen.symbolTable.lookup(variable.name)
-        val (dt, numElements) = when(symbol) {
-            is StStaticVariable  -> symbol.dt to symbol.length!!
-            is StMemVar -> symbol.dt to symbol.length!!
-            else -> DataType.UNDEFINED to 0
-        }
-        val varName = asmgen.asmVariableName(variable)
-        when (dt) {
-            DataType.ARRAY_UB, DataType.ARRAY_B -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements
-                    jsr  prog8_lib.func_reverse_b""")
-            }
-            DataType.ARRAY_UW, DataType.ARRAY_W -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements
-                    jsr  prog8_lib.func_reverse_w""")
-            }
-            DataType.STR -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #${numElements-1}
-                    jsr  prog8_lib.func_reverse_b""")
-            }
-            DataType.ARRAY_F -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements
-                    jsr  floats.func_reverse_f""")
-            }
-            in SplitWordArrayTypes -> {
-                // reverse the lsb and msb arrays both, independently
-                asmgen.out("""
-                    lda  #<${varName}_lsb
-                    ldy  #>${varName}_lsb
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements
-                    jsr  prog8_lib.func_reverse_b
-                    lda  #<${varName}_msb
-                    ldy  #>${varName}_msb
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements
-                    jsr  prog8_lib.func_reverse_b""")
-            }
-            else -> throw AssemblyError("weird type")
-        }
-    }
-
-    private fun funcSort(fcall: PtBuiltinFunctionCall) {
-        val variable = fcall.args.single() as PtIdentifier
-        val symbol = asmgen.symbolTable.lookup(variable.name)
-        val varName = asmgen.asmVariableName(variable)
-        val (dt, numElements) = when(symbol) {
-            is StStaticVariable  -> symbol.dt to symbol.length!!
-            is StMemVar -> symbol.dt to symbol.length!!
-            else -> DataType.UNDEFINED to 0
-        }
-        when (dt) {
-            DataType.ARRAY_UB, DataType.ARRAY_B -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements""")
-                asmgen.out(if (dt == DataType.ARRAY_UB) "  jsr  prog8_lib.func_sort_ub" else "  jsr  prog8_lib.func_sort_b")
-            }
-            DataType.ARRAY_UW, DataType.ARRAY_W -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #$numElements""")
-                asmgen.out(if (dt == DataType.ARRAY_UW) "  jsr  prog8_lib.func_sort_uw" else "  jsr  prog8_lib.func_sort_w")
-            }
-            DataType.STR -> {
-                asmgen.out("""
-                    lda  #<$varName
-                    ldy  #>$varName
-                    sta  P8ZP_SCRATCH_W1
-                    sty  P8ZP_SCRATCH_W1+1
-                    lda  #${numElements-1}
-                    jsr  prog8_lib.func_sort_ub""")
-            }
-            DataType.ARRAY_F -> throw AssemblyError("sorting of floating point array is not supported")
-            in SplitWordArrayTypes -> TODO("split words sort")
-            else -> throw AssemblyError("weird type")
-        }
-    }
-
     private fun funcRor2(fcall: PtBuiltinFunctionCall) {
         val what = fcall.args.single()
         when (what.type) {
@@ -837,63 +805,6 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, true, true)
     }
 
-    private fun funcAnyAll(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
-        val dt = fcall.args.single().type
-        val array = fcall.args[0] as PtIdentifier
-        when (dt) {
-            DataType.ARRAY_B, DataType.ARRAY_UB, DataType.STR -> {
-                outputAddressAndLengthOfArray(array)
-                asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_into_A")
-            }
-            DataType.ARRAY_UW, DataType.ARRAY_W -> {
-                outputAddressAndLengthOfArray(array)
-                asmgen.out("  jsr  prog8_lib.func_${fcall.name}_w_into_A")
-            }
-            DataType.ARRAY_F -> {
-                outputAddressAndLengthOfArray(array)
-                asmgen.out("  jsr  floats.func_${fcall.name}_f_into_A")
-            }
-            in SplitWordArrayTypes -> {
-                val numElements = (asmgen.symbolTable.lookup(array.name) as StStaticVariable).length
-                when(fcall.name) {
-                    "any" -> {
-                        // any(lsb-array) or any(msb-array)
-                        val arrayName = asmgen.asmVariableName(array)
-                        asmgen.out("""
-                            lda  #<${arrayName}_lsb
-                            ldy  #>${arrayName}_lsb
-                            sta  P8ZP_SCRATCH_W1
-                            sty  P8ZP_SCRATCH_W1+1
-                            lda  #$numElements
-                        """)
-                        asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_into_A")
-                        asmgen.out("  bne  +")      // shortcircuit
-                        asmgen.out("""
-                            pha
-                            lda  #<${arrayName}_msb
-                            ldy  #>${arrayName}_msb
-                            sta  P8ZP_SCRATCH_W1
-                            sty  P8ZP_SCRATCH_W1+1
-                            lda  #$numElements
-                        """)
-                        asmgen.out("  jsr  prog8_lib.func_${fcall.name}_b_into_A")
-                        asmgen.out("""
-                            sta  P8ZP_SCRATCH_REG
-                            pla
-                            ora  P8ZP_SCRATCH_REG
-+""")
-                    }
-                    "all" -> {
-                        TODO("split words all")
-                    }
-                    else -> throw AssemblyError("weird call")
-                }
-            }
-            else -> throw AssemblyError("weird type $dt")
-        }
-        assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, dt in SignedDatatypes, true)
-    }
-
     private fun funcAbs(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, scope: IPtSubroutine?) {
         translateArguments(fcall, scope)
         val dt = fcall.args.single().type
@@ -1289,7 +1200,44 @@ 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  $arrayVar+1,y")
+                        }
+                        else -> throw AssemblyError("invalid reg")
+                    }
+                }
+            } else when(resultRegister) {
                 null, RegisterOrPair.A -> {
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  tya")
@@ -1304,6 +1252,23 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  pla")
                 }
+                RegisterOrPair.AY -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
+                    asmgen.out("  tya |  ldy  #0 |  cmp  #0")
+                }
+                RegisterOrPair.AX -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.AX)
+                    asmgen.out("  txa |  ldx  #0 |  cmp  #0")
+                }
+                RegisterOrPair.XY -> {
+                    asmgen.assignExpressionToRegister(arg, RegisterOrPair.XY)
+                    asmgen.out("  tya |  tax |  ldy  #0 |  cpx  #0")
+                }
+                in Cx16VirtualRegisters -> {
+                    val reg = "cx16.${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")
             }
         }
@@ -1345,7 +1310,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     }
                     RegisterOrPair.Y -> {
                         asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.X)
-                        asmgen.out("  lda  $arrayVar,x")
+                        asmgen.out("  ldy  $arrayVar,x")
                     }
                     RegisterOrPair.X -> {
                         asmgen.loadScaledArrayIndexIntoRegister(arg, CpuRegister.Y)
@@ -1383,11 +1348,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.out("  ldy  #0 |  cpx  #0")
                 }
                 in Cx16VirtualRegisters -> {
-                    asmgen.assignExpressionToRegister(arg, resultRegister)
-                    val zero = PtNumber(DataType.UBYTE, 0.0, Position.DUMMY)
-                    zero.parent=fcall
-                    assignAsmGen.assignExpressionToVariable(zero, "cx16.${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")
             }

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

@@ -82,7 +82,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
             is PtAddressOf -> false
             is PtIdentifier -> false
             is PtIrRegister -> false
-            is PtMemoryByte -> return usesOtherRegistersWhileEvaluating(arg.address)
+            is PtMemoryByte -> arg.address !is PtNumber && arg.address !is PtIdentifier
             is PtNumber -> false
             is PtBool -> false
             else -> true
@@ -110,9 +110,9 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
                         throw AssemblyError("call argument evaluation problem: can't save cpu statusregister parameter ${call.position}")
                     }
                     else {
-                        if(usedX()) asmgen.saveRegisterStack(CpuRegister.X, false)
-                        if(usedY()) asmgen.saveRegisterStack(CpuRegister.Y, false)
-                        if(usedA()) asmgen.saveRegisterStack(CpuRegister.A, false)
+                        if(usedX()) asmgen.saveRegisterStack(CpuRegister.X, usedA())
+                        if(usedY()) asmgen.saveRegisterStack(CpuRegister.Y, usedA())
+                        if(usedA()) asmgen.saveRegisterStack(CpuRegister.A, usedA())
                         val used = argumentViaRegister(sub, IndexedValue(it, param.second), arg)
                         if(usedA()) asmgen.restoreRegisterStack(CpuRegister.A, false)
                         if(usedY()) asmgen.restoreRegisterStack(CpuRegister.Y, true)
@@ -205,7 +205,7 @@ 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 +213,7 @@ 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

@@ -83,9 +83,80 @@ internal class IfElseAsmGen(private val program: PtProgram,
     }
 
     private fun fallbackTranslateForSimpleCondition(ifElse: PtIfElse) {
+        val bittest = ifElse.condition as? PtBuiltinFunctionCall
+        val jumpAfterIf = ifElse.ifScope.children.singleOrNull() as? PtJump
+
+        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) {
+                        if(jumpAfterIf!=null) {
+                            val (asmLabel, indirect) = asmgen.getJumpTarget(jumpAfterIf)
+                            if(indirect)
+                                throw AssemblyError("cannot BIT to indirect label ${ifElse.position}")
+                            if(ifElse.hasElse())
+                                throw AssemblyError("didn't expect else part here ${ifElse.position}")
+                            else
+                                asmgen.out("  bmi  $asmLabel")
+                        }
+                        else
+                            translateIfElseBodies("bpl", ifElse)
+                    } else {
+                        if(jumpAfterIf!=null) {
+                            val (asmLabel, indirect) = asmgen.getJumpTarget(jumpAfterIf)
+                            if(indirect)
+                                throw AssemblyError("cannot BIT to indirect label ${ifElse.position}")
+                            if(ifElse.hasElse())
+                                throw AssemblyError("didn't expect else part here ${ifElse.position}")
+                            else
+                                asmgen.out("  bpl  $asmLabel")
+                        }
+                        else
+                            translateIfElseBodies("bmi", ifElse)
+                    }
+                    return
+                }
+                6 -> {
+                    // test via bit + V flag
+                    asmgen.out("  bit  ${variable.name}")
+                    if(testForBitSet) {
+                        if(jumpAfterIf!=null) {
+                            val (asmLabel, indirect) = asmgen.getJumpTarget(jumpAfterIf)
+                            if(indirect)
+                                throw AssemblyError("cannot BIT to indirect label ${ifElse.position}")
+                            if(ifElse.hasElse())
+                                throw AssemblyError("didn't expect else part here ${ifElse.position}")
+                            else
+                                asmgen.out("  bvs  $asmLabel")
+                        }
+                        else
+                            translateIfElseBodies("bvc", ifElse)
+                    } else {
+                        if(jumpAfterIf!=null) {
+                            val (asmLabel, indirect) = asmgen.getJumpTarget(jumpAfterIf)
+                            if(indirect)
+                                throw AssemblyError("cannot BIT to indirect label ${ifElse.position}")
+                            if(ifElse.hasElse())
+                                throw AssemblyError("didn't expect else part here ${ifElse.position}")
+                            else
+                                asmgen.out("  bvc  $asmLabel")
+                        }
+                        else
+                            translateIfElseBodies("bvs", ifElse)
+                    }
+                    return
+                }
+                else -> throw AssemblyError("prog8_ifelse_bittest can only work on bits 7 and 6")
+            }
+        }
+
         // the condition is "simple" enough to just assign its 0/1 value to a register and branch on that
         assignConditionValueToRegisterAndTest(ifElse.condition)
-        val jumpAfterIf = ifElse.ifScope.children.singleOrNull() as? PtJump
         if(jumpAfterIf!=null)
             translateJumpElseBodies("bne", "beq", jumpAfterIf, ifElse.elseScope)
         else
@@ -405,14 +476,14 @@ internal class IfElseAsmGen(private val program: PtProgram,
         val condition = stmt.condition as PtBinaryExpression
         if(signed) {
             // X>Y --> Y<X
-            asmgen.assignExpressionToRegister(condition.right, RegisterOrPair.A, signed)
+            asmgen.assignExpressionToRegister(condition.right, RegisterOrPair.A, true)
             cmpAwithByteValue(condition.left, true)
             if (jumpAfterIf != null)
                 translateJumpElseBodies("bmi", "bpl", jumpAfterIf, stmt.elseScope)
             else
                 translateIfElseBodies("bpl", stmt)
         } else {
-            asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.A, signed)
+            asmgen.assignExpressionToRegister(condition.left, RegisterOrPair.A, false)
             cmpAwithByteValue(condition.right, false)
             if(jumpAfterIf!=null) {
                 val (asmLabel, indirect) = asmgen.getJumpTarget(jumpAfterIf)
@@ -848,10 +919,10 @@ _jump                       jmp  ($asmLabel)
                     }
                 } else {
                     asmgen.loadScaledArrayIndexIntoRegister(value, CpuRegister.Y)
-                    if(value.splitWords) {
-                        return compareLsbMsb("${varname}_lsb,y", "${varname}_msb,y")
+                    return if(value.splitWords) {
+                        compareLsbMsb("${varname}_lsb,y", "${varname}_msb,y")
                     } else {
-                        return compareLsbMsb("$varname,y", "$varname+1,y")
+                        compareLsbMsb("$varname,y", "$varname+1,y")
                     }
                 }
             }
@@ -985,10 +1056,10 @@ _jump                       jmp  ($asmLabel)
                     }
                 } else {
                     asmgen.loadScaledArrayIndexIntoRegister(value, CpuRegister.Y)
-                    if(value.splitWords) {
-                        return compareLsbMsb("${varname}_lsb,y", "${varname}_msb,y")
+                    return if(value.splitWords) {
+                        compareLsbMsb("${varname}_lsb,y", "${varname}_msb,y")
                     } else {
-                        return compareLsbMsb("$varname,y", "$varname+1,y")
+                        compareLsbMsb("$varname,y", "$varname+1,y")
                     }
                 }
             }

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

@@ -89,6 +89,12 @@ internal class ProgramAndVarsGen(
             OutputType.RAW -> {
                 asmgen.out("; ---- raw assembler program ----")
                 asmgen.out("* = ${options.loadAddress.toHex()}")
+                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")
             }
             OutputType.PRG -> {
                 when(options.launcher) {
@@ -103,13 +109,20 @@ internal class ProgramAndVarsGen(
                         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("  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")
                     }
                     CbmPrgLauncherType.NONE -> {
+                        // this is the same as RAW
                         asmgen.out("; ---- program without basic sys call ----")
                         asmgen.out("* = ${options.loadAddress.toHex()}")
+                        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")
@@ -119,6 +132,9 @@ internal class ProgramAndVarsGen(
             OutputType.XEX -> {
                 asmgen.out("; ---- atari xex program ----")
                 asmgen.out("* = ${options.loadAddress.toHex()}")
+                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")
@@ -149,7 +165,10 @@ internal class ProgramAndVarsGen(
                 asmgen.out("  jsr  p8b_main.p8s_start |  lda  #0 |  sta ${"$"}ff00")
                 asmgen.out("  jmp  sys.cleanup_at_exit")
             }
-            else -> asmgen.jmp("p8b_main.p8s_start")
+            else -> {
+                asmgen.out("  jsr  p8b_main.p8s_start")
+                asmgen.out("  jmp  sys.cleanup_at_exit")
+            }
         }
     }
 
@@ -451,22 +470,10 @@ 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
+        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")
@@ -517,9 +524,7 @@ internal class ProgramAndVarsGen(
             arrayVariable2asm(varname, it.alloc.dt, it.value, null)
         }
 
-        asmgen.out("""+        
-                    clv
-                    clc""")
+        asmgen.out("+")
     }
 
     private class ZpStringWithInitial(
@@ -579,7 +584,7 @@ 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 {
                 uninitializedVariable2asm(it)
@@ -588,7 +593,7 @@ internal class ProgramAndVarsGen(
         }
 
         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 {
                 outputStringvar(

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

@@ -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) {
@@ -47,6 +48,26 @@ internal class AssignmentAsmGen(private val program: PtProgram,
 
         asmgen.translate(values)
 
+        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<StRomSubParameter, PtNode>>,
+        registersResults: List<Pair<StRomSubParameter, PtNode>>
+    ) {
+
         fun needsToSaveA(registersResults: List<Pair<StRomSubParameter, PtNode>>): Boolean =
             if(registersResults.isEmpty())
                 false
@@ -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<StRomSubParameter, 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<StRomSubParameter, 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<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
+            }
+        }
+    }
+
+    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)
@@ -284,12 +304,6 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 }
             }
             SourceStorageKind.MEMORY -> {
-                fun assignViaExprEval(expression: PtExpression) {
-                    assignExpressionToVariable(expression, "P8ZP_SCRATCH_W2", DataType.UWORD)
-                    asmgen.loadAFromZpPointerVar("P8ZP_SCRATCH_W2", false)
-                    assignRegisterByte(assign.target, CpuRegister.A, false, true)
-                }
-
                 val value = assign.source.memory!!
                 when (value.address) {
                     is PtNumber -> {
@@ -304,10 +318,10 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                         if(asmgen.tryOptimizedPointerAccessWithA(addrExpr, false)) {
                             assignRegisterByte(assign.target, CpuRegister.A, false, true)
                         } else {
-                            assignViaExprEval(value.address)
+                            assignByteFromAddressExpression(value.address, assign.target)
                         }
                     }
-                    else -> assignViaExprEval(value.address)
+                    else -> assignByteFromAddressExpression(value.address, assign.target)
                 }
             }
             SourceStorageKind.EXPRESSION -> {
@@ -319,6 +333,106 @@ internal class AssignmentAsmGen(private val program: PtProgram,
         }
     }
 
