implemented last remaining codegen for word-byte division and remainders.

Add support for built-in functions to Vim syntax file (and other fixes)

.gitignore

@@ -29,3 +29,5 @@ parsetab.py
 
 .gradle
 /prog8compiler.jar
+sd*.img
+

.idea/inspectionProfiles/Project_Default.xml

@@ -3,14 +3,11 @@
     <option name="myName" value="Project Default" />
     <inspection_tool class="DuplicatedCode" enabled="true" level="WEAK WARNING" enabled_by_default="true">
       <Languages>
-        <language minSize="100" isEnabled="false" name="JavaScript" />
-        <language isEnabled="false" name="Groovy" />
-        <language isEnabled="false" name="Style Sheets" />
         <language minSize="70" name="Kotlin" />
-        <language isEnabled="false" name="TypeScript" />
-        <language isEnabled="false" name="ActionScript" />
+        <language isEnabled="false" name="Groovy" />
       </Languages>
     </inspection_tool>
+    <inspection_tool class="PyInterpreterInspection" enabled="false" level="WARNING" enabled_by_default="false" />
     <inspection_tool class="SpellCheckingInspection" enabled="true" level="TYPO" enabled_by_default="true">
       <option name="processCode" value="false" />
       <option name="processLiterals" value="true" />

.idea/kotlinc.xml

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

.idea/libraries/antlr_4_7_2_complete.xml

@@ -1,9 +0,0 @@
-<component name="libraryTable">
-  <library name="antlr-4.7.2-complete">
-    <CLASSES>
-      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-4.7.2-complete.jar!/" />
-    </CLASSES>
-    <JAVADOC />
-    <SOURCES />
-  </library>
-</component>

.idea/libraries/antlr_4_9_complete.xml

@@ -1,7 +1,7 @@
 <component name="libraryTable">
-  <library name="antlr-4.8-complete">
+  <library name="antlr-4.9-complete">
     <CLASSES>
-      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-4.8-complete.jar!/" />
+      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-4.9-complete.jar!/" />
     </CLASSES>
     <JAVADOC />
     <SOURCES />

.idea/libraries/antlr_runtime_4_7_2.xml

@@ -1,9 +0,0 @@
-<component name="libraryTable">
-  <library name="antlr-runtime-4.7.2">
-    <CLASSES>
-      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-runtime-4.7.2.jar!/" />
-    </CLASSES>
-    <JAVADOC />
-    <SOURCES />
-  </library>
-</component>

.idea/libraries/antlr_runtime_4_9.xml

@@ -1,7 +1,7 @@
 <component name="libraryTable">
-  <library name="antlr-runtime-4.8">
+  <library name="antlr-runtime-4.9">
     <CLASSES>
-      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-runtime-4.8.jar!/" />
+      <root url="jar://$PROJECT_DIR$/parser/antlr/lib/antlr-runtime-4.9.jar!/" />
     </CLASSES>
     <JAVADOC />
     <SOURCES />

.idea/libraries/dbus_java_3_2_4.xml

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

.idea/libraries/javax_json_api_1_1_4.xml

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

.idea/libraries/kotlinx_cli_jvm.xml

@@ -1,7 +1,7 @@
 <component name="libraryTable">
-  <library name="kotlinx-cli-jvm-0.1.0-dev-5">
+  <library name="kotlinx-cli-jvm">
     <CLASSES>
-      <root url="jar://$PROJECT_DIR$/compiler/lib/kotlinx-cli-jvm-0.1.0-dev-5.jar!/" />
+      <root url="jar://$PROJECT_DIR$/compiler/lib/kotlinx-cli-jvm-0.3.1.jar!/" />
     </CLASSES>
     <JAVADOC />
     <SOURCES />