+    private fun assignByteFromAddressExpression(address: PtExpression, target: AsmAssignTarget) {
+        if(address is PtBinaryExpression) {
+            if(address.operator=="+" && address.right.type==DataType.UWORD) {
+                if (address.left is PtIdentifier) {
+                    // use (zp),Y instead of explicitly calculating the full zp pointer value
+                    val pointer = (address.left as PtIdentifier).name
+                    when(val index=address.right) {
+                        is PtIdentifier -> {
+                            val indexName = index.name
+                            asmgen.out("""
+                                lda  $pointer
+                                sta  P8ZP_SCRATCH_W2
+                                lda  $pointer+1
+                                clc
+                                adc  $indexName+1
+                                sta  P8ZP_SCRATCH_W2+1
+                                ldy  $indexName
+                                lda  (P8ZP_SCRATCH_W2),y""")
+                            assignRegisterByte(target, CpuRegister.A, false, true)
+                            return
+                        }
+                        is PtNumber -> {
+                            val indexValue = index.number.toInt().toString()
+                            asmgen.out("""
+                                lda  $pointer
+                                sta  P8ZP_SCRATCH_W2
+                                lda  $pointer+1
+                                clc
+                                adc  #>$indexValue
+                                sta  P8ZP_SCRATCH_W2+1
+                                ldy  #<$indexValue
+                                lda  (P8ZP_SCRATCH_W2),y""")
+                            assignRegisterByte(target, CpuRegister.A, false, true)
+                            return
+                        }
+                        else -> {}
+                    }
+                }
+            }
+//          else if(address.operator=="-") {
+//              // does this ever occur? we could optimize it too, but it seems like a pathological case
+//          }
+        }
+        assignExpressionToVariable(address, "P8ZP_SCRATCH_W2", DataType.UWORD)
+        asmgen.loadAFromZpPointerVar("P8ZP_SCRATCH_W2", false)
+        assignRegisterByte(target, CpuRegister.A, false, true)
+    }
+
+    private fun storeByteInAToAddressExpression(address: PtExpression, saveA: Boolean) {
+        if(address is PtBinaryExpression) {
+            if(address.operator=="+") {
+                if (address.left is PtIdentifier && address.right.type==DataType.UWORD) {
+                    // use (zp),Y instead of explicitly calculating the full zp pointer value
+                    val pointer = (address.left as PtIdentifier).name
+                    when(val index=address.right) {
+                        is PtIdentifier -> {
+                            val indexName = index.name
+                            asmgen.out("""
+                                tax
+                                lda  $pointer
+                                sta  P8ZP_SCRATCH_W2
+                                lda  $pointer+1
+                                clc
+                                adc  $indexName+1
+                                sta  P8ZP_SCRATCH_W2+1
+                                ldy  $indexName
+                                txa
+                                sta  (P8ZP_SCRATCH_W2),y""")
+                            return
+                        }
+                        is PtNumber -> {
+                            val indexValue = index.number.toInt().toString()
+                            asmgen.out("""
+                                tax
+                                lda  $pointer
+                                sta  P8ZP_SCRATCH_W2
+                                lda  $pointer+1
+                                clc
+                                adc  #>$indexValue
+                                sta  P8ZP_SCRATCH_W2+1
+                                ldy  #<$indexValue
+                                txa
+                                sta  (P8ZP_SCRATCH_W2),y""")
+                            return
+                        }
+                        else -> {}
+                    }
+                }
+            }
+//          else if(address.operator=="-") {
+//              // does this ever occur? we could optimize it too, but it seems like a pathological case
+//          }
+        }
+        if(saveA) asmgen.out("  pha")
+        assignExpressionToVariable(address, "P8ZP_SCRATCH_W2", DataType.UWORD)
+        if(saveA) asmgen.out("  pla")
+        asmgen.storeAIntoZpPointerVar("P8ZP_SCRATCH_W2", false)
+    }
+
+
     private fun assignExpression(assign: AsmAssignment, scope: IPtSubroutine?) {
         when(val value = assign.source.expression!!) {
             is PtAddressOf -> {
@@ -703,8 +817,8 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 RegisterOrPair.Y -> "y"
                 else -> TODO("comparison to word register")
             }
-            val assignTrue = PtInlineAssembly("  ld${reg}  #1", false, assign.target.position)
-            val assignFalse = PtInlineAssembly("  ld${reg}  #0", false, assign.target.position)
+            val assignTrue = PtInlineAssembly("\tld${reg}  #1", false, assign.target.position)
+            val assignFalse = PtInlineAssembly("\tld${reg}  #0", false, assign.target.position)
             ifPart.add(assignTrue)
             elsePart.add(assignFalse)
             val ifelse = PtIfElse(assign.position)
@@ -754,7 +868,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
   sta  ${assign.target.asmVarname}+1""", false, assign.target.position)
                         }
                     } else {
-                        assignTrue = PtInlineAssembly("  lda  #1\n  sta  ${assign.target.asmVarname}", false, assign.target.position)
+                        assignTrue = PtInlineAssembly("\tlda  #1\n  sta  ${assign.target.asmVarname}", false, assign.target.position)
                     }
                 }
                 TargetStorageKind.MEMORY -> {
@@ -839,6 +953,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
     }
 
     private fun optimizedDivideExpr(expr: PtBinaryExpression, target: AsmAssignTarget): Boolean {
+        // replacing division by shifting is done in an optimizer step.
         when(expr.type) {
             DataType.UBYTE -> {
                 assignExpressionToRegister(expr.left, RegisterOrPair.A, false)
@@ -1010,7 +1125,10 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                             }
                         } else {
                             if (signed && shifts > 0) {
-                                asmgen.out("  ldy  #$shifts |  jsr  math.lsr_byte_A")
+                                if(shifts==1)
+                                    asmgen.out("  cmp  #$80 |  ror  a")
+                                else
+                                    asmgen.out("  ldy  #$shifts |  jsr  math.lsr_byte_A")
                             } else {
                                 repeat(shifts) {
                                     asmgen.out("  lsr  a")
@@ -1044,7 +1162,18 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                             }
                         } else {
                             if(signed && shifts>0) {
-                                asmgen.out("  ldx  #$shifts |  jsr  math.lsr_word_AY")
+                                if(shifts==1) {
+                                    asmgen.out("""
+                                        pha
+                                        tya
+                                        cmp  #$80
+                                        ror  a
+                                        tay
+                                        pla
+                                        ror  a""")
+                                }
+                                else
+                                    asmgen.out("  ldx  #$shifts |  jsr  math.lsr_word_AY")
                             } else {
                                 if(shifts>0) {
                                     asmgen.out("  sty  P8ZP_SCRATCH_B1")
@@ -1108,10 +1237,17 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 }
                 is PtNumber -> {
                     assignExpressionToRegister(left, RegisterOrPair.A, dt==DataType.BYTE)
-                    if(expr.operator=="+")
-                        asmgen.out("  clc |  adc  #${right.number.toHex()}")
-                    else
-                        asmgen.out("  sec |  sbc  #${right.number.toHex()}")
+                    if(right.number==1.0 && asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                        if (expr.operator == "+")
+                            asmgen.out("  inc  a")
+                        else
+                            asmgen.out("  dec  a")
+                    } else {
+                        if (expr.operator == "+")
+                            asmgen.out("  clc |  adc  #${right.number.toHex()}")
+                        else
+                            asmgen.out("  sec |  sbc  #${right.number.toHex()}")
+                    }
                     assignRegisterByte(target, CpuRegister.A, dt in SignedDatatypes, true)
                     return true
                 }
@@ -1246,8 +1382,39 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 }
                 is PtNumber -> {
                     assignExpressionToRegister(left, RegisterOrPair.AY, dt==DataType.WORD)
-                    if(expr.operator=="+") {
-                        asmgen.out("""
+                    if(right.number==1.0 && asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                        if(expr.operator=="+") {
+                            asmgen.out("""
+                                inc  a
+                                bne  +
+                                iny
++""")
+                        } else {
+                            asmgen.out("""
+                                dec  a
+                                cmp  #255
+                                bne  +
+                                dey
++""")
+                        }
+                    } else if(dt!=DataType.WORD && right.number.toInt() in 0..255) {
+                        if(expr.operator=="+") {
+                            asmgen.out("""
+                                clc
+                                adc  #${right.number.toHex()}
+                                bcc  +
+                                iny
++""")                   } else if(expr.operator=="-") {
+                            asmgen.out("""
+                                sec
+                                sbc  #${right.number.toHex()}
+                                bcs  +
+                                dey
++""")
+                        }
+                    } else {
+                        if(expr.operator=="+") {
+                            asmgen.out("""
                                 clc
                                 adc  #<${right.number.toHex()}
                                 tax
@@ -1255,8 +1422,8 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                                 adc  #>${right.number.toHex()}
                                 tay
                                 txa""")
-                    } else if(expr.operator=="-") {
-                        asmgen.out("""
+                        } else if(expr.operator=="-") {
+                            asmgen.out("""
                                 sec
                                 sbc  #<${right.number.toHex()}
                                 tax
@@ -1264,6 +1431,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                                 sbc  #>${right.number.toHex()}
                                 tay
                                 txa""")
+                        }
                     }
                     assignRegisterpairWord(target, RegisterOrPair.AY)
                     return true
@@ -1335,11 +1503,16 @@ 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")
@@ -1657,9 +1830,56 @@ 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 $elementDt")
+            }
+
+            return
+        }
+
+        val symbol = asmgen.symbolTable.lookup(containment.haystackHeapVar!!.name)!!
+        val symbolName = asmgen.asmVariableName(symbol, containment.definingISub())
         val (dt, numElements) = when(symbol) {
             is StStaticVariable  -> symbol.dt to symbol.length!!
             is StMemVar -> symbol.dt to symbol.length!!
@@ -1667,7 +1887,7 @@ internal class AssignmentAsmGen(private val program: PtProgram,
         }
         when(dt) {
             DataType.STR -> {
-                assignExpressionToRegister(containment.element, RegisterOrPair.A, elementDt == DataType.BYTE)
+                assignExpressionToRegister(containment.needle, RegisterOrPair.A, elementDt == DataType.BYTE)
                 asmgen.out("  pha")     // need to keep the scratch var safe so we have to do it in this order
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position,"P8ZP_SCRATCH_W1"), symbolName, null, null)
                 asmgen.out("  pla")
@@ -1675,13 +1895,13 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                 asmgen.out("  jsr  prog8_lib.containment_bytearray")
             }
             DataType.ARRAY_F -> {
-                assignExpressionToRegister(containment.element, RegisterOrPair.FAC1, true)
+                assignExpressionToRegister(containment.needle, RegisterOrPair.FAC1, true)
                 assignAddressOf(AsmAssignTarget(TargetStorageKind.VARIABLE, asmgen, DataType.UWORD, containment.definingISub(), containment.position, "P8ZP_SCRATCH_W1"), symbolName, null, null)
                 asmgen.out("  ldy  #$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")
@@ -1689,7 +1909,7 @@ 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")
@@ -3143,11 +3363,20 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                     }
                     else {
                         if (regs !in Cx16VirtualRegisters) {
-                            when (regs) {
-                                RegisterOrPair.AX -> asmgen.out("  pha |  txa |  pha")
-                                RegisterOrPair.AY -> asmgen.out("  pha |  tya |  pha")
-                                RegisterOrPair.XY -> asmgen.out("  txa |  pha |  tya |  pha")
-                                else -> throw AssemblyError("expected reg pair")
+                            if (asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                                when (regs) {
+                                    RegisterOrPair.AX -> asmgen.out("  pha |  phx")
+                                    RegisterOrPair.AY -> asmgen.out("  pha |  phy")
+                                    RegisterOrPair.XY -> asmgen.out("  phx |  phy")
+                                    else -> throw AssemblyError("expected reg pair")
+                                }
+                            } else {
+                                when (regs) {
+                                    RegisterOrPair.AX -> asmgen.out("  pha |  txa |  pha")
+                                    RegisterOrPair.AY -> asmgen.out("  pha |  tya |  pha")
+                                    RegisterOrPair.XY -> asmgen.out("  txa |  pha |  tya |  pha")
+                                    else -> throw AssemblyError("expected reg pair")
+                                }
                             }
                             asmgen.loadScaledArrayIndexIntoRegister(target.array, CpuRegister.Y)
                             asmgen.out("""
@@ -3186,11 +3415,20 @@ internal class AssignmentAsmGen(private val program: PtProgram,
                     }
                     else {
                         if (regs !in Cx16VirtualRegisters) {
-                            when (regs) {
-                                RegisterOrPair.AX -> asmgen.out("  pha |  txa |  pha")
-                                RegisterOrPair.AY -> asmgen.out("  pha |  tya |  pha")
-                                RegisterOrPair.XY -> asmgen.out("  txa |  pha |  tya |  pha")
-                                else -> throw AssemblyError("expected reg pair")
+                            if (asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                                when (regs) {
+                                    RegisterOrPair.AX -> asmgen.out("  pha |  phx")
+                                    RegisterOrPair.AY -> asmgen.out("  pha |  phy")
+                                    RegisterOrPair.XY -> asmgen.out("  phx |  phy")
+                                    else -> throw AssemblyError("expected reg pair")
+                                }
+                            } else {
+                                when (regs) {
+                                    RegisterOrPair.AX -> asmgen.out("  pha |  txa |  pha")
+                                    RegisterOrPair.AY -> asmgen.out("  pha |  tya |  pha")
+                                    RegisterOrPair.XY -> asmgen.out("  txa |  pha |  tya |  pha")
+                                    else -> throw AssemblyError("expected reg pair")
+                                }
                             }
                             asmgen.loadScaledArrayIndexIntoRegister(target.array, CpuRegister.Y)
                             asmgen.out("""
@@ -3689,17 +3927,8 @@ internal class AssignmentAsmGen(private val program: PtProgram,
 
         fun storeViaExprEval() {
             when(addressExpr) {
-                is PtNumber, is PtIdentifier -> {
-                    assignExpressionToVariable(addressExpr, "P8ZP_SCRATCH_W2", DataType.UWORD)
-                    asmgen.storeAIntoZpPointerVar("P8ZP_SCRATCH_W2", false)
-                }
-                else -> {
-                    // same as above but we need to save the A register
-                    asmgen.out("  pha")
-                    assignExpressionToVariable(addressExpr, "P8ZP_SCRATCH_W2", DataType.UWORD)
-                    asmgen.out("  pla")
-                    asmgen.storeAIntoZpPointerVar("P8ZP_SCRATCH_W2", false)
-                }
+                is PtNumber, is PtIdentifier -> storeByteInAToAddressExpression(addressExpr, false)
+                else -> storeByteInAToAddressExpression(addressExpr, true)
             }
         }
 

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)
@@ -1086,6 +1086,21 @@ $shortcutLabel:""")
             }
         }
 
+        if(asmgen.isTargetCpu(CpuType.CPU65c02)) {
+            if(operator=="&" && value is PtPrefix && value.operator=="~") {
+                // M &= ~A  -->  use TRB 65c02 instruction for that
+                asmgen.assignExpressionToRegister(value.value, RegisterOrPair.A, dt in SignedDatatypes)
+                asmgen.out("  trb  $name")
+                return
+            }
+            else if(operator=="|") {
+                // M |= A --> use TSB 65c02 instruction for that
+                asmgen.assignExpressionToRegister(value, RegisterOrPair.A, dt in SignedDatatypes)
+                asmgen.out("  tsb  $name")
+                return
+            }
+        }
+
         // normal evaluation
         asmgen.assignExpressionToRegister(value, RegisterOrPair.A, dt in SignedDatatypes)
         inplacemodificationRegisterAwithVariableWithSwappedOperands(operator, name, dt in SignedDatatypes)
@@ -1094,6 +1109,15 @@ $shortcutLabel:""")
 
     private fun inplacemodificationByteVariableWithVariable(name: String, dt: DataType, operator: String, otherName: String) {
         // note: no logical and/or shortcut here, not worth it due to simple right operand
+
+        if(asmgen.isTargetCpu(CpuType.CPU65c02)) {
+            if(operator=="|") {
+                // M |= A  -->  use TSB 65c02 instruction for that
+                asmgen.out("  lda  $otherName |  tsb  $name")
+                return
+            }
+        }
+
         asmgen.out("  lda  $name")
         inplacemodificationRegisterAwithVariable(operator, otherName, dt in SignedDatatypes)
         asmgen.out("  sta  $name")
@@ -1459,6 +1483,7 @@ $shortcutLabel:""")
                     asmgen.out("  lda  $name |  ldy  #$value |  jsr  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")
                 else
@@ -1642,7 +1667,7 @@ $shortcutLabel:""")
     }
 
     private fun immediateOrInplace(name: String, value: Int) {
-        if(asmgen.isTargetCpu(CpuType.CPU65c02) && ((value and (value-1))==0)) {
+        if(asmgen.isTargetCpu(CpuType.CPU65c02)) {
             asmgen.out(" lda  #$value |  tsb  $name")       // set bit
         } else  {
             asmgen.out(" lda  $name |  ora  #$value |  sta  $name")
@@ -1807,33 +1832,36 @@ $shortcutLabel:""")
                 }
             }
             "/" -> {
-                if(value==0)
+                // replacing division by shifting is done in an optimizer step.
+                if(value==0) {
                     throw AssemblyError("division by zero")
-                if(dt==DataType.WORD) {
-                    asmgen.out("""
-                        lda  $lsb
-                        ldy  $msb
-                        sta  P8ZP_SCRATCH_W1
-                        sty  P8ZP_SCRATCH_W1+1
-                        lda  #<$value
-                        ldy  #>$value
-                        jsr  math.divmod_w_asm
-                        sta  $lsb
-                        sty  $msb
-                    """)
-                }
-                else {
-                    asmgen.out("""
-                        lda  $lsb
-                        ldy  $msb
-                        sta  P8ZP_SCRATCH_W1
-                        sty  P8ZP_SCRATCH_W1+1
-                        lda  #<$value
-                        ldy  #>$value
-                        jsr  math.divmod_uw_asm
-                        sta  $lsb
-                        sty  $msb
-                    """)
+                } else {
+                    if(dt==DataType.WORD) {
+                        asmgen.out("""
+                            lda  $lsb
+                            ldy  $msb
+                            sta  P8ZP_SCRATCH_W1
+                            sty  P8ZP_SCRATCH_W1+1
+                            lda  #<$value
+                            ldy  #>$value
+                            jsr  math.divmod_w_asm
+                            sta  $lsb
+                            sty  $msb
+                        """)
+                    }
+                    else {
+                        asmgen.out("""
+                            lda  $lsb
+                            ldy  $msb
+                            sta  P8ZP_SCRATCH_W1
+                            sty  P8ZP_SCRATCH_W1+1
+                            lda  #<$value
+                            ldy  #>$value
+                            jsr  math.divmod_uw_asm
+                            sta  $lsb
+                            sty  $msb
+                        """)
+                    }
                 }
             }
             "%" -> {

codeGenIntermediate/build.gradle

@@ -29,8 +29,9 @@ dependencies {
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation "com.michael-bull.kotlin-result:kotlin-result-jvm:2.0.0"
 
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.0'
+    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

@@ -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")
         }
@@ -1390,9 +1387,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 val valueReg = codeGen.registers.nextFree()
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADM, vmDt, reg1=valueReg, labelSymbol = array.variable.name, symbolOffset = constIndex*eltSize)
-                    when(comparisonOperator) {
-                        "==" -> it += IRInstruction(Opcode.SZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
-                        "!=" -> it += IRInstruction(Opcode.SNZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
+                    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?
@@ -1413,9 +1410,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 addToResult(result, indexTr, indexTr.resultReg, -1)
                 result += IRCodeChunk(null, null).also {
                     it += IRInstruction(Opcode.LOADX, vmDt, reg1=valueReg, reg2=indexTr.resultReg, labelSymbol = array.variable.name)
-                    when(comparisonOperator) {
-                        "==" -> it += IRInstruction(Opcode.SZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
-                        "!=" -> it += IRInstruction(Opcode.SNZ, vmDt, reg1=cmpResultReg, reg2=valueReg)
+                    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?

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

@@ -9,8 +9,6 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     fun translate(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         return when(call.name) {
-            "any" -> funcAny(call)
-            "all" -> funcAll(call)
             "abs__byte", "abs__word", "abs__float" -> funcAbs(call)
             "cmp" -> funcCmp(call)
             "sgn" -> funcSgn(call)
@@ -19,6 +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)
@@ -34,14 +33,14 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(call)
             "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(call)
             "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(call)
-            "sort" -> funcSort(call)
-            "reverse" -> funcReverse(call)
             "setlsb" -> funcSetLsbMsb(call, false)
             "setmsb" -> funcSetLsbMsb(call, true)
             "rol" -> funcRolRor(call)
             "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_lib_stringcompare" -> funcStringCompare(call)
             "prog8_lib_square_byte" -> funcSquare(call, IRDataType.BYTE)
             "prog8_lib_square_word" -> funcSquare(call, IRDataType.WORD)
@@ -132,10 +131,10 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val addressTr = exprGen.translateExpression(call.args[0])
         addToResult(result, addressTr, addressTr.resultReg, -1)
         addInstr(result, IRInstruction(Opcode.CALLI, reg1 = addressTr.resultReg), null)
-        if(call.void)
-            return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
+        return if(call.void)
+            ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
         else
-            return ExpressionCodeResult(result, IRDataType.WORD, codeGen.registers.nextFree(), -1)      // TODO actually the result is returned in CPU registers AY...
+            ExpressionCodeResult(result, IRDataType.WORD, codeGen.registers.nextFree(), -1)      // TODO actually the result is returned in CPU registers AY...
     }
 
     private fun funcCallfar(call: PtBuiltinFunctionCall): ExpressionCodeResult {
@@ -151,6 +150,29 @@ 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]
@@ -204,79 +226,6 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         return ExpressionCodeResult(result, dt, leftTr.resultReg, -1)
     }
 
-    private fun funcAny(call: PtBuiltinFunctionCall): ExpressionCodeResult {
-        val arrayName = call.args[0] as PtIdentifier
-        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
-        val result = mutableListOf<IRCodeChunkBase>()
-        val lengthReg = codeGen.registers.nextFree()
-
-        if(arrayName.type in SplitWordArrayTypes) {
-            // any(lsb-array) or any(msb-array)
-            addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-            val trLsb = exprGen.translateExpression(PtIdentifier(arrayName.name+"_lsb", DataType.ARRAY_UB, call.position))
-            addToResult(result, trLsb, trLsb.resultReg, -1)
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength), null)
-            result += codeGen.makeSyscall(IMSyscall.ANY_BYTE, listOf(IRDataType.WORD to trLsb.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to trLsb.resultReg)
-            val shortcircuitLabel = codeGen.createLabelName()
-            result += IRCodeChunk(null, null).also {
-                it += IRInstruction(Opcode.CMPI, IRDataType.BYTE, reg1 = trLsb.resultReg, immediate = 0)
-                it += IRInstruction(Opcode.BSTNE, labelSymbol = shortcircuitLabel)
-                it += IRInstruction(Opcode.PREPARECALL, immediate = 2)
-            }
-            val trMsb = exprGen.translateExpression(PtIdentifier(arrayName.name+"_msb", DataType.ARRAY_UB, call.position))
-            addToResult(result, trMsb, trMsb.resultReg, -1)
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength), null)
-            result += codeGen.makeSyscall(IMSyscall.ANY_BYTE, listOf(IRDataType.WORD to trMsb.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to trMsb.resultReg)
-            addInstr(result, IRInstruction(Opcode.ORR, IRDataType.BYTE, reg1=trLsb.resultReg, reg2=trMsb.resultReg), null)
-            result += IRCodeChunk(shortcircuitLabel, null)
-            return ExpressionCodeResult(result, IRDataType.BYTE, trLsb.resultReg, -1)
-        }
-
-        val syscall =
-            when (arrayName.type) {
-                DataType.ARRAY_UB,
-                DataType.ARRAY_B -> IMSyscall.ANY_BYTE
-                DataType.ARRAY_UW,
-                DataType.ARRAY_W -> IMSyscall.ANY_WORD
-                DataType.ARRAY_F -> IMSyscall.ANY_FLOAT
-                else -> throw IllegalArgumentException("weird type")
-            }
-        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-        val tr = exprGen.translateExpression(arrayName)
-        addToResult(result, tr, tr.resultReg, -1)
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength!! and 255), null)
-        result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to tr.resultReg)
-        return ExpressionCodeResult(result, IRDataType.BYTE, tr.resultReg, -1)
-    }
-
-    private fun funcAll(call: PtBuiltinFunctionCall): ExpressionCodeResult {
-        val arrayName = call.args[0] as PtIdentifier
-        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
-
-        if(arrayName.type in SplitWordArrayTypes) {
-            // this is a bit complicated to calculate.... have to check all recombined (lsb,msb) words for $0000
-            TODO("all(split words $arrayName)")
-        }
-
-        val syscall =
-            when(arrayName.type) {
-                DataType.ARRAY_UB,
-                DataType.ARRAY_B -> IMSyscall.ALL_BYTE
-                DataType.ARRAY_UW,
-                DataType.ARRAY_W -> IMSyscall.ALL_WORD
-                DataType.ARRAY_F -> IMSyscall.ALL_FLOAT
-                else -> throw IllegalArgumentException("weird type")
-            }
-        val result = mutableListOf<IRCodeChunkBase>()
-        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-        val tr = exprGen.translateExpression(arrayName)
-        addToResult(result, tr, tr.resultReg, -1)
-        val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = arrayLength!! and 255), null)
-        result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), IRDataType.BYTE to tr.resultReg)
-        return ExpressionCodeResult(result, IRDataType.BYTE, tr.resultReg, -1)
-    }
-
     private fun funcAbs(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val sourceDt = call.args.single().type
         val result = mutableListOf<IRCodeChunkBase>()
@@ -369,65 +318,6 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         }
     }
 
-    private fun funcReverse(call: PtBuiltinFunctionCall): ExpressionCodeResult {
-        val arrayName = call.args[0] as PtIdentifier
-        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
-        val lengthReg = codeGen.registers.nextFree()
-        val result = mutableListOf<IRCodeChunkBase>()
-
-        if(arrayName.type in SplitWordArrayTypes) {
-            // reverse the lsb and msb arrays both, independently
-            addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-            val trLsb = exprGen.translateExpression(PtIdentifier(arrayName.name+"_lsb", DataType.ARRAY_UB, call.position))
-            addToResult(result, trLsb, trLsb.resultReg, -1)
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
-            result += codeGen.makeSyscall(IMSyscall.REVERSE_BYTES, listOf(IRDataType.WORD to trLsb.resultReg, IRDataType.BYTE to lengthReg), null)
-            val trMsb = exprGen.translateExpression(PtIdentifier(arrayName.name+"_msb", DataType.ARRAY_UB, call.position))
-            addToResult(result, trMsb, trMsb.resultReg, -1)
-            addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
-            result += codeGen.makeSyscall(IMSyscall.REVERSE_BYTES, listOf(IRDataType.WORD to trMsb.resultReg, IRDataType.BYTE to lengthReg), null)
-            return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
-        }
-
-        val syscall =
-            when(arrayName.type) {
-                DataType.ARRAY_UB, DataType.ARRAY_B, DataType.STR -> IMSyscall.REVERSE_BYTES
-                DataType.ARRAY_UW, DataType.ARRAY_W -> IMSyscall.REVERSE_WORDS
-                DataType.ARRAY_F -> IMSyscall.REVERSE_FLOATS
-                else -> throw IllegalArgumentException("weird type to reverse")
-            }
-        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-        val tr = exprGen.translateExpression(arrayName)
-        addToResult(result, tr, tr.resultReg, -1)
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
-        result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), null)
-        return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
-    }
-
-    private fun funcSort(call: PtBuiltinFunctionCall): ExpressionCodeResult {
-        val arrayName = call.args[0] as PtIdentifier
-        val arrayLength = codeGen.symbolTable.getLength(arrayName.name)
-        val syscall =
-            when(arrayName.type) {
-                DataType.ARRAY_UB -> IMSyscall.SORT_UBYTE
-                DataType.ARRAY_B -> IMSyscall.SORT_BYTE
-                DataType.ARRAY_UW -> IMSyscall.SORT_UWORD
-                DataType.ARRAY_W -> IMSyscall.SORT_WORD
-                DataType.STR -> IMSyscall.SORT_UBYTE
-                DataType.ARRAY_F -> throw IllegalArgumentException("sorting a floating point array is not supported")
-                in SplitWordArrayTypes -> TODO("split word sort")
-                else -> throw IllegalArgumentException("weird type to sort")
-            }
-        val result = mutableListOf<IRCodeChunkBase>()
-        addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = 2), null)
-        val tr = exprGen.translateExpression(arrayName)
-        addToResult(result, tr, tr.resultReg, -1)
-        val lengthReg = codeGen.registers.nextFree()
-        addInstr(result, IRInstruction(Opcode.LOAD, IRDataType.BYTE, reg1 = lengthReg, immediate = if(arrayName.type==DataType.STR) arrayLength!!-1 else arrayLength), null)
-        result += codeGen.makeSyscall(syscall, listOf(IRDataType.WORD to tr.resultReg, IRDataType.BYTE to lengthReg), null)
-        return ExpressionCodeResult(result, IRDataType.BYTE, -1, -1)
-    }
-
     private fun funcMkword(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val resultReg = codeGen.registers.nextFree()
@@ -714,6 +604,7 @@ 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

@@ -1,5 +1,6 @@
 package prog8.codegen.intermediate
 
+import prog8.code.StNode
 import prog8.code.StRomSub
 import prog8.code.StSub
 import prog8.code.ast.*
@@ -167,55 +168,93 @@ 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}")
         }
     }
 
@@ -451,7 +490,36 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
     }
 
     fun translate(fcall: PtFunctionCall): ExpressionCodeResult {
-        when (val callTarget = codeGen.symbolTable.flat.getValue(fcall.name)) {
+        val callTarget = codeGen.symbolTable.flat.getValue(fcall.name)
+
+        if(callTarget.scopedName in listOf("sys.push", "sys.pushw", "sys.pop", "sys.popw")) {
+            // special case, these should be inlined, or even use specialized instructions. Instead of doing a normal subroutine call.
+            return translateStackFunctions(fcall, callTarget)
+        }
+        when(callTarget.scopedName) {
+            "sys.clear_carry" -> {
+                val chunk = mutableListOf<IRCodeChunkBase>()
+                addInstr(chunk, IRInstruction(Opcode.CLC), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, -1, -1)
+            }
+            "sys.set_carry" -> {
+                val chunk = mutableListOf<IRCodeChunkBase>()
+                addInstr(chunk, IRInstruction(Opcode.SEC), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, -1, -1)
+            }
+            "sys.clear_irqd" -> {
+                val chunk = mutableListOf<IRCodeChunkBase>()
+                addInstr(chunk, IRInstruction(Opcode.CLI), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, -1, -1)
+            }
+            "sys.set_irqd" -> {
+                val chunk = mutableListOf<IRCodeChunkBase>()
+                addInstr(chunk, IRInstruction(Opcode.SEI), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, -1, -1)
+            }
+        }
+
+        when (callTarget) {
             is StSub -> {
                 val result = mutableListOf<IRCodeChunkBase>()
                 addInstr(result, IRInstruction(Opcode.PREPARECALL, immediate = callTarget.parameters.size), null)
@@ -590,6 +658,39 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         }
     }
 
+    private fun translateStackFunctions(fcall: PtFunctionCall, callTarget: StNode): ExpressionCodeResult {
+        val chunk = mutableListOf<IRCodeChunkBase>()
+        when(callTarget.scopedName) {
+            "sys.push" -> {
+                // push byte
+                val tr = translateExpression(fcall.args.single())
+                chunk += tr.chunks
+                addInstr(chunk, IRInstruction(Opcode.PUSH, IRDataType.BYTE, reg1=tr.resultReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, -1, -1)
+            }
+            "sys.pushw" -> {
+                // push word
+                val tr = translateExpression(fcall.args.single())
+                chunk += tr.chunks
+                addInstr(chunk, IRInstruction(Opcode.PUSH, IRDataType.WORD, reg1=tr.resultReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.WORD, -1, -1)
+            }
+            "sys.pop" -> {
+                // pop byte
+                val popReg = codeGen.registers.nextFree()
+                addInstr(chunk, IRInstruction(Opcode.POP, IRDataType.BYTE, reg1=popReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.BYTE, popReg, -1)
+            }
+            "sys.popw" -> {
+                // pop word
+                val popReg = codeGen.registers.nextFree()
+                addInstr(chunk, IRInstruction(Opcode.POP, IRDataType.WORD, reg1=popReg), null)
+                return ExpressionCodeResult(chunk, IRDataType.WORD, popReg, -1)
+            }
+            else -> throw AssemblyError("unknown stack subroutine called")
+        }
+    }
+
     private fun callRomSubWithMultipleReturnValues(
         callTarget: StRomSub,
         fcall: PtFunctionCall,

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

@@ -5,7 +5,6 @@ import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.intermediate.*
 import kotlin.io.path.readBytes
-import kotlin.math.pow
 
 
 class IRCodeGen(
@@ -391,10 +390,10 @@ class IRCodeGen(
         val result = mutableListOf<IRCodeChunkBase>()
         when(iterable) {
             is PtRange -> {
-                if(iterable.from is PtNumber && iterable.to is PtNumber)
-                    result += translateForInConstantRange(forLoop, loopvar)
+                result += if(iterable.from is PtNumber && iterable.to is PtNumber)
+                    translateForInConstantRange(forLoop, loopvar)
                 else
-                    result += translateForInNonConstantRange(forLoop, loopvar)
+                    translateForInNonConstantRange(forLoop, loopvar)
             }
             is PtIdentifier -> {
                 require(forLoop.variable.name == loopvar.scopedName)
@@ -488,49 +487,76 @@ 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
     }
 
@@ -559,9 +585,25 @@ 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
     }
@@ -682,13 +724,11 @@ class IRCodeGen(
         return code
     }
 
-    internal val powersOfTwo = (0..16).map { 2.0.pow(it.toDouble()).toInt() }
-
     internal fun multiplyByConst(dt: IRDataType, reg: Int, factor: Int): IRCodeChunk {
         val code = IRCodeChunk(null, null)
         if(factor==1)
             return code
-        val pow2 = powersOfTwo.indexOf(factor)
+        val pow2 = powersOfTwoInt.indexOf(factor)
         if(pow2==1) {
             // just shift 1 bit
             code += IRInstruction(Opcode.LSL, dt, reg1=reg)
@@ -712,7 +752,7 @@ class IRCodeGen(
         val code = IRCodeChunk(null, null)
         if(factor==1)
             return code
-        val pow2 = powersOfTwo.indexOf(factor)
+        val pow2 = powersOfTwoInt.indexOf(factor)
         if(pow2==1) {
             // just shift 1 bit
             code += if(knownAddress!=null)
@@ -785,19 +825,32 @@ class IRCodeGen(
         val code = IRCodeChunk(null, null)
         if(factor==1)
             return code
-        val pow2 = powersOfTwo.indexOf(factor)
-        if(pow2==1 && !signed) {
-            code += IRInstruction(Opcode.LSR, dt, reg1=reg)     // simple single bit shift
-        }
-        else if(pow2>=1 &&!signed) {
-            // just shift multiple bits
-            val pow2reg = registers.nextFree()
-            code += IRInstruction(Opcode.LOAD, dt, reg1=pow2reg, immediate = pow2)
-            code += if(signed)
-                IRInstruction(Opcode.ASRN, dt, reg1=reg, reg2=pow2reg)
-            else
-                IRInstruction(Opcode.LSRN, dt, reg1=reg, reg2=pow2reg)
+        val pow2 = powersOfTwoInt.indexOf(factor)
+        if(pow2>=0) {
+            if(signed) {
+                if(pow2==1) {
+                    // simple single bit shift (signed)
+                    code += IRInstruction(Opcode.ASR, dt, reg1=reg)
+                } else {
+                    // just shift multiple bits (signed)
+                    val pow2reg = registers.nextFree()
+                    code += IRInstruction(Opcode.LOAD, dt, reg1=pow2reg, immediate = pow2)
+                    code += IRInstruction(Opcode.ASRN, dt, reg1=reg, reg2=pow2reg)
+                }
+            } else {
+                if(pow2==1) {
+                    // simple single bit shift (unsigned)
+                    code += IRInstruction(Opcode.LSR, dt, reg1=reg)
+                } else {
+                    // just shift multiple bits (unsigned)
+                    val pow2reg = registers.nextFree()
+                    code += IRInstruction(Opcode.LOAD, dt, reg1 = pow2reg, immediate = pow2)
+                    code += IRInstruction(Opcode.LSRN, dt, reg1 = reg, reg2 = pow2reg)
+                }
+            }
+            return code
         } else {
+            // regular div
             code += if (factor == 0) {
                 IRInstruction(Opcode.LOAD, dt, reg1=reg, immediate = 0xffff)
             } else {
@@ -806,39 +859,56 @@ class IRCodeGen(
                 else
                     IRInstruction(Opcode.DIV, dt, reg1=reg, immediate = factor)
             }
+            return code
         }
-        return code
     }
 