.idea/libraries/slf4j_api_1_7_30.xml

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

.idea/libraries/takes_http.xml

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

.idea/misc.xml

@@ -16,7 +16,7 @@
       </list>
     </option>
   </component>
-  <component name="ProjectRootManager" version="2" languageLevel="JDK_1_8" default="false" project-jdk-name="Kotlin SDK" project-jdk-type="KotlinSDK">
+  <component name="ProjectRootManager" version="2" languageLevel="JDK_11" default="false" project-jdk-name="Kotlin SDK" project-jdk-type="KotlinSDK">
     <output url="file://$PROJECT_DIR$/out" />
   </component>
 </project>

.idea/modules.xml

@@ -3,8 +3,11 @@
   <component name="ProjectModuleManager">
     <modules>
       <module fileurl="file://$PROJECT_DIR$/compiler/compiler.iml" filepath="$PROJECT_DIR$/compiler/compiler.iml" />
+      <module fileurl="file://$PROJECT_DIR$/compilerAst/compilerAst.iml" filepath="$PROJECT_DIR$/compilerAst/compilerAst.iml" />
+      <module fileurl="file://$PROJECT_DIR$/dbusCompilerService/dbusCompilerService.iml" filepath="$PROJECT_DIR$/dbusCompilerService/dbusCompilerService.iml" />
       <module fileurl="file://$PROJECT_DIR$/docs/docs.iml" filepath="$PROJECT_DIR$/docs/docs.iml" />
       <module fileurl="file://$PROJECT_DIR$/examples/examples.iml" filepath="$PROJECT_DIR$/examples/examples.iml" />
+      <module fileurl="file://$PROJECT_DIR$/httpCompilerService/httpCompilerService.iml" filepath="$PROJECT_DIR$/httpCompilerService/httpCompilerService.iml" />
       <module fileurl="file://$PROJECT_DIR$/parser/parser.iml" filepath="$PROJECT_DIR$/parser/parser.iml" />
     </modules>
   </component>

README.md

@@ -20,37 +20,42 @@ Full documentation (syntax reference, how to use the language and the compiler,
 https://prog8.readthedocs.io/
 
 
-What use Prog8 provide?
------------------------
+What does Prog8 provide?
+------------------------
 
 - reduction of source code length over raw assembly
 - modularity, symbol scoping, subroutines
 - various data types other than just bytes (16-bit words, floats, strings)
 - automatic variable allocations, automatic string and array variables and string sharing
-- subroutines with an input- and output parameter signature
-- constant folding in expressions
+- subroutines with input parameters and result values
+- high-level program optimizations
+- small program boilerplate/compilersupport overhead
+- sane variable initialization, programs can be restarted again just fine after exiting to basic
 - conditional branches
+- floating point operations  (requires the C64 Basic ROM routines for this)
 - 'when' statement to provide a concise jump table alternative to if/elseif chains
 - structs to group together sets of variables and manipulate them at once
-- floating point operations  (requires the C64 Basic ROM routines for this)
-- abstracting away low level aspects such as ZeroPage handling, program startup, explicit memory addresses
-- various code optimizations (code structure, logical and numerical expressions, unused code removal...)
+- many built-in functions such as ``sin``, ``cos``, ``rnd``, ``abs``, ``min``, ``max``, ``sqrt``, ``msb``, ``rol``, ``ror``, ``swap``, ``sort`` and ``reverse``
+- various powerful built-in libraries to do I/O, number conversions, graphics and more  
+- convenience abstractions for low level aspects such as ZeroPage handling, program startup, explicit memory addresses
+- fast execution speed due to compilation to native assembly code
+- variables are allocated statically
 - inline assembly allows you to have full control when every cycle or byte matters
-- many built-in functions such as ``sin``, ``cos``, ``rnd``, ``abs``, ``min``, ``max``, ``sqrt``, ``msb``, ``rol``, ``ror``, ``swap``, ``memset``, ``memcopy``, ``sort`` and ``reverse``
+- supports the sixteen 'virtual' 16-bit registers R0 .. R15 from the Commander X16, and provides them also on the C64.
+- encode strings and characters into petscii or screencodes as desired (C64/Cx16)
 