     internal fun divideByConstInplace(dt: IRDataType, knownAddress: Int?, symbol: String?, factor: Int, signed: Boolean): IRCodeChunk {
         val code = IRCodeChunk(null, null)
         if(factor==1)
             return code
-        val pow2 = powersOfTwo.indexOf(factor)
-        if(pow2==1 && !signed) {
-            // just simple bit shift
-            code += if(knownAddress!=null)
-                IRInstruction(Opcode.LSRM, dt, address = knownAddress)
-            else
-                IRInstruction(Opcode.LSRM, dt, labelSymbol = symbol)
+        val pow2 = powersOfTwoInt.indexOf(factor)
+        if(pow2>=0) {
+            // can do bit shift instead of division
+            if(signed) {
+                if(pow2==1) {
+                    // just simple bit shift (signed)
+                    code += if (knownAddress != null)
+                        IRInstruction(Opcode.ASRM, dt, address = knownAddress)
+                    else
+                        IRInstruction(Opcode.ASRM, dt, labelSymbol = symbol)
+                } else {
+                    // just shift multiple bits (signed)
+                    val pow2reg = registers.nextFree()
+                    code += IRInstruction(Opcode.LOAD, dt, reg1 = pow2reg, immediate = pow2)
+                    code += if (knownAddress != null)
+                                IRInstruction(Opcode.ASRNM, dt, reg1 = pow2reg, address = knownAddress)
+                            else
+                                IRInstruction(Opcode.ASRNM, dt, reg1 = pow2reg, labelSymbol = symbol)
+                }
+            } else {
+                if(pow2==1) {
+                    // just simple bit shift (unsigned)
+                    code += if(knownAddress!=null)
+                        IRInstruction(Opcode.LSRM, dt, address = knownAddress)
+                    else
+                        IRInstruction(Opcode.LSRM, dt, labelSymbol = symbol)
+                }
+                else {
+                    // just shift multiple bits (unsigned)
+                    val pow2reg = registers.nextFree()
+                    code += IRInstruction(Opcode.LOAD, dt, reg1=pow2reg, immediate = pow2)
+                    code += if(knownAddress!=null)
+                                IRInstruction(Opcode.LSRNM, dt, reg1 = pow2reg, address = knownAddress)
+                            else
+                                IRInstruction(Opcode.LSRNM, dt, reg1 = pow2reg, labelSymbol = symbol)
+                }
+            }
+            return code
         }
-        else if(pow2>=1 && !signed) {
-            // just shift multiple bits
-            val pow2reg = registers.nextFree()
-            code += IRInstruction(Opcode.LOAD, dt, reg1=pow2reg, immediate = pow2)
-            code += if(signed) {
-                if(knownAddress!=null)
-                    IRInstruction(Opcode.ASRNM, dt, reg1 = pow2reg, address = knownAddress)
-                else
-                    IRInstruction(Opcode.ASRNM, dt, reg1 = pow2reg, labelSymbol = symbol)
-            }
-            else {
-                if(knownAddress!=null)
-                    IRInstruction(Opcode.LSRNM, dt, reg1 = pow2reg, address = knownAddress)
-                else
-                    IRInstruction(Opcode.LSRNM, dt, reg1 = pow2reg, labelSymbol = symbol)
-            }
-        } else {
+        else
+        {
+            // regular div
             if (factor == 0) {
                 val reg = registers.nextFree()
                 code += IRInstruction(Opcode.LOAD, dt, reg1=reg, immediate = 0xffff)
@@ -863,8 +933,8 @@ class IRCodeGen(
                         IRInstruction(Opcode.DIVM, dt, reg1 = factorReg, labelSymbol = symbol)
                 }
             }
+            return code
         }
-        return code
     }
 
     private fun translate(ifElse: PtIfElse): IRCodeChunks {
@@ -1291,6 +1361,10 @@ class IRCodeGen(
         val result = mutableListOf<IRCodeChunkBase>()
 
         fun translateSimple(condition: PtExpression, jumpFalseOpcode: Opcode) {
+
+            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)
             result += tr.chunks
             if(ifElse.hasElse()) {
@@ -1503,17 +1577,16 @@ class IRCodeGen(
     private fun translate(jump: PtJump): IRCodeChunks {
         val result = mutableListOf<IRCodeChunkBase>()
         val chunk = IRCodeChunk(null, null)
-        if(jump.address!=null) {
-            chunk += IRInstruction(Opcode.JUMP, address = jump.address!!.toInt())
+        chunk += if(jump.address!=null) {
+            IRInstruction(Opcode.JUMP, address = jump.address!!.toInt())
         } else {
             if (jump.identifier != null) {
                 if(isIndirectJump(jump)) {
-                    chunk += IRInstruction(Opcode.JUMPI, labelSymbol = jump.identifier!!.name)
+                    IRInstruction(Opcode.JUMPI, labelSymbol = jump.identifier!!.name)
                 } else {
-                    chunk += IRInstruction(Opcode.JUMP, labelSymbol = jump.identifier!!.name)
+                    IRInstruction(Opcode.JUMP, labelSymbol = jump.identifier!!.name)
                 }
-            }
-            else
+            } else
                 throw AssemblyError("weird jump")
         }
         result += chunk
@@ -1618,7 +1691,7 @@ 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

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

@@ -263,6 +263,24 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
                     }
                 }
             }
+
+            if(ins.opcode== Opcode.SEI || ins.opcode== Opcode.CLI) {
+                if(idx < chunk.instructions.size-1) {
+                    val insAfter = chunk.instructions[idx+1]
+                    if(insAfter.opcode == ins.opcode) {
+                        chunk.instructions.removeAt(idx)
+                        changed = true
+                    }
+                    else if(ins.opcode== Opcode.SEI && insAfter.opcode== Opcode.CLI) {
+                        chunk.instructions.removeAt(idx)
+                        changed = true
+                    }
+                    else if(ins.opcode== Opcode.CLI && insAfter.opcode== Opcode.SEI) {
+                        chunk.instructions.removeAt(idx)
+                        changed = true
+                    }
+                }
+            }
         }
         return changed
     }

codeGenIntermediate/test/TestIRPeepholeOpt.kt

@@ -80,21 +80,52 @@ class TestIRPeepholeOpt: FunSpec({
         instr[1].opcode shouldBe Opcode.INC
     }
 
-    test("remove double sec/clc") {
+    test("remove double sec/clc/sei/cli") {
         val irProg = makeIRProgram(listOf(
             IRInstruction(Opcode.SEC),
             IRInstruction(Opcode.SEC),
             IRInstruction(Opcode.SEC),
             IRInstruction(Opcode.CLC),
             IRInstruction(Opcode.CLC),
-            IRInstruction(Opcode.CLC)
+            IRInstruction(Opcode.CLC),
+            IRInstruction(Opcode.SEI),
+            IRInstruction(Opcode.SEI),
+            IRInstruction(Opcode.SEI),
+            IRInstruction(Opcode.CLI),
+            IRInstruction(Opcode.CLI),
+            IRInstruction(Opcode.CLI),
         ))
-        irProg.chunks().single().instructions.size shouldBe 6
+        irProg.chunks().single().instructions.size shouldBe 12
         val opt = IRPeepholeOptimizer(irProg)
         opt.optimize(true, ErrorReporterForTests())
         val instr = irProg.chunks().single().instructions
-        instr.size shouldBe 1
+        instr.size shouldBe 2
         instr[0].opcode shouldBe Opcode.CLC
+        instr[1].opcode shouldBe Opcode.CLI
+    }
+
+    test("remove double sec/clc/sei/cli reversed") {
+        val irProg = makeIRProgram(listOf(
+            IRInstruction(Opcode.CLC),
+            IRInstruction(Opcode.CLC),
+            IRInstruction(Opcode.CLC),
+            IRInstruction(Opcode.SEC),
+            IRInstruction(Opcode.SEC),
+            IRInstruction(Opcode.SEC),
+            IRInstruction(Opcode.CLI),
+            IRInstruction(Opcode.CLI),
+            IRInstruction(Opcode.CLI),
+            IRInstruction(Opcode.SEI),
+            IRInstruction(Opcode.SEI),
+            IRInstruction(Opcode.SEI),
+        ))
+        irProg.chunks().single().instructions.size shouldBe 12
+        val opt = IRPeepholeOptimizer(irProg)
+        opt.optimize(true, ErrorReporterForTests())
+        val instr = irProg.chunks().single().instructions
+        instr.size shouldBe 2
+        instr[0].opcode shouldBe Opcode.SEC
+        instr[1].opcode shouldBe Opcode.SEI
     }
 
     test("push followed by pop") {

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

@@ -33,6 +33,14 @@ 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)) {
@@ -91,7 +99,7 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
                         program.encoding.encodeString(leftString.value, leftString.encoding) + program.encoding.encodeString(rightString.value, rightString.encoding),
                         leftString.encoding)
                 }
-                val concatStr = StringLiteral(concatenated, leftString.encoding, expr.position)
+                val concatStr = StringLiteral.create(concatenated, leftString.encoding, expr.position)
                 return listOf(IAstModification.ReplaceNode(expr, concatStr, parent))
             }
             else if(expr.operator=="*" && rightconst!=null && expr.left is StringLiteral) {
@@ -99,7 +107,7 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
                 val part = expr.left as StringLiteral
                 if(part.value.isEmpty())
                     errors.warn("resulting string has length zero", part.position)
-                val newStr = StringLiteral(part.value.repeat(rightconst.number.toInt()), part.encoding, expr.position)
+                val newStr = StringLiteral.create(part.value.repeat(rightconst.number.toInt()), part.encoding, expr.position)
                 return listOf(IAstModification.ReplaceNode(expr, newStr, parent))
             }
         }
@@ -164,13 +172,13 @@ 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))
             }
@@ -384,12 +392,13 @@ 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

@@ -154,11 +154,10 @@ class VarConstantValueTypeAdjuster(
         if(func==listOf("clamp")) {
             val t1 = functionCallExpr.args[0].inferType(program)
             if(t1.isKnown) {
-                val replaceFunc: String
-                if(t1.isBytes) {
-                    replaceFunc = if(t1.istype(DataType.BYTE)) "clamp__byte" else "clamp__ubyte"
+                val replaceFunc = if(t1.isBytes) {
+                    if(t1.istype(DataType.BYTE)) "clamp__byte" else "clamp__ubyte"
                 } else if(t1.isInteger) {
-                    replaceFunc = if(t1.istype(DataType.WORD)) "clamp__word" else "clamp__uword"
+                    if(t1.istype(DataType.WORD)) "clamp__word" else "clamp__uword"
                 } else {
                     errors.err("clamp builtin not supported for floats, use floats.clamp", functionCallExpr.position)
                     return noModifications
@@ -393,11 +392,35 @@ 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, 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.
 
+        val rangeSize=(decl.value as? RangeExpression)?.size()
+        if(rangeSize!=null && rangeSize>65535) {
+            errors.err("range size overflow", decl.value!!.position)
+            return null
+        }
+
         // convert the initializer range expression from a range or int, to an actual array.
         when(decl.datatype) {
             DataType.ARRAY_UB, DataType.ARRAY_B, DataType.ARRAY_UW, DataType.ARRAY_W, DataType.ARRAY_W_SPLIT, DataType.ARRAY_UW_SPLIT -> {
@@ -491,8 +514,7 @@ internal class ConstantIdentifierReplacer(
                 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)
+                        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>()

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -11,12 +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() {
-    private val powersOfTwo = (1..16).map { (2.0).pow(it) }.toSet()
-    private val negativePowersOfTwo = powersOfTwo.map { -it }.toSet()
 
+class ExpressionSimplifier(private val program: Program, private val errors: IErrorReporter) : AstWalker() {
     override fun after(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
         val mods = mutableListOf<IAstModification>()
 
@@ -104,7 +101,7 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
         val leftIDt = expr.left.inferType(program)
         val rightIDt = expr.right.inferType(program)
         if (!leftIDt.isKnown || !rightIDt.isKnown)
-            throw FatalAstException("can't determine datatype of both expression operands $expr")
+            return noModifications
 
         // X + (-A)  -->  X - A
         if (expr.operator == "+" && (expr.right as? PrefixExpression)?.operator == "-") {
@@ -249,20 +246,15 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                 }
             }
             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))
                 }
             }
         }
@@ -277,21 +269,16 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                 }
             }
             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))
+                }
             }
         }
 
@@ -686,7 +673,7 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                     if(!idt.isKnown)
                         throw FatalAstException("unknown dt")
                     return NumericLiteral(idt.getOr(DataType.UNDEFINED), 0.0, expr.position)
-                } else if (cv in powersOfTwo) {
+                } else if (cv in powersOfTwoFloat) {
                     expr.operator = "&"
                     expr.right = NumericLiteral.optimalInteger(cv!!.toInt()-1, expr.position)
                     return null
@@ -711,6 +698,7 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                 return null
             val leftDt = leftIDt.getOr(DataType.UNDEFINED)
             when (cv) {
+                0.0 -> return null // fall through to regular float division to properly deal with division by zero
                 -1.0 -> {
                     //  '/' -> -left
                     if (expr.operator == "/") {
@@ -738,12 +726,10 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                         else -> return null
                     }
                 }
-                in powersOfTwo -> {
-                    if (leftDt==DataType.UBYTE || leftDt==DataType.UWORD) {
-                        // Unsigned number divided by a power of two => shift right
-                        // Signed number can't simply be bitshifted in this case (due to rounding issues for negative values),
-                        // so we leave that as is and let the code generator deal with it.
-                        val numshifts = log2(cv).toInt()
+                in powersOfTwoFloat -> {
+                    val numshifts = powersOfTwoFloat.indexOf(cv)
+                    if (leftDt in IntegerDatatypes) {
+                        // division by a power of two => shift right (signed and unsigned)
                         return BinaryExpression(expr.left, ">>", NumericLiteral.optimalInteger(numshifts, expr.position), expr.position)
                     }
                 }
@@ -795,14 +781,14 @@ class ExpressionSimplifier(private val program: Program, private val options: Co
                     // left
                     return expr2.left
                 }
-                in powersOfTwo -> {
+                in powersOfTwoFloat -> {
                     if (leftValue.inferType(program).isInteger) {
                         // times a power of two => shift left
                         val numshifts = log2(cv).toInt()
                         return BinaryExpression(expr2.left, "<<", NumericLiteral.optimalInteger(numshifts, expr.position), expr.position)
                     }
                 }
-                in negativePowersOfTwo -> {
+                in negativePowersOfTwoFloat -> {
                     if (leftValue.inferType(program).isInteger) {
                         // times a negative power of two => negate, then shift
                         val numshifts = log2(-cv).toInt()

codeOptimizers/src/prog8/optimizer/Extensions.kt

@@ -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

@@ -30,11 +30,11 @@ class Inliner(private val program: Program, private val options: CompilationOpti
         }
 
         override fun visit(subroutine: Subroutine) {
-            if(!subroutine.isAsmSubroutine && !subroutine.inline && subroutine.parameters.isEmpty()) {
-                val containsSubsOrVariables = subroutine.statements.any { it is VarDecl || it is Subroutine}
-                if(!containsSubsOrVariables) {
-                    if(subroutine.statements.size==1 || (subroutine.statements.size==2 && isEmptyReturn(subroutine.statements[1]))) {
-                        if(subroutine !== program.entrypoint) {
+            if (!subroutine.isAsmSubroutine && !subroutine.inline && subroutine.parameters.isEmpty()) {
+                val containsSubsOrVariables = subroutine.statements.any { it is VarDecl || it is Subroutine }
+                if (!containsSubsOrVariables) {
+                    if (subroutine.statements.size == 1 || (subroutine.statements.size == 2 && isEmptyReturn(subroutine.statements[1]))) {
+                        if (subroutine !== program.entrypoint) {
                             // subroutine is possible candidate to be inlined
                             subroutine.inline =
                                 when (val stmt = subroutine.statements[0]) {
@@ -87,9 +87,9 @@ class Inliner(private val program: Program, private val options: CompilationOpti
                                                 } else
                                                     false
                                             targetInline || valueInline
-                                        } else if(stmt.target.identifier!=null && stmt.isAugmentable) {
+                                        } else if (stmt.target.identifier != null && stmt.isAugmentable) {
                                             val binExpr = stmt.value as BinaryExpression
-                                            if(binExpr.operator in "+-" && binExpr.right.constValue(program)?.number==1.0) {
+                                            if (binExpr.operator in "+-" && binExpr.right.constValue(program)?.number == 1.0) {
                                                 makeFullyScoped(stmt.target.identifier!!)
                                                 makeFullyScoped(binExpr.left as IdentifierReference)
                                                 true
@@ -121,8 +121,8 @@ class Inliner(private val program: Program, private val options: CompilationOpti
                         }
                     }
 
-                    if(subroutine.inline && subroutine.statements.size>1) {
-                        require(subroutine.statements.size==2 && isEmptyReturn(subroutine.statements[1]))
+                    if (subroutine.inline && subroutine.statements.size > 1) {
+                        require(subroutine.statements.size == 2 && isEmptyReturn(subroutine.statements[1]))
                         subroutine.statements.removeLast()      // get rid of the Return, to be able to inline the (single) statement preceding it.
                     }
                 }
@@ -147,6 +147,7 @@ class Inliner(private val program: Program, private val options: CompilationOpti
         }
 
         private fun makeFullyScoped(call: FunctionCallStatement) {
+            makeFullyScoped(call.target)
             call.target.targetSubroutine(program)?.let { sub ->
                 val scopedName = IdentifierReference(sub.scopedName, call.target.position)
                 val scopedArgs = makeScopedArgs(call.args)
@@ -169,6 +170,7 @@ class Inliner(private val program: Program, private val options: CompilationOpti
         }
 
         private fun makeFullyScoped(call: FunctionCallExpression) {
+            makeFullyScoped(call.target)
             call.target.targetSubroutine(program)?.let { sub ->
                 val scopedName = IdentifierReference(sub.scopedName, call.target.position)
                 val scopedArgs = makeScopedArgs(call.args)
@@ -222,8 +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
                     }
                 }
             }
@@ -253,7 +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

@@ -21,7 +21,8 @@ class StatementOptimizer(private val program: Program,
         if(functionCallStatement.target.nameInSource.size==1) {
             val functionName = functionCallStatement.target.nameInSource[0]
             if (functionName in functions.purefunctionNames) {
-                errors.warn("statement has no effect (function return value is discarded)", functionCallStatement.position)
+                if("ignore_unused" !in parent.definingBlock.options())
+                    errors.warn("statement has no effect (function return value is discarded)", functionCallStatement.position)
                 return listOf(IAstModification.Remove(functionCallStatement, parent as IStatementContainer))
             }
         }
@@ -199,35 +200,6 @@ 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

@@ -149,7 +149,8 @@ class UnusedCodeRemover(private val program: Program,
                                         val declIndex = (parent as IStatementContainer).statements.indexOf(decl)
                                         val singleUseIndex = (parent as IStatementContainer).statements.indexOf(singleUse.parent)
                                         if(declIndex==singleUseIndex-1) {
-                                            errors.info("replaced unused variable '${decl.name}' with void call, maybe this can be removed altogether", decl.position)
+                                            if("ignore_unused" !in decl.definingBlock.options())
+                                                errors.info("replaced unused variable '${decl.name}' with void call, maybe this can be removed altogether", decl.position)
                                             val fcall = assignment.value as IFunctionCall
                                             val voidCall = FunctionCallStatement(fcall.target, fcall.args, true, fcall.position)
                                             return listOf(

compiler/build.gradle

@@ -2,7 +2,8 @@ plugins {
     id 'java'
     id 'application'
     id 'org.jetbrains.kotlin.jvm'
-    id 'com.github.johnrengelman.shadow' version '8.1.1'
+    // id 'com.github.johnrengelman.shadow' version '8.1.1'
+    id 'io.github.goooler.shadow' version '8.1.7'
     id 'com.peterabeles.gversion' version '1.10.2'
 }
 
@@ -31,7 +32,7 @@ dependencies {
     implementation project(':codeGenIntermediate')
     implementation project(':codeGenExperimental')
     implementation project(':virtualmachine')
-    implementation "org.antlr:antlr4-runtime:4.13.1"
+    implementation "org.antlr:antlr4-runtime:4.13.2"
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
     implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.6'
@@ -39,7 +40,8 @@ dependencies {
 
     testImplementation project(':codeCore')
     testImplementation project(':intermediate')
-    testImplementation 'io.kotest:kotest-runner-junit5-jvm:5.9.0'
+    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'
 

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/anyall.p8

@@ -0,0 +1,82 @@
+; any() and all()  checks on arrays/memory buffers.
+; These were builtin functions in older versions of the language.
+
+%option no_symbol_prefixing, ignore_unused
+
+anyall {
+    sub any(uword arrayptr, uword num_elements) -> bool {
+        ; -- returns true if any byte in the array is not zero.
+        cx16.r1 = arrayptr
+        if msb(num_elements)==0 {
+            for cx16.r0L in 0 to lsb(num_elements)-1 {
+                if cx16.r1[cx16.r0L]!=0
+                    return true
+            }
+            return false
+        }
+        repeat num_elements {
+            if @(cx16.r1)!=0
+                return true
+            cx16.r1++
+        }
+        return false
+    }
+
+    sub all(uword arrayptr, uword num_elements) -> bool {
+        ; -- returns true if all bytes in the array are not zero.
+        cx16.r1 = arrayptr
+        if msb(num_elements)==0 {
+            for cx16.r0L in 0 to lsb(num_elements)-1 {
+                if cx16.r1[cx16.r0L]==0
+                    return false
+            }
+            return true
+        }
+        repeat num_elements {
+            if @(cx16.r1)==0
+                return false
+            cx16.r1++
+        }
+        return true
+    }
+
+    sub anyw(uword arrayptr, uword num_elements) -> bool {
+        ; -- returns true if any word in the array is not zero.
+        ;    doesn't work on @split arrays.
+        cx16.r1 = arrayptr
+        if msb(num_elements)==0 {
+            repeat lsb(num_elements) {
+                if peekw(cx16.r1)!=0
+                    return true
+                cx16.r1+=2
+            }
+            return false
+        }
+        repeat num_elements {
+            if peekw(cx16.r1)!=0
+                return true
+            cx16.r1+=2
+        }
+        return false
+    }
+
+    sub allw(uword arrayptr, uword num_elements) -> bool {
+        ; -- returns true if all words in the array are not zero.
+        ;    doesn't work on @split arrays.
+        cx16.r1 = arrayptr
+        if msb(num_elements)==0 {
+            repeat lsb(num_elements) {
+                if peekw(cx16.r1)==0
+                    return false
+                cx16.r1+=2
+            }
+            return true
+        }
+        repeat num_elements {
+            if peekw(cx16.r1)==0
+                return false
+            cx16.r1+=2
+        }
+        return true
+    }
+}

compiler/res/prog8lib/atari/syslib.p8

@@ -23,16 +23,20 @@ sys {
 
     const ubyte target = 8         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 8 = atari800XL
 
+    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  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
         }}
@@ -40,7 +44,26 @@ sys {
 
     asmsub  init_system_phase2()  {
         %asm {{
-            rts     ; no phase 2 steps on the Atari
+            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
         }}
     }
 
@@ -248,27 +271,38 @@ 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  cleanup_at_exit._exitcode
             ldx  prog8_lib.orig_stackpointer
             txs
-            rts		; return to original caller
+            jmp  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  cleanup_at_exit._exitcode
+            stx  cleanup_at_exit._exitcodeX
+            sty  cleanup_at_exit._exitcodeY
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            jmp  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  cleanup_at_exit._exitcode
+            lda  #0
+            rol  a
+            sta  cleanup_at_exit._exitcodeCarry
+            stx  cleanup_at_exit._exitcodeX
+            sty  cleanup_at_exit._exitcodeY
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            jmp  cleanup_at_exit
         }}
     }
 

compiler/res/prog8lib/buffers.p8

@@ -0,0 +1,152 @@
+; **experimental** buffer data structures, API subject to change!!
+
+%option no_symbol_prefixing, ignore_unused
+
+smallringbuffer {
+    ; -- A ringbuffer (FIFO queue) that occupies a single page in memory, containing 255 bytes maximum.
+    ;    You can store and retrieve words too.
+    ;    It's optimized for speed and depends on the byte-wrap-around feature when doing incs and decs.
+
+    ubyte fill
+    ubyte head
+    ubyte tail
+    ubyte[256] buffer
+
+    sub init() {
+        ; -- (re)initialize the ringbuffer, you must call this before using the other routines
+        head = fill = 0
+        tail = 255
+    }
+
+    sub put(ubyte value) -> bool {
+        ; -- store a byte in the buffer, returns success
+        if fill==255
+            return false
+        buffer[head] = value
+        head++
+        fill++
+        return true
+    }
+
+    sub putw(uword value) -> bool {
+        ; -- store a word in the buffer, returns success
+        if fill>=254
+            return false
+        fill += 2
+        buffer[head] = lsb(value)
+        head++
+        buffer[head] = msb(value)
+        head++
+        return true
+    }
+
+    sub get() -> ubyte {
+        ; -- retrieves a byte from the buffer. Also sets Carry flag: set=success, clear=buffer was empty
+        if fill==0 {
+            sys.clear_carry()
+            return 0
+        }
+        fill--
+        tail++
+        sys.set_carry()
+        return buffer[tail]
+    }
+
+    sub getw() -> uword {
+        ; -- retrieves a word from the buffer. Also sets Carry flag: set=success, clear=buffer was empty
+        if fill<2 {
+            sys.clear_carry()
+            return 0
+        }
+        fill -= 2
+        tail++
+        cx16.r0L = buffer[tail]
+        tail++
+        cx16.r0H = buffer[tail]
+        sys.set_carry()
+        return cx16.r0
+    }
+}
+
+
+ringbuffer {
+    ; -- A ringbuffer (FIFO queue) that occupies a single page in memory, containing 8 KB maximum.
+    ;    You can store and retrieve words too.
+
+    uword fill
+    uword head
+    uword tail
+    uword buffer_ptr = memory("ringbuffer", 8192, 0)
+
+    sub init() {
+        ; -- (re)initialize the ringbuffer, you must call this before using the other routines
+        head = fill = 0
+        tail = 8191
+    }
+
+    sub put(ubyte value) -> bool {
+        ; -- store a byte in the buffer, returns success
+        if fill==8192
+            return false
+        buffer_ptr[head] = value
+        inc_head()
+        fill++
+        return true
+    }
+
+    sub putw(uword value) -> bool {
+        ; -- store a word in the buffer, returns success
+        if fill>=8191
+            return false
+        fill += 2
+        buffer_ptr[head] = lsb(value)
+        inc_head()
+        buffer_ptr[head] = msb(value)
+        inc_head()
+        return true
+    }
+
+    sub get() -> ubyte {
+        ; -- retrieves a byte from the buffer. Also sets Carry flag: set=success, clear=buffer was empty
+        if fill==0 {
+            sys.clear_carry()
+            return 0
+        }
+        fill--
+        inc_tail()
+        cx16.r0L = buffer_ptr[tail]
+        sys.set_carry()
+        return cx16.r0L
+    }
+
+    sub getw() -> uword {
+        ; -- retrieves a word from the buffer. Also sets Carry flag: set=success, clear=buffer was empty
+        if fill<2 {
+            sys.clear_carry()
+            return 0
+        }
+        fill -= 2
+        inc_tail()
+        cx16.r0L = buffer_ptr[tail]
+        inc_tail()
+        cx16.r0H = buffer_ptr[tail]
+        sys.set_carry()
+        return cx16.r0
+    }
+
+    sub inc_head() {
+        head++
+        if msb(head)==$20
+            head=0
+    }
+
+    sub inc_tail() {
+        tail++
+        if msb(tail)==$20
+            tail=0
+    }
+}
+
+
+; TODO ringbuffer (FIFO queue) should use banked ram on the X16, but still work on virtual
+; TODO stack (LIFO queue) using more than 1 page of ram (maybe even banked ram on the x16)

compiler/res/prog8lib/c128/syslib.p8

@@ -351,6 +351,14 @@ sys {
 
     const ubyte target = 128         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16.
 
+    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.
@@ -360,7 +368,6 @@ asmsub  init_system()  {
     ; Uppercase charset is activated.
     %asm {{
         sei
-        cld
         lda  #0
         sta  $ff00      ; select default bank 15
         jsr  cbm.IOINIT
@@ -373,8 +380,6 @@ asmsub  init_system()  {
         lda  #0
         sta  c64.BGCOL0
         jsr  disable_runstop_and_charsetswitch
-        clc
-        clv
         cli
         rts
     }}
@@ -382,7 +387,10 @@ asmsub  init_system()  {
 
 asmsub  init_system_phase2()  {
     %asm {{
-        rts     ; no phase 2 steps on the C128
+        cld
+        clc
+        clv
+        rts
     }}
 }
 

compiler/res/prog8lib/c64/floats.p8

@@ -36,7 +36,7 @@ romsub $b1bf = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 100-101
 ; 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)
+; there is also floats.FREAD{S,U}24AXY  that read (un)signed int24 into fac1 from A/X/Y (lo/mid/hi bytes)
 romsub $b391 = GIVAYF(ubyte lo @ Y, ubyte hi @ A) clobbers(A,X,Y)
 
 romsub $b3a2 = FREADUY(ubyte value @ Y) clobbers(A,X,Y)     ; 8 bit unsigned Y -> float in fac1
@@ -104,7 +104,6 @@ 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
@@ -119,6 +118,39 @@ 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 {{
@@ -177,6 +209,13 @@ 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/floats_funcs.asm

@@ -6,108 +6,6 @@ func_sign_f_into_A	.proc
 		jmp  SIGN
 		.pend
 
-func_swap_f	.proc
-		; -- swap floats pointed to by SCRATCH_ZPWORD1, SCRATCH_ZPWORD2
-		ldy  #4
--               lda  (P8ZP_SCRATCH_W1),y
-		pha
-		lda  (P8ZP_SCRATCH_W2),y
-		sta  (P8ZP_SCRATCH_W1),y
-		pla
-		sta  (P8ZP_SCRATCH_W2),y
-		dey
-		bpl  -
-		rts
-		.pend
-
-func_reverse_f	.proc
-		; --- reverse an array of floats (array in P8ZP_SCRATCH_W1, num elements in A)
-_left_index = P8ZP_SCRATCH_W2
-_right_index = P8ZP_SCRATCH_W2+1
-_loop_count = P8ZP_SCRATCH_REG
-		pha
-		jsr  a_times_5
-		sec
-		sbc  #5
-		sta  _right_index
-		lda  #0
-		sta  _left_index
-		pla
-		lsr  a
-		sta  _loop_count
-_loop		; push the left indexed float on the stack
-		ldy  _left_index
-		lda  (P8ZP_SCRATCH_W1),y
-		pha
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		pha
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		pha
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		pha
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		pha
-		; copy right index float to left index float
-		ldy  _right_index
-		lda  (P8ZP_SCRATCH_W1),y
-		ldy  _left_index
-		sta  (P8ZP_SCRATCH_W1),y
-		inc  _left_index
-		inc  _right_index
-		ldy  _right_index
-		lda  (P8ZP_SCRATCH_W1),y
-		ldy  _left_index
-		sta  (P8ZP_SCRATCH_W1),y
-		inc  _left_index
-		inc  _right_index
-		ldy  _right_index
-		lda  (P8ZP_SCRATCH_W1),y
-		ldy  _left_index
-		sta  (P8ZP_SCRATCH_W1),y
-		inc  _left_index
-		inc  _right_index
-		ldy  _right_index
-		lda  (P8ZP_SCRATCH_W1),y
-		ldy  _left_index
-		sta  (P8ZP_SCRATCH_W1),y
-		inc  _left_index
-		inc  _right_index
-		ldy  _right_index
-		lda  (P8ZP_SCRATCH_W1),y
-		ldy  _left_index
-		sta  (P8ZP_SCRATCH_W1),y
-		; pop the float off the stack into the right index float
-		ldy  _right_index
-		pla
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		pla
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		pla
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		pla
-		sta  (P8ZP_SCRATCH_W1),y
-		dey
-		pla
-		sta  (P8ZP_SCRATCH_W1),y
-		inc  _left_index
-		lda  _right_index
-		sec
-		sbc  #9
-		sta  _right_index
-		dec  _loop_count
-		bne  _loop
-		rts
-
-		.pend
-
-
 
 a_times_5	.proc
 		sta  P8ZP_SCRATCH_B1
@@ -118,26 +16,6 @@ a_times_5	.proc
 		rts
 		.pend
 
-func_any_f_into_A	.proc
-		jsr  a_times_5
-		jmp  prog8_lib.func_any_b_into_A
-		.pend
-
-func_all_f_into_A	.proc
-		jsr  a_times_5
-		jmp  prog8_lib.func_all_b_into_A
-		.pend
-
-func_any_f_stack	.proc
-		jsr  a_times_5
-		jmp  prog8_lib.func_any_b_stack
-		.pend
-
-func_all_f_stack	.proc
-		jsr  a_times_5
-		jmp  prog8_lib.func_all_b_stack
-		.pend
-
 func_abs_f_into_FAC1    .proc
         jsr  MOVFM
         jmp  ABS

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)
     }
 
 

compiler/res/prog8lib/c64/syslib.p8

@@ -347,6 +347,13 @@ sys {
 
     const ubyte target = 64         ;  compilation target specifier.  64 = C64, 128 = C128, 16 = CommanderX16.
 
+    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.
@@ -357,7 +364,6 @@ asmsub  init_system()  {
     ; Uppercase charset is activated.
     %asm {{
         sei
-        cld
         lda  #%00101111
         sta  $00
         lda  #%00100111
@@ -372,8 +378,6 @@ asmsub  init_system()  {
         lda  #0
         sta  c64.BGCOL0
         jsr  disable_runstop_and_charsetswitch
-        clc
-        clv
         cli
         rts
     }}
@@ -381,7 +385,10 @@ asmsub  init_system()  {
 
 asmsub  init_system_phase2()  {
     %asm {{
-        rts     ; no phase 2 steps on the C64
+        cld
+        clc
+        clv
+        rts
     }}
 }
 

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

@@ -517,16 +517,16 @@ _stop
 
 asmsub internal_ubyte2decimal(ubyte value @A) -> ubyte @Y, ubyte @X, ubyte @A {
     %asm {{
-        ldy #'0'-1
-        ldx #'9'+1
+        ldy  #'0'-1
+        ldx  #'9'+1
         sec
 -       iny
-        sbc #100
-        bcs -
+        sbc  #100
+        bcs  -
 -       dex
-        adc #10
-        bmi -
-        adc #'0'-1
+        adc  #10
+        bmi  -
+        adc  #'0'-1
         rts
     }}
 }

compiler/res/prog8lib/cx16/audio.p8

@@ -0,0 +1,459 @@
+%import syslib
+
+; Stubs for the audio routines in rom bank 10
+; They use JSRFAR to select the correct bank, so user code doesn't have to bother.
+;
+; The stubs here still use the romsub definitions of the routines from syslib, but those are
+; just the jump addresses and are unaware of the kernal bank. Wrapping in JSRFAR fixes this.
+
+audio {
+    const ubyte rom_bank = $0a
+    asmsub audio_init() clobbers(A,X,Y) -> bool @Pc { ; (re)initialize both vera PSG and YM audio chips
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.audio_init
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_fmfreq(ubyte channel @A, uword freq @XY, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_fmfreq
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_fmnote(ubyte channel @A, ubyte note @X, ubyte fracsemitone @Y, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_fmnote
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_fmplaystring(ubyte length @A, str string @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_fmplaystring
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_fmvib(ubyte speed @A, ubyte depth @X) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_fmvib
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_playstringvoice(ubyte channel @A) clobbers(Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_playstringvoice
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_psgfreq(ubyte voice @A, uword freq @XY) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_psgfreq
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_psgnote(ubyte voice @A, ubyte note @X, ubyte fracsemitone @Y) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_psgnote
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_psgwav(ubyte voice @A, ubyte waveform @X) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_psgwav
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_psgplaystring(ubyte length @A, str string @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_psgplaystring
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_fmchordstring(ubyte length @A, str string @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_fmchordstring
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub bas_psgchordstring(ubyte length @A, str string @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.bas_psgchordstring
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_bas2fm(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_bas2fm
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_bas2midi(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_bas2midi
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_bas2psg(ubyte note @X, ubyte fracsemitone @Y) clobbers(A) -> uword @XY, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_bas2psg
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_fm2bas(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_fm2bas
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_fm2midi(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_fm2midi
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_fm2psg(ubyte note @X, ubyte fracsemitone @Y) clobbers(A) -> uword @XY, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_fm2psg
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_freq2bas(uword freqHz @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_freq2bas
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_freq2fm(uword freqHz @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_freq2fm
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_freq2midi(uword freqHz @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_freq2midi
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_freq2psg(uword freqHz @XY) clobbers(A) -> uword @XY, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_freq2psg
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_midi2bas(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_midi2bas
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_midi2fm(ubyte note @X) clobbers(A) -> ubyte @X, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_midi2fm
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_midi2psg(ubyte note @X, ubyte fracsemitone @Y) clobbers(A) -> uword @XY, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_midi2psg
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_psg2bas(uword freq @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_psg2bas
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_psg2fm(uword freq @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_psg2fm
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub notecon_psg2midi(uword freq @XY) clobbers(A) -> ubyte @X, ubyte @Y, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.notecon_psg2midi
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_init() clobbers(A,X,Y) {              ; (re)init Vera PSG
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_init
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_playfreq(ubyte voice @A, uword freq @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_playfreq
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_read(ubyte offset @X, bool cookedVol @Pc) clobbers(Y) -> ubyte @A{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_read
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_setatten(ubyte voice @A, ubyte attenuation @X) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_setatten
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_setfreq(ubyte voice @A, uword freq @XY) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_setfreq
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_setpan(ubyte voice @A, ubyte panning @X) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_setpan
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_setvol(ubyte voice @A, ubyte volume @X) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_setvol
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_write(ubyte value @A, ubyte offset @X) clobbers(Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_write
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_write_fast(ubyte value @A, ubyte offset @X) clobbers(Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_write_fast
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_getatten(ubyte voice @A) clobbers(Y) -> ubyte @X{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_getatten
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub psg_getpan(ubyte voice @A) clobbers(Y) -> ubyte @X{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.psg_getpan
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_init() clobbers(A,X,Y) -> bool @Pc {            ; (re)init YM chip
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_init
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_loaddefpatches() clobbers(A,X,Y) -> bool @Pc {   ; load default YM patches
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_loaddefpatches
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_loadpatch(ubyte channel @A, uword patchOrAddress @XY, bool what @Pc) clobbers(A,X,Y){
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_loadpatch
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_loadpatchlfn(ubyte channel @A, ubyte lfn @X) clobbers(X,Y) -> ubyte @A, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_loadpatchlfn
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_playdrum(ubyte channel @A, ubyte note @X) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_playdrum
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_playnote(ubyte channel @A, ubyte kc @X, ubyte kf @Y, bool notrigger @Pc) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_playnote
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_setatten(ubyte channel @A, ubyte attenuation @X) clobbers(Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_setatten
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_setdrum(ubyte channel @A, ubyte note @X) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_setdrum
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_setnote(ubyte channel @A, ubyte kc @X, ubyte kf @Y) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_setnote
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_setpan(ubyte channel @A, ubyte panning @X) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_setpan
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_read(ubyte register @X, bool cooked @Pc) clobbers(Y) -> ubyte @A, bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_read
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_release(ubyte channel @A) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_release
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_trigger(ubyte channel @A, bool noRelease @Pc) clobbers(A,X,Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_trigger
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_write(ubyte value @A, ubyte register @X) clobbers(Y) -> bool @Pc{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_write
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_getatten(ubyte channel @A) clobbers(Y) -> ubyte @X{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_getatten
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_getpan(ubyte channel @A) clobbers(Y) -> ubyte @X{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_getpan
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+    asmsub ym_get_chip_type() clobbers(X) -> ubyte @A{
+        %asm{{
+            jsr cx16.JSRFAR
+            .word cx16.ym_get_chip_type
+            .byte p8c_rom_bank
+            rts
+        }}
+    }
+}

compiler/res/prog8lib/cx16/diskio.p8

@@ -379,6 +379,7 @@ m_in_buffer     sta  $ffff
     ; optimized for Commander X16 to use MACPTR block read kernal call
     sub f_read_all(uword bufferpointer) -> uword {
         ; -- read the full contents of the file, returns number of bytes read.
+        ;    It is assumed the file size is less than 64 K.
         ;    NOTE: cannot be used to load into VRAM.  Use vload() or call cx16.MACPTR() yourself with the vera data register as address.
         if not iteration_in_progress
             return 0
@@ -393,12 +394,12 @@ m_in_buffer     sta  $ffff
         return total_read
     }
 
-    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y {
+    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y, ubyte @A {
         ; Routine to read text lines from a text file. Lines must be less than 255 characters.
         ; Reads characters from the input file UNTIL a newline or return character (or EOF).
         ; The line read will be 0-terminated in the buffer (and not contain the end of line character).
         ; The length of the line is returned in Y. Note that an empty line is okay and is length 0!
-        ; I/O error status should be checked by the caller itself via READST() routine.
+        ; The I/O error status byte is returned in A.
         %asm {{
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
@@ -415,7 +416,8 @@ _loop       jsr  cbm.CHRIN
 _line_end   dey     ; get rid of the trailing end-of-line char
             lda  #0
             sta  (P8ZP_SCRATCH_W1),y
-_end        rts
+_end        jsr  cbm.READST
+            rts
         }}
     }
 
@@ -467,6 +469,7 @@ _end        rts
 
     sub f_write(uword bufferpointer, uword num_bytes) -> bool {
         ; -- write the given number of bytes to the currently open file
+        ;    you can call this multiple times to append more data
         if num_bytes!=0 {
             reset_write_channel()
             do {
@@ -754,6 +757,11 @@ 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'
@@ -863,6 +871,7 @@ io_error:
 
     sub f_seek(uword pos_hiword, uword pos_loword) {
         ; -- seek in the reading file opened with f_open, to the given 32-bits position
+        ;    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.
         ubyte[6] command = ['p',0,0,0,0,0]
         command[1] = READ_IO_CHANNEL       ; f_open uses this secondary address
         command[2] = lsb(pos_loword)
@@ -931,15 +940,14 @@ 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

@@ -90,6 +90,38 @@ 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
 	%asm {{
@@ -157,7 +189,6 @@ _msg    .text 13,"?rom 47+ required for val1",13,0
     }}
 }
 
-
 &uword AYINT_facmo = $c6      ; $c6/$c7 contain result of AYINT   (See "basic.sym" kernal symbol file)
 
 sub rnd() -> float {
@@ -173,6 +204,14 @@ 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/gfx2.p8

@@ -13,6 +13,10 @@
 ; 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 $0:0000
+;
+; NOTE: In the future, these routines might be split out to separate modules, 1 for each screen mode,
+;       so they can be optimized a lot better.  There's already a "gfx_lores" module with a few routines for lores 256C mode.
 ;
 ; SCREEN MODE LIST:
 ;   mode 0 = reset back to default text mode
@@ -254,6 +258,7 @@ gfx2 {
                 ;; 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]
@@ -272,10 +277,10 @@ gfx2 {
 
         sub set_both_strides(ubyte stride) {
             stride <<= 4
-            cx16.VERA_CTRL = 0
-            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
             cx16.VERA_CTRL = 1
             cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
+            cx16.VERA_CTRL = 0
+            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
         }
 
     }
@@ -283,6 +288,9 @@ gfx2 {
     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
@@ -477,12 +485,25 @@ 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--
@@ -491,6 +512,11 @@ 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) {
@@ -500,12 +526,23 @@ 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--
@@ -514,6 +551,10 @@ 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) {
@@ -547,6 +588,7 @@ gfx2 {
                 ; 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]
@@ -626,10 +668,18 @@ gfx2 {
     }
 
     sub fill(uword x, uword 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 = 64
+        const ubyte MAXDEPTH = 100
         word @zp xx = x as word
         word @zp yy = y as word
         word[MAXDEPTH] @split @shared stack_xl
@@ -641,6 +691,7 @@ gfx2 {
         word x2
         byte dy
         cx16.r10L = new_color
+
         sub push_stack(word sxl, word sxr, word sy, byte sdy) {
             if cx16.r12L==MAXDEPTH
                 return
@@ -704,39 +755,27 @@ gfx2 {
             return
         if xx<0 or xx>width-1 or yy<0 or yy>height-1
             return
+        if gfx2.active_mode == 2 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
-            ; possible speed optimization: if mode==1 (256c) use vera autodecrement instead of pget(), but code bloat not worth it?
-            while xx >= 0 {
-                if pget(xx as uword, yy as uword) != cx16.r11L
-                    break
-                xx--
+            when active_mode {
+                1 -> if fill_scanline_left_8bpp() goto skip
+                2 -> if fill_scanline_left_2bpp() goto skip
             }
-            if x1!=xx
-                horizontal_line(xx as uword+1, yy as uword, x1-xx as uword, cx16.r10L)
-            else
-                goto skip
-
             left = xx + 1
             if left < x1
                 push_stack(left, x1 - 1, yy, -dy)
             xx = x1 + 1
 
             do {
-                cx16.r9s = xx
-                ; possible speed optimization: if mode==1 (256c) use vera autoincrement instead of pget(), but code bloat not worth it?
-                while xx <= width-1 {
-                    if pget(xx as uword, yy as uword) != cx16.r11L
-                        break
-                    xx++
+                when active_mode {
+                    1 -> fill_scanline_right_8bpp()
+                    2 -> fill_scanline_right_2bpp()
                 }
-                if cx16.r9s!=xx
-                    horizontal_line(cx16.r9, yy as uword, xx-cx16.r9s as uword, cx16.r10L)
-
                 push_stack(left, xx - 1, yy, dy)
                 if xx > x2 + 1
                     push_stack(x2 + 1, xx - 1, yy, -dy)
@@ -750,6 +789,133 @@ skip:
                 left = xx
             } until xx>x2
         }
+
+        sub set_vera_address() {
+            ; set both data0 and data1 addresses (expects H in R1L, M/L in R0)
+            cx16.VERA_CTRL = 0
+            cx16.VERA_ADDR_H = cx16.r1L
+            cx16.VERA_ADDR = cx16.r0
+            cx16.VERA_CTRL = 1
+            cx16.VERA_ADDR_H = cx16.r1L
+            cx16.VERA_ADDR = cx16.r0
+            cx16.VERA_CTRL = 0
+        }
+
+        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++
+            }
+        }
+
+        sub fill_scanline_left_2bpp() -> bool {
+            uword vx = xx as uword
+            void gfx2.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
+
+            repeat {
+                vub = cx16.VERA_DATA0  ; read the VERA color data for 4 pixels
+                if cmask == $03 {     ; only speed fill from far right
+                    ; speed fill
+                    if vub == seedm { ; all four colors match the seed
+                        nvub = fillm  ; replace all four colors at once
+                        xx -= 4
+                        goto set_byte  ; go on
+                    }
+                }
+
+                ; replace one color at a time
+                nvub = vub
+                while cmask != 0 {
+                    if vub & cmask == seedm & cmask {
+                        nvub &= ~cmask
+                        nvub |= cmask & fillm
+;                        %asm{{
+;                            lda  p8v_cmask
+;                            trb  p8v_nvub
+;                            and  p8v_fillm
+;                            tsb  p8v_nvub
+;                        }}
+                        xx--
+                        cmask <<= 2
+                    } else { ; not the seed color, finish here
+                        cx16.VERA_DATA1 = nvub
+                        return vx == xx
+                    }
+                }
+set_byte:
+                cx16.VERA_DATA1 = nvub
+                if xx <= 0 break
+                cmask = $03
+            }
+            return vx == xx
+        }
+
+        sub fill_scanline_right_2bpp() {
+            void gfx2.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
+
+            repeat {
+                vub = cx16.VERA_DATA0  ; read the VERA color data for 4 pixels
+                ; speed fill
+                if vub == seedm { ; all four colors match the seed
+                    nvub = fillm  ; replace all four colors at once
+                    xx += 4
+                    goto set_byte   ; go on
+                }
+                ; replace one color at a time
+                nvub = vub
+                while cmask != 0 {
+                    if vub & cmask == seedm & cmask {
+                        nvub &= ~cmask
+                        nvub |= cmask & fillm
+;                        %asm{{
+;                            lda  p8v_cmask
+;                            trb  p8v_nvub
+;                            and  p8v_fillm
+;                            tsb  p8v_nvub
+;                        }}
+                        xx++
+                        cmask >>= 2
+                    } else { ; not the seed color finish here
+                        cx16.VERA_DATA1 = nvub
+                        return
+                    }
+                }
+set_byte:
+                cx16.VERA_DATA1 = nvub
+                if xx >= gfx2.width-1 break
+                cmask = $C0
+            }
+        }
+
+        sub set_color_masks() {
+            seedm = cx16.r11L | (cx16.r11L<<2) | (cx16.r11L<<4) | (cx16.r11L<<6) ; seed mask
+            fillm = cx16.r10L | (cx16.r10L<<2) | (cx16.r10L<<4) | (cx16.r10L<<6) ; fill mask
+        }
     }
 
     sub position(uword @zp xx, uword yy) {

compiler/res/prog8lib/cx16/gfx_lores.p8

@@ -0,0 +1,268 @@
+; 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 verafx
+
+
+gfx_lores {
+
+    %option ignore_unused
+
+    sub set_screen_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)
+    }
+
+    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 240 {
+            %asm {{
+                lda  p8v_color
+                ldy  #320/8
+-               .rept 8
+                sta  cx16.VERA_DATA0
+                .endrept
+                dey
+                bne  -
+            }}
+        }
+        cx16.VERA_ADDR=0
+        cx16.VERA_ADDR_H = 0
+    }
+
+    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  cx16.r0L
+            sta  cx16.VERA_DATA0
+            rts
+        }}
+    }
+
+    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, and ~3-4 times as fast as gfx2.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_color
+                sta  cx16.VERA_DATA0
+                rts
+            }}
+        }
+    }
+
+    sub horizontal_line(uword xx, ubyte yy, uword length, ubyte color) {
+        if length==0
+            return
+        plot(xx, yy, color)  ; set starting position by reusing plot routine
+        ; set vera auto-increment to 1 pixel
+        cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (1<<4)
+
+        %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 vertical_line(uword xx, ubyte yy, ubyte lheight, ubyte color) {
+        if lheight==0
+            return
+        plot(xx, yy, color)  ; set starting position by reusing plot routine
+        ; set vera auto-increment to 320 pixel increment (=next line)
+        cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
+        %asm {{
+            ldy  p8v_lheight
+            lda  p8v_color
+-           sta  cx16.VERA_DATA0
+            dey
+            bne  -
+        }}
+    }
+
+
+    %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

@@ -1,5 +1,4 @@
 %import syslib
-%import textio
 
 ; Bitmap pixel graphics module for the CommanderX16
 ; Wraps the graphics functions that are in ROM.
@@ -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) {

compiler/res/prog8lib/cx16/monogfx.p8

@@ -32,7 +32,7 @@ monogfx {
         width = 320
         height = 240
         mode = MODE_NORMAL
-        clear_screen(0)
+        clear_screen(false)
     }
 
     sub hires() {
@@ -47,7 +47,7 @@ monogfx {
         width = 640
         height = 480
         mode = MODE_NORMAL
-        clear_screen(0)
+        clear_screen(false)
     }
 
     sub textmode() {
@@ -61,16 +61,16 @@ monogfx {
         mode = dm
     }
 
-    sub clear_screen(ubyte color) {
+    sub clear_screen(bool draw) {
         position(0, 0)
         when width {
             320 -> {
                 repeat 240/2/8
-                    cs_innerloop640(color)
+                    cs_innerloop640(draw)
             }
             640 -> {
                 repeat 480/8
-                    cs_innerloop640(color)
+                    cs_innerloop640(draw)
             }
         }
         position(0, 0)
@@ -90,8 +90,8 @@ monogfx {
     }
 
     sub fillrect(uword xx, uword yy, uword rwidth, uword rheight, bool draw) {
-        ; Draw a filled rectangle of the given size and color.
-        ; To fill the whole screen, use clear_screen(color) instead - it is much faster.
+        ; Draw a filled rectangle of the given size.
+        ; To fill the whole screen, use clear_screen(draw) instead - it is much faster.
         if rwidth==0
             return
         repeat rheight {
@@ -358,10 +358,10 @@ drawmode:               ora  cx16.r15L
 
         sub set_both_strides(ubyte stride) {
             stride <<= 4
-            cx16.VERA_CTRL = 0
-            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
             cx16.VERA_CTRL = 1
             cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
+            cx16.VERA_CTRL = 0
+            cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | stride
         }
 
     }
@@ -563,12 +563,23 @@ drawmode:               ora  cx16.r15L
             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, draw)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
-            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
+                last_y4 = new_y4
+            }
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -577,6 +588,10 @@ drawmode:               ora  cx16.r15L
             decisionOver2 += yy*$0002
             decisionOver2++
         }
+        ; draw the final two spans
+        yy--
+        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
+        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
@@ -586,12 +601,23 @@ drawmode:               ora  cx16.r15L
             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, draw)
             safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
-            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
+            new_y3 = ycenter+radius
+            if new_y3 != last_y3 {
+                safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
+                last_y3 = new_y3
+            }
+            new_y4 = ycenter-radius
+            if new_y4 != last_y4 {
+                safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
+                last_y4 = new_y4
+            }
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -600,6 +626,10 @@ drawmode:               ora  cx16.r15L
             decisionOver2 += yy*$0002
             decisionOver2++
         }
+        ; draw the final two spans
+        yy--
+        safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
+        safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub plot(uword @zp xx, uword @zp yy, bool @zp draw) {
@@ -708,7 +738,7 @@ invert:
         ; Non-recursive scanline flood fill.
         ; based loosely on code found here https://www.codeproject.com/Articles/6017/QuickFill-An-efficient-flood-fill-algorithm
         ; with the fixes applied to the seedfill_4 routine as mentioned in the comments.
-        const ubyte MAXDEPTH = 64
+        const ubyte MAXDEPTH = 100
         word @zp xx = x as word
         word @zp yy = y as word
         word[MAXDEPTH] @split @shared stack_xl
@@ -789,31 +819,14 @@ invert:
         while cx16.r12L!=0 {
             pop_stack()
             xx = x1
-            while xx >= 0 {
-                if pget(xx as uword, yy as uword) as ubyte != cx16.r11L
-                    break
-                xx--
-            }
-            if x1!=xx
-                horizontal_line(xx as uword+1, yy as uword, x1-xx as uword, cx16.r10L as bool)
-            else
-                goto skip
-
+            if fill_scanline_left() goto skip
             left = xx + 1
             if left < x1
                 push_stack(left, x1 - 1, yy, -dy)
             xx = x1 + 1
 
             do {
-                cx16.r9s = xx
-                while xx <= width-1 {
-                    if pget(xx as uword, yy as uword) as ubyte != cx16.r11L
-                        break
-                    xx++
-                }
-                if cx16.r9s!=xx
-                    horizontal_line(cx16.r9, yy as uword, xx-cx16.r9s as uword, cx16.r10L as bool)
-
+                fill_scanline_right()
                 push_stack(left, xx - 1, yy, dy)
                 if xx > x2 + 1
                     push_stack(x2 + 1, xx - 1, yy, -dy)
@@ -827,6 +840,70 @@ skip:
                 left = xx
             } until xx>x2
         }
+
+        sub fill_scanline_left() -> bool {
+            ; TODO maybe this could use vera auto decrement, but that would require some clever masking calculations
+            cx16.r9s = xx
+            while xx >= 0 {
+                if pgetset()
+                    break
+                xx--
+            }
+            return xx==cx16.r9s
+        }
+
+        sub fill_scanline_right() {
+             ; TODO maybe this could use vera auto increment, but that would require some clever masking calculations
+             cx16.r9s = xx
+            while xx <= width-1 {
+                if pgetset()
+                    break
+                xx++
+            }
+        }
+
+        sub pgetset() -> bool {
+            ; test and optionally set a pixel
+            word @zp xpos = xx
+            %asm {{
+                lda  p8v_xpos
+                and  #7
+                pha     ; xbits
+            }}
+            xpos /= 8
+            if width==320
+                xpos += yy*(320/8) as uword
+            else
+                xpos += yy*(640/8) as uword
+
+            %asm {{
+                stz  cx16.VERA_CTRL
+                stz  cx16.VERA_ADDR_H
+                lda  p8v_xpos+1
+                sta  cx16.VERA_ADDR_M
+                lda  p8v_xpos
+                sta  cx16.VERA_ADDR_L
+                ply         ; xbits
+                lda  p8s_plot.p8v_maskbits,y
+                and  cx16.VERA_DATA0
+                beq  +
+                lda  #1
++
+                ; cx16.r11L = seed color to check against
+                eor  cx16.r11L
+                beq  +
+                rts
++               ; cx16.r10L = new color to set
+                lda  p8s_plot.p8v_maskbits,y
+                ldx  cx16.r10L
+                beq  +
+                tsb  cx16.VERA_DATA0
+                bra  ++
++               trb  cx16.VERA_DATA0
++               lda  #0
+                rts
+            }}
+        }
     }
 
     sub position(uword @zp xx, uword yy) {
@@ -957,20 +1034,20 @@ cdraw_mod2              ora  cx16.VERA_DATA1
         sub set_autoincrs() {
             ; set autoincrements to go to next pixel row (40 or 80 increment)
             if width==320 {
-                cx16.VERA_CTRL = 0
-                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
                 cx16.VERA_CTRL = 1
                 cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (11<<4)
             } else {
-                cx16.VERA_CTRL = 0
-                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
                 cx16.VERA_CTRL = 1
                 cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
+                cx16.VERA_CTRL = 0
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & $0f | (12<<4)
             }
         }
     }
 
-    asmsub cs_innerloop640(ubyte color @A) clobbers(Y) {
+    asmsub cs_innerloop640(bool draw @A) clobbers(Y) {
         ; using verafx 32 bits writes here would make this faster but it's safer to
         ; use verafx only explicitly when you know what you're doing.
         %asm {{

compiler/res/prog8lib/cx16/palette.p8

@@ -106,6 +106,90 @@ palette {
         }
     }
 
+    sub fade_step_multi(ubyte startindex, ubyte endindex, uword target_rgb) -> bool {
+        ; Perform one color fade step for multiple consecutive palette entries.
+        ;   startindex = palette index of first color to fade
+        ;   endindex = palette index of last color to fade
+        ;   target_rgb = $RGB color value to fade towards
+        ; Returns true if one or more colors were changed, false if no fade steps were done anymore.
+        ; So you usually keep calling this until it returns false.
+        bool changed = false
+        while startindex <= endindex {
+            if fade_step(startindex, target_rgb)
+                changed=true
+            startindex++
+            if_z
+                break
+        }
+        return changed
+    }
+
+    sub fade_step_colors(ubyte startindex, ubyte endindex, uword target_colors) -> bool {
+        ; Perform one color fade step for multiple consecutive palette entries, to different target colors.
+        ;   startindex = palette index of first color to fade
+        ;   endindex = palette index of last color to fade
+        ;   target_colors = address of uword $RGB array of colors to fade towards
+        ; Returns true if one or more colors were changed, false if no fade steps were done anymore.
+        ; So you usually keep calling this until it returns false.
+        bool changed = false
+        ubyte target_index = 0
+        while startindex <= endindex {
+            if fade_step(startindex, peekw(target_colors+target_index))
+                changed=true
+            target_index += 2
+            startindex++
+            if_z
+                break
+        }
+        return changed
+    }
+
+    sub fade_step(ubyte index, uword target_rgb) -> bool {
+        ; Perform one color fade step for a single palette entry.
+        ;   index = palette index of the color to fade
+        ;   target_rgb = $RGB color value to fade towards
+        ; Returns true if the color was changed, false if no fade step was done anymore.
+        ; So you usually keep calling this until it returns false.
+        uword color = palette.get_color(index)
+        cx16.r0L = msb(color)            ; r
+        cx16.r1L = lsb(color) >> 4       ; g
+        cx16.r2L = lsb(color) & 15       ; b
+        cx16.r0H = msb(target_rgb) & 15  ; r2
+        cx16.r1H = lsb(target_rgb) >> 4  ; g2
+        cx16.r2H = lsb(target_rgb) & 15  ; b2
+
+        ubyte changed
+
+        ; use cmp() + status bits branches, to avoid multiple compares that could be done just once
+        cmp(cx16.r0L, cx16.r0H)
+        if_ne {
+            if_cc
+                cx16.r0L++
+            else
+                cx16.r0L--
+            changed++
+        }
+        cmp(cx16.r1L, cx16.r1H)
+        if_ne {
+            if_cc
+                cx16.r1L++
+            else
+                cx16.r1L--
+            changed++
+        }
+        cmp(cx16.r2L, cx16.r2H)
+        if_ne {
+            if_cc
+                cx16.r2L++
+            else
+                cx16.r2L--
+            changed++
+        }
+
+        palette.set_color(index, mkword(cx16.r0L, cx16.r1L<<4 | cx16.r2L))
+        return changed!=0
+    }
+
     sub set_c64pepto() {
         ; set first 16 colors to the "Pepto" PAL commodore-64 palette  http://www.pepto.de/projects/colorvic/
         uword[] colors = [

compiler/res/prog8lib/cx16/sprites.p8

@@ -23,7 +23,7 @@ sprites {
              ubyte databank, uword dataaddr,
              ubyte width_flag, ubyte height_flag,
              ubyte colors_flag, ubyte palette_offset) {
-        hide(spritenum)
+        pos(spritenum, -64, -64)                    ; move sprite off-screen initially
         cx16.VERA_DC_VIDEO |= %01000000             ; enable sprites globally
         dataaddr >>= 5
         dataaddr |= (databank as uword)<<11
@@ -42,9 +42,15 @@ sprites {
         cx16.vpoke_mask(1, sprite_reg+1, %11110000, msb(addr))    ; address 16:13
     }
 
-    sub get_data_ptr(ubyte spritenum) {
+    inline asmsub get_data_ptr(ubyte spritenum @A) -> ubyte @R1, uword @R0 {
         ; -- returns the VRAM address where the sprite's bitmap data is stored
         ;    R1 (byte) = the vera bank (0 or 1), R0 (word) = the address.
+        %asm {{
+            jsr  p8b_sprites.p8s_get_data_ptr_internal
+        }}
+    }
+
+    sub get_data_ptr_internal(ubyte spritenum) {
         sprite_reg = VERA_SPRITEREGS + spritenum*$0008
         cx16.r0L = cx16.vpeek(1, sprite_reg)
         cx16.r0H = cx16.vpeek(1, sprite_reg+1)
@@ -132,7 +138,15 @@ sprites {
     }
 
     sub hide(ubyte spritenum) {
-        pos(spritenum, -64, -64)
+        cx16.vpoke_and(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11110011)
+    }
+
+    sub show(ubyte spritenum) {
+        cx16.vpoke_or(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %00001100)
+    }
+
+    sub zdepth(ubyte spritenum, ubyte depth) {
+        cx16.vpoke_mask(1, VERA_SPRITEREGS + 6 + spritenum*$0008, %11110011, depth<<2)
     }
 
     sub flipx(ubyte spritenum, bool flipped) {
@@ -176,7 +190,7 @@ sprites {
     }
 
     sub set_mousepointer_image(uword data, bool compressed) {
-        get_data_ptr(0)  ; the mouse cursor is sprite 0
+        get_data_ptr_internal(0)  ; the mouse cursor is sprite 0
         if cx16.r1L==0 and cx16.r0==0
             return    ; mouse cursor not enabled
         ubyte vbank = cx16.r1L

compiler/res/prog8lib/cx16/syslib.p8

@@ -6,6 +6,39 @@
 cbm {
     ; Commodore (CBM) common variables, vectors and kernal routines
 
+; irq, system and hardware vectors  (common across cbm machines):
+    &uword  IERROR      = $0300
+    &uword  IMAIN       = $0302
+    &uword  ICRNCH      = $0304
+    &uword  IQPLOP      = $0306
+    &uword  IGONE       = $0308
+    &uword  IEVAL       = $030a
+    &ubyte  SAREG       = $030c     ; register storage for A for SYS calls
+    &ubyte  SXREG       = $030d     ; register storage for X for SYS calls
+    &ubyte  SYREG       = $030e     ; register storage for Y for SYS calls
+    &ubyte  SPREG       = $030f     ; register storage for P (status register) for SYS calls
+    &uword  USRADD      = $0311     ; vector for the USR() basic command
+    ; $0313 is unused.
+    &uword  CINV        = $0314     ; IRQ vector (in ram)
+    &uword  CBINV       = $0316     ; BRK vector (in ram)
+    &uword  NMINV       = $0318     ; NMI vector (in ram)
+    &uword  IOPEN       = $031a
+    &uword  ICLOSE      = $031c
+    &uword  ICHKIN      = $031e
+    &uword  ICKOUT      = $0320
+    &uword  ICLRCH      = $0322
+    &uword  IBASIN      = $0324
+    &uword  IBSOUT      = $0326
+    &uword  ISTOP       = $0328
+    &uword  IGETIN      = $032a
+    &uword  ICLALL      = $032c
+    ; $032e has a X16 specific function (KEYHDL) so you'll find this as cx16.KEYHDL
+    &uword  ILOAD       = $0330
+    &uword  ISAVE       = $0332
+    &uword  NMI_VEC     = $FFFA     ; 65c02 nmi vector, determined by the kernal if banked in
+    &uword  RESET_VEC   = $FFFC     ; 65c02 reset vector, determined by the kernal if banked in
+    &uword  IRQ_VEC     = $FFFE     ; 65c02 interrupt vector, determined by the kernal if banked in
+
 
 ; STROUT --> use txt.print
 ; CLEARSCR -> use txt.clear_screen
@@ -117,42 +150,14 @@ asmsub kbdbuf_clear() {
 
 cx16 {
 
-; irq, system and hardware vectors:
-    &uword  IERROR      = $0300
-    &uword  IMAIN       = $0302
-    &uword  ICRNCH      = $0304
-    &uword  IQPLOP      = $0306
-    &uword  IGONE       = $0308
-    &uword  IEVAL       = $030a
-    &ubyte  SAREG       = $030c     ; register storage for A for SYS calls
-    &ubyte  SXREG       = $030d     ; register storage for X for SYS calls
-    &ubyte  SYREG       = $030e     ; register storage for Y for SYS calls
-    &ubyte  SPREG       = $030f     ; register storage for P (status register) for SYS calls
-    &uword  USRADD      = $0311     ; vector for the USR() basic command
-    ; $0313 is unused.
-    &uword  CINV        = $0314     ; IRQ vector (in ram)
-    &uword  CBINV       = $0316     ; BRK vector (in ram)
-    &uword  NMINV       = $0318     ; NMI vector (in ram)
-    &uword  IOPEN       = $031a
-    &uword  ICLOSE      = $031c
-    &uword  ICHKIN      = $031e
-    &uword  ICKOUT      = $0320
-    &uword  ICLRCH      = $0322
-    &uword  IBASIN      = $0324
-    &uword  IBSOUT      = $0326
-    &uword  ISTOP       = $0328
-    &uword  IGETIN      = $032a
-    &uword  ICLALL      = $032c
+; cx16 specific vectors and variables
     &uword  KEYHDL      = $032e     ; keyboard scan code handler see examples/cx16/keyboardhandler.p8
-    &uword  ILOAD       = $0330
-    &uword  ISAVE       = $0332
-    &uword  NMI_VEC     = $FFFA     ; 65c02 nmi vector, determined by the kernal if banked in
-    &uword  RESET_VEC   = $FFFC     ; 65c02 reset vector, determined by the kernal if banked in
-    &uword  IRQ_VEC     = $FFFE     ; 65c02 interrupt vector, determined by the kernal if banked in
 
     &uword  edkeyvec    = $ac03     ; (ram bank 0): for intercepting BASIN/CHRIN key strokes. See set_chrin_keyhandler()
     &ubyte  edkeybk     = $ac05     ; ...the RAM bank of the handler routine, if not in low ram
 
+    &ubyte  stavec      = $03b2     ; argument for stash()
+
 
 ; the sixteen virtual 16-bit registers in both normal unsigned mode and signed mode (s)
     &uword r0  = $0002
@@ -395,8 +400,8 @@ romsub $ff5c = LKUPSA(ubyte sa @Y)  clobbers(A,X,Y)
 romsub $ff5f = screen_mode(ubyte mode @A, bool getCurrent @Pc) -> ubyte @A, ubyte @X, ubyte @Y, bool @Pc        ; also see SCREEN or get_screen_mode()
 romsub $ff62 = screen_set_charset(ubyte charset @A, uword charsetptr @XY)  clobbers(A,X,Y)
 romsub $ff6e = JSRFAR()  ; following word = address to call, byte after that=rom/ram bank it is in
-romsub $ff74 = fetch(ubyte bank @X, ubyte index @Y)  clobbers(X) -> ubyte @A
-romsub $ff77 = stash(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)
+romsub $ff74 = fetch(ubyte zp_startaddr @A, ubyte bank @X, ubyte index @Y)  clobbers(X) -> ubyte @A
+romsub $ff77 = stash(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)     ;  note: The the zero page address containing the base address is passed in stavec ($03B2)
 romsub $ff7d = PRIMM()
 
 ; high level graphics & fonts
@@ -478,7 +483,7 @@ romsub $C006 = x16edit_loadfile_options(ubyte firstbank @X, ubyte lastbank @Y, s
                 uword filenameLengthAndOptions @R1, uword tabstopAndWordwrap @R2,
                 uword disknumberAndColors @R3, uword headerAndStatusColors @R4) clobbers(A,X,Y)
 
-; Audio (rom bank 10 - you have to activate this bank manually first! Or use callfar/JSRFAR.)
+; Audio (rom bank 10 - you have to activate this bank manually first! Or use the stubs in the audio module. Or use callfar/JSRFAR manually.)
 romsub $C09F = audio_init() clobbers(A,X,Y) -> bool @Pc     ; (re)initialize both vera PSG and YM audio chips
 romsub $C000 = bas_fmfreq(ubyte channel @A, uword freq @XY, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc
 romsub $C003 = bas_fmnote(ubyte channel @A, ubyte note @X, ubyte fracsemitone @Y, bool noretrigger @Pc) clobbers(A,X,Y) -> bool @Pc
@@ -549,7 +554,8 @@ const ubyte  EXTAPI_ps2data_raw = $09
 const ubyte  EXTAPI_cursor_blink = $0A
 const ubyte  EXTAPI_led_update = $0B
 const ubyte  EXTAPI_mouse_set_position = $0C
-const ubyte  EXTAPI_scnsiz = $0D        ; rom R48+
+const ubyte  EXTAPI_scnsiz = $0D
+const ubyte  EXTAPI_kbd_leds = $0E
 
 ; extapi16 call numbers
 const ubyte  EXTAPI16_test = $00
@@ -598,6 +604,14 @@ asmsub mouse_pos() -> ubyte @A, uword @R0, uword @R1, byte @X {
     }}
 }
 
+sub mouse_present() -> bool {
+    ; -- check if a mouse is connected to the machine
+    cx16.r0L, void = cx16.i2c_read_byte($42, $22)         ; $22 = I2C_GET_MOUSE_DEVICE_ID
+    if_cs
+        return false
+    return cx16.r0L != $fc       ; $fc = BAT_FAIL
+}
+
 ; shims for the kernal routines called via the extapi call:
 
 asmsub mouse_set_pos(uword xpos @R0, uword ypos @R1) clobbers(X) {
@@ -1070,8 +1084,8 @@ asmsub  enable_irq_handlers(bool disable_all_irq_sources @Pc) clobbers(A,Y)  {
         trb  cx16.VERA_IEN      ; disable all IRQ sources
 +       lda  #<_irq_dispatcher
         ldy  #>_irq_dispatcher
-        sta  cx16.CINV
-        sty  cx16.CINV+1
+        sta  cbm.CINV
+        sty  cbm.CINV+1
         plp
         rts
 
@@ -1195,6 +1209,7 @@ asmsub set_sprcol_irq_handler(uword address @AY) clobbers(A) {
 asmsub set_aflow_irq_handler(uword address @AY) clobbers(A) {
     ; Sets the AFLOW irq handler to use with enable_irq_handlers().  Also enables AFLOW irqs.
     ; NOTE: unless a proper irq handler is already running, you should enclose this call in set_irqd() / clear_irqd() to avoid system crashes.
+    ; NOTE: the handler itself must fill the audio fifo buffer to at least 25% full again (1 KB) or the aflow irq will keep triggering!
     %asm {{
         php
         sei
@@ -1223,11 +1238,10 @@ sub search_x16edit() -> ubyte {
     str @shared signature = petscii:"x16edit"
     for cx16.r1L in 31 downto 0  {
         cx16.rombank(cx16.r1L)
-        cx16.r2 = $fff0
         %asm {{
             ldy  #0
 -           lda  signature,y
-            cmp  (cx16.r2),y
+            cmp  $fff0,y
             bne  +
             iny
             cpy #7
@@ -1322,6 +1336,23 @@ _continue   iny
         void cx16.i2c_write_byte($42, $05, cx16.r0L)
     }
 
+    asmsub rom_version() clobbers(Y) -> ubyte @A, bool @Pc {
+        ; Returns the KERNEL ROM version. Carry set if pre-release, clear if offical release.
+        %asm{{
+            ; the ROM BANK is unknown on entry
+            ldy  $01
+            stz  $01        ; KERNEL ROM
+            clc             ; prepare for released ROM
+            lda  $FF80
+            bpl  _final     ; pre-release versions are negative
+            eor  #$FF       ; twos complement
+            ina
+            sec
+    _final:
+            sty  $01
+            rts
+        }}
+    }
 }
 
 sys {
@@ -1329,6 +1360,13 @@ sys {
 
     const ubyte target = 16         ;  compilation target specifier.  64 = C64,  128 = C128,  16 = CommanderX16.
 
+    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.
@@ -1338,7 +1376,6 @@ asmsub  init_system()  {
         tax
         tay
         jsr  cx16.mouse_config  ; disable mouse
-        cld
         lda  cx16.VERA_DC_VIDEO
         and  #%00000111 ; retain chroma + output mode
         sta  P8ZP_SCRATCH_REG
@@ -1366,10 +1403,9 @@ asmsub  init_system()  {
         lda  #PROG8_VARSHIGH_RAMBANK
         sta  $00    ; select ram bank
         lda  #0
+        sta  $01    ; set ROM bank to kernal bank to speed up kernal calls
         tax
         tay
-        clc
-        clv
         cli
         rts
     }}
@@ -1378,13 +1414,17 @@ asmsub  init_system()  {
 asmsub  init_system_phase2()  {
     %asm {{
         sei
-        lda  cx16.CINV
+        lda  cbm.CINV
         sta  restore_irq._orig_irqvec
-        lda  cx16.CINV+1
+        lda  cbm.CINV+1
         sta  restore_irq._orig_irqvec+1
         lda  #PROG8_VARSHIGH_RAMBANK
         sta  $00    ; select ram bank
+        stz  $01    ; set ROM bank to kernal bank to speed up kernal calls
         cli
+        cld
+        clc
+        clv
         rts
     }}
 }
@@ -1418,9 +1458,9 @@ asmsub  set_irq(uword handler @AY) clobbers(A)  {
         sta  _modified+1
         sty  _modified+2
         lda  #<_irq_handler
-        sta  cx16.CINV
+        sta  cbm.CINV
         lda  #>_irq_handler
-        sta  cx16.CINV+1
+        sta  cbm.CINV+1
         lda  #1
         tsb  cx16.VERA_IEN      ; enable the vsync irq
         cli
@@ -1449,9 +1489,9 @@ asmsub  restore_irq() clobbers(A) {
 	%asm {{
 	    sei
 	    lda  _orig_irqvec
-	    sta  cx16.CINV
+	    sta  cbm.CINV
 	    lda  _orig_irqvec+1
-	    sta  cx16.CINV+1
+	    sta  cbm.CINV+1
 	    lda  cx16.VERA_IEN
 	    and  #%11110000     ; disable all Vera IRQs but the vsync
 	    ora  #%00000001
@@ -1478,9 +1518,9 @@ asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
             ldy  cx16.r0+1
             jsr  set_rasterline
             lda  #<_raster_irq_handler
-            sta  cx16.CINV
+            sta  cbm.CINV
             lda  #>_raster_irq_handler
-            sta  cx16.CINV+1
+            sta  cbm.CINV+1
             cli
             rts
 

compiler/res/prog8lib/cx16/textio.p8

@@ -262,6 +262,12 @@ sub iso16() {
     cx16.screen_set_charset(10, 0)  ; charset
 }
 
+sub kata() {
+    ; -- switch to katakana character set (requires rom 48+)
+    cbm.CHROUT($0f)                 ; iso mode
+    cx16.screen_set_charset(12, 0)  ; charset
+}
+
 asmsub  scroll_left() clobbers(A, X, Y)  {
 	; ---- scroll the whole screen 1 character to the left
 	;      contents of the rightmost column are unchanged, you should clear/refill this yourself

compiler/res/prog8lib/cx16/verafx.p8

@@ -1,6 +1,10 @@
-; Somewhat experimental Vera FX support.
+; Partial Vera FX support:
+; - fast 32 bit cached writes (clear, copy)
+; - transparent write setting
+; - hardware 16 bits multiplications
+;
 ; Docs:
-; https://github.com/X16Community/x16-docs/blob/101759f3bfa5e6cce4e8c5a0b67cb0f2f1c6341e/X16%20Reference%20-%2010%20-%20VERA%20FX%20Reference.md
+; https://github.com/X16Community/x16-docs/blob/fb63156cca2d6de98be0577aacbe4ddef458f896/X16%20Reference%20-%2010%20-%20VERA%20FX%20Reference.md
 ; https://docs.google.com/document/d/1q34uWOiM3Be2pnaHRVgSdHySI-qsiQWPTo_gfE54PTg
 
 verafx {
@@ -8,17 +12,18 @@ verafx {
 
     sub available() -> bool {
         ; returns true if Vera FX is available (Vera V0.3.1 or later), false if not.
-        cx16.r1L = 0
         cx16.r0L = cx16.VERA_CTRL
+        cx16.r0H = 0
         cx16.VERA_CTRL = $7e
         if cx16.VERA_DC_VER0 == $56 {
-            ; Vera version number is valid.
-            ; Vera fx is available on Vera version 0.3.1 and later,
-            ; so no need to even check VERA_DC_VER1, which contains 0 (or higher)
-            cx16.r1L = mkword(cx16.VERA_DC_VER2, cx16.VERA_DC_VER3) >= $0301 as ubyte
+            ; Vera version number is valid. Vera fx is available on Vera version 0.3.1 and later.
+            if cx16.VERA_DC_VER1>0
+                cx16.r0H = 1
+            else
+                cx16.r0H = mkword(cx16.VERA_DC_VER2, cx16.VERA_DC_VER3) >= $0301 as ubyte
         }
         cx16.VERA_CTRL = cx16.r0L
-        return cx16.r1L as bool
+        return cx16.r0H as bool
     }
 
     sub clear(ubyte vbank, uword vaddr, ubyte data, uword num_longwords) {
@@ -109,17 +114,28 @@ verafx {
         cx16.VERA_CTRL = 0
     }
 
-    ; unsigned multiplication just passes the values as signed to muls
-    ; if you do this yourself in your call to muls, it will save a few instructions.
-    inline asmsub mult(uword value1 @R0, uword value2 @R1) clobbers(X) -> uword @AY, uword @R0 {
-        ; Returns the 32 bits unsigned result in AY and R0  (lower word, upper word).
+
+    asmsub mult16(uword value1 @R0, uword value2 @R1) clobbers(X) -> uword @AY {
+        ; Returns the 16 bits unsigned result of R0*R1 in AY.
+        ; Note: only the lower 16 bits!   (the upper 16 bits are not valid for unsigned word multiplications, only for signed)
+        ; Verafx doesn't support unsigned values like this for full 32 bit result.
         %asm {{
+            lda  cx16.r0
+            sta  P8ZP_SCRATCH_W1
+            lda  cx16.r0+1
+            sta  P8ZP_SCRATCH_W1+1
             jsr  verafx.muls
+            ldx  P8ZP_SCRATCH_W1
+            stx  cx16.r0
+            ldx  P8ZP_SCRATCH_W1+1
+            stx  cx16.r0+1
+            rts
         }}
     }
 
     asmsub muls(word value1 @R0, word value2 @R1) clobbers(X) -> word @AY, word @R0 {
         ; Returns the 32 bits signed result in AY and R0  (lower word, upper word).
+        ; Vera Fx multiplication support only works on signed values!
         %asm {{
             lda  #(2 << 1)
             sta  cx16.VERA_CTRL        ; $9F25
@@ -160,12 +176,15 @@ verafx {
             ldx  cx16.VERA_DATA0
             stx  cx16.r0s+1
             stz  cx16.VERA_FX_CTRL    ; Cache write disable
+            stz  cx16.VERA_FX_MULT    ; $9F2C  reset multiply bit
             stz  cx16.VERA_CTRL       ; reset DCSEL
             rts
         }}
     }
 
     sub transparency(bool enable) {
+        ; Set transparent write mode for VeraFX cached writes and also for normal writes to DATA0/DATA.
+        ; If enabled, pixels with value 0 do not modify VRAM when written (so they are "transparent")
         cx16.VERA_CTRL = 2<<1       ; dcsel = 2
         if enable
             cx16.VERA_FX_CTRL |= %10000000

compiler/res/prog8lib/diskio.p8

@@ -333,6 +333,7 @@ m_in_buffer     sta  $ffff
 
     sub f_read_all(uword bufferpointer) -> uword {
         ; -- read the full contents of the file, returns number of bytes read.
+        ;    It is assumed the file size is less than 64 K.
         if not iteration_in_progress
             return 0
 
@@ -346,12 +347,13 @@ m_in_buffer     sta  $ffff
         return total_read
     }
 
-    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y {
+    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y, ubyte @A {
         ; Routine to read text lines from a text file. Lines must be less than 255 characters.
         ; Reads characters from the input file UNTIL a newline or return character (or EOF).
         ; The line read will be 0-terminated in the buffer (and not contain the end of line character).
         ; The length of the line is returned in Y. Note that an empty line is okay and is length 0!
         ; I/O error status should be checked by the caller itself via READST() routine.
+        ; The I/O error status byte is returned in A.
         %asm {{
             sta  P8ZP_SCRATCH_W1
             sty  P8ZP_SCRATCH_W1+1
@@ -368,7 +370,8 @@ _loop       jsr  cbm.CHRIN
 _line_end   dey     ; get rid of the trailing end-of-line char
             lda  #0
             sta  (P8ZP_SCRATCH_W1),y
-_end        rts
+_end        jsr  cbm.READST
+            rts
         }}
     }
 
@@ -405,6 +408,7 @@ _end        rts
 
     sub f_write(uword bufferpointer, uword num_bytes) -> bool {
         ; -- write the given number of bytes to the currently open file
+        ;    you can call this multiple times to append more data
         if num_bytes!=0 {
             reset_write_channel()
             repeat num_bytes {

compiler/res/prog8lib/math.asm

@@ -55,13 +55,14 @@ _multiplier      = P8ZP_SCRATCH_REG
 
 
 multiply_words	.proc
-	; -- multiply two 16-bit words into a 32-bit result  (signed and unsigned)
+	; -- multiply two 16-bit words into a 32-bit result  (UNSIGNED)
 	;      input: A/Y = first 16-bit number, multiply_words.multiplier = second 16-bit number
 	;      output: multiply_words.result, 4-bytes/32-bits product, LSB order (low-to-high)  low 16 bits also in AY.
+	;      you can retrieve the upper 16 bits via math.mul16_last_upper()
 
-	; NOTE: the result (which includes the multiplier parameter on entry) is a 4-byte array.
-	;       this routine could be faster if we could stick that into zeropage,
-	;       but there currently is no way to use 4 consecutive bytes in ZP (without disabling irq and saving/restoring them)...
+	; NOTE FOR NEGATIVE VALUES:
+	;      The routine also works for NEGATIVE (signed) word values, but ONLY the lower 16 bits of the result are correct then!
+	;      Prog8 only uses those so that's not an issue, but math.mul16_last_upper() no longer gives the correct result here.
 
 ; mult62.a
 ; from: https://github.com/TobyLobster/multiply_test/blob/main/tests/mult62.a
@@ -179,7 +180,7 @@ _inner_loop2
     ldy  result+1
     rts
 
-result		.byte  0,0,0,0
+result		.byte  0,0,0,0       ; routine could be faster if this were in Zeropage...
 
 		.pend
 

compiler/res/prog8lib/math.p8

@@ -168,6 +168,9 @@ _sinecosR8	.char  trunc(127.0 * sin(range(180+45) * rad(360.0/180.0)))
         ;     for instance, simply printing a number may already result in new multiplication calls being performed
         ;   - not all multiplications in the source code result in an actual multiplication call:
         ;     some simpler multiplications will be optimized away into faster routines. These will not set the upper 16 bits at all!
+        ;   - THE RESULT IS ONLY VALID IF THE MULTIPLICATION WAS DONE WITH UWORD ARGUMENTS (or two positive WORD arguments)
+        ;     as soon as a negative word value (or 2) was used in the multiplication, these upper 16 bits are not valid!!
+        ;     Suggestion (if you are on the Commander X16): use verafx.muls() to get a hardware accelerated 32 bit signed multplication.
         %asm {{
             lda  multiply_words.result+2
             ldy  multiply_words.result+3

compiler/res/prog8lib/pet32/syslib.p8

@@ -98,19 +98,24 @@ sys {
 
     const ubyte target = 32         ;  compilation target specifier.  64 = C64, 128 = C128,  16 = CommanderX16, 32=PET
 
+    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 floats supported
+
+
 asmsub  init_system()  {
     ; Initializes the machine to a sane starting state.
     ; Called automatically by the loader program logic.
     ; Uppercase charset is activated.
     %asm {{
         sei
-        cld
         lda  #142
         jsr  cbm.CHROUT     ; uppercase
         lda  #147
         jsr  cbm.CHROUT     ; clear screen
-        clc
-        clv
         cli
         rts
     }}
@@ -118,7 +123,10 @@ asmsub  init_system()  {
 
 asmsub  init_system_phase2()  {
     %asm {{
-        rts     ; no phase 2 steps on the PET
+        cld
+        clc
+        clv
+        rts
     }}
 }
 

compiler/res/prog8lib/prog8_funcs.asm

@@ -1,60 +1,5 @@
 ; ---- builtin functions
 
-
-func_any_b_into_A	.proc
-		; -- any(array),  array in P8ZP_SCRATCH_W1, num bytes in A
-		sta  _cmp_mod+1		; self-modifying code
-		ldy  #0
--		lda  (P8ZP_SCRATCH_W1),y
-		bne  _got_any
-		iny
-_cmp_mod	cpy  #255		; modified
-		bne  -
-		lda  #0
-		rts
-_got_any	lda  #1
-		rts
-		.pend
-
-
-func_all_b_into_A	.proc
-		; -- all(array),  array in P8ZP_SCRATCH_W1, num bytes in A
-		sta  _cmp_mod+1		; self-modifying code
-		ldy  #0
--		lda  (P8ZP_SCRATCH_W1),y
-		beq  _got_not_all
-		iny
-_cmp_mod	cpy  #255		; modified
-		bne  -
-		lda  #1
-_got_not_all	rts
-		.pend
-
-func_any_w_into_A	.proc
-		asl  a
-		jmp  func_any_b_into_A
-		.pend
-
-func_all_w_into_A	.proc
-		; -- all(warray),  array in P8ZP_SCRATCH_W1, num bytes in A
-		asl  a			; times 2 because of word
-		sta  _cmp_mod+1		; self-modifying code
-		ldy  #0
--		lda  (P8ZP_SCRATCH_W1),y
-		bne  +
-		iny
-		lda  (P8ZP_SCRATCH_W1),y
-		bne  ++
-		lda  #0
-		rts
-+		iny
-+		iny
-_cmp_mod	cpy  #255		; modified
-		bne  -
-		lda  #1
-		rts
-		.pend
-
 abs_b_into_A	.proc
 	; -- A = abs(A)
 		cmp  #0

compiler/res/prog8lib/prog8_lib.asm

@@ -5,6 +5,24 @@
 orig_stackpointer	.byte  0	; stores the Stack pointer register at program start
 
 
+program_startup_clear_bss    .proc
+	; this is always ran first thing from the start routine to clear out the BSS area
+	.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
+		rts
+		.pend
+
+
+
 read_byte_from_address_in_AY_into_A	.proc
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
@@ -248,31 +266,25 @@ _arg_s2		.word  0
 
 strcmp_mem	.proc
 		; --   compares strings in s1 (AY) and s2 (P8ZP_SCRATCH_W2).
-		;      Returns -1,0,1 in A, depeding on the ordering. Clobbers Y.
+		;      Returns -1,0,1 in A, depending on the ordering. Clobbers Y.
 		sta  P8ZP_SCRATCH_W1
 		sty  P8ZP_SCRATCH_W1+1
 		ldy  #0
-_loop		lda  (P8ZP_SCRATCH_W1),y
-		bne  +
-		lda  (P8ZP_SCRATCH_W2),y
-		bne  _return_minusone
-		beq  _return
-+		cmp  (P8ZP_SCRATCH_W2),y
-		bcc  _return_minusone
-		bne  _return_one
-		inc  P8ZP_SCRATCH_W1
-		bne  +
-		inc  P8ZP_SCRATCH_W1+1
-+		inc  P8ZP_SCRATCH_W2
-		bne  _loop
-		inc  P8ZP_SCRATCH_W2+1
-		bne  _loop
-_return_one
+_loop           lda  (P8ZP_SCRATCH_W1),y
+		beq  _c1_zero
+		cmp  (P8ZP_SCRATCH_W2),y
+		beq  _equal
+		bmi  _less
 		lda  #1
-_return		rts
-_return_minusone
-		lda  #-1
 		rts
+_less           lda  #-1
+		rts
+_equal          iny
+		bne  _loop
+_c1_zero        lda  (P8ZP_SCRATCH_W2),y
+		beq  +
+		lda  #-1
++		rts
 		.pend
 
 

compiler/res/prog8lib/shared_floats_functions.p8

@@ -96,7 +96,16 @@ sub atan(float value) -> float {
 ; two-argument arctangent that returns an angle in the correct quadrant
 ; for the signs of x and y, normalized to the range [0, 2π]
 sub atan2(float y, float x) -> float {
-    float atn = atan(y / x)
+    float atn
+    if x == 0 {
+        atn = π/2
+        if y == 0 return 0
+        if y < 0 {
+            atn += π
+        }
+    } else {
+        atn = atan(y / x)
+    }
     if x < 0 atn += π
     if atn < 0 atn += 2*π
     return atn

compiler/res/prog8lib/string.p8

@@ -151,6 +151,37 @@ _found	tya
         }}
     }
 
+    asmsub rfind(uword string @AY, ubyte character @X) -> ubyte @A, bool @Pc {
+        ; Locates the first position of the given character in the string, starting from the right.
+        ; returns Carry set if found + index in A, or Carry clear if not found (and A will be 255, an invalid index).
+        %asm {{
+            stx  P8ZP_SCRATCH_B1
+            sta  _str
+            sty  _str+1
+            jsr  string.length
+            dey
+            lda  _str
+    		sta  P8ZP_SCRATCH_W1
+    		lda  _str+1
+    		sta  P8ZP_SCRATCH_W1+1
+-           lda  (P8ZP_SCRATCH_W1),y
+            cmp  P8ZP_SCRATCH_B1
+            beq  _found
+            dey
+            cpy  #255
+            bne  -
+_notfound   lda  #255
+            clc
+            rts
+_found      tya
+            sec
+            rts
+
+_str    .word 0
+
+        }}
+    }
+
     asmsub contains(uword string @AY, ubyte character @X) -> bool @Pc {
         ; Just return true/false if the character is in the given string or not.
         %asm {{
@@ -199,9 +230,9 @@ _found	tya
 
     asmsub compare(uword string1 @R0, uword string2 @AY) clobbers(Y) -> byte @A {
         ; Compares two strings for sorting.
-        ; Returns -1 (255), 0 or 1 depending on wether string1 sorts before, equal or after string2.
+        ; Returns -1 (255), 0 or 1, meaning: string1 sorts before, equal or after string2.
         ; Note that you can also directly compare strings and string values with eachother using
-        ; comparison operators ==, < etcetera (it will use strcmp for you under water automatically).
+        ; comparison operators ==, < etcetera (this will use strcmp automatically).
         %asm {{
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1