 *Rapid edit-compile-run-debug cycle:*
 
-- use a modern PC to do the work on
-- very quick compilation times
+- use a modern PC to do the work on, use nice editors and enjoy quick compilation times
 - can automatically run the program in the Vice emulator after succesful compilation
 - breakpoints, that let the Vice emulator drop into the monitor if execution hits them
 - source code labels automatically loaded in Vice emulator so it can show them in disassembly
 
 *Two supported compiler targets* (contributions to improve these or to add support for other machines are welcome!):
 
-- "c64": Commodore-64  (6510 CPU = almost a 6502)  premium support.
-- "cx16": [CommanderX16](https://www.commanderx16.com)  (65c02 CPU)  experimental support.
-- If you only use standard kernel and prog8 library routines, it is possible to compile the *exact same program* for both machines (just change the compiler target flag)!
+- "c64": Commodore-64  (6510 CPU = almost a 6502)
+- "cx16": [CommanderX16](https://www.commanderx16.com)  (65c02 CPU)
+- If you only use standard kernal and prog8 library routines, it is possible to compile the *exact same program* for both machines (just change the compiler target flag)!
 
 
 
@@ -61,7 +66,7 @@ Additional required tools
 A recent .exe version of this tool for Windows can be obtained from my [clone](https://github.com/irmen/64tass/releases) of this project.
 For other platforms it is very easy to compile it yourself (make ; make install).
 
-A **Java runtime (jre or jdk), version 8 or newer**  is required to run a prepackaged version of the compiler.
+A **Java runtime (jre or jdk), version 11 or newer**  is required to run a prepackaged version of the compiler.
 If you want to build it from source, you'll need a Java SDK + Kotlin 1.3.x SDK (or for instance,
 IntelliJ IDEA with the Kotlin plugin).
 
@@ -84,9 +89,7 @@ This code calculates prime numbers using the Sieve of Eratosthenes algorithm::
         ubyte candidate_prime = 2       ; is increased in the loop
     
         sub start() {
-            ; clear the sieve, to reset starting situation on subsequent runs
-            memset(sieve, 256, false)   
-            ; calculate primes
+            sys.memset(sieve, 256, false)   ; clear the sieve
             txt.print("prime numbers up to 255:\n\n")
             ubyte amount=0
             repeat {
@@ -97,17 +100,17 @@ This code calculates prime numbers using the Sieve of Eratosthenes algorithm::
                 txt.print(", ")
                 amount++
             }
-            txt.chrout('\n')
+            txt.nl()
             txt.print("number of primes (expected 54): ")
             txt.print_ub(amount)
-            txt.chrout('\n')
+            txt.nl()
         }
     
         sub find_next_prime() -> ubyte {
             while sieve[candidate_prime] {
                 candidate_prime++
                 if candidate_prime==0
-                    return 0        ; we wrapped; no more primes available in the sieve
+                    return 0        ; we wrapped; no more primes
             }
     
             ; found next one, mark the multiples and return it.
@@ -124,6 +127,7 @@ This code calculates prime numbers using the Sieve of Eratosthenes algorithm::
 
 
 
+
 when compiled an ran on a C-64 you'll get:
 
 ![c64 screen](docs/source/_static/primes_example.png)
@@ -140,7 +144,8 @@ If you want to play a video game, a fully working Tetris clone is included in th
 
 ![tehtriz_screen](docs/source/_static/tehtriz.png)
 
-The CommanderX16 compiler target is quite capable already too, here's a well known space ship 
-animated in 3D with hidden line removal, in the CommanderX16 emulator:
+There are a couple of examples specially made for the CommanderX16 compiler target.
+For instance here's a well known space ship animated in 3D with hidden line removal,
+in the CommanderX16 emulator:
 
 ![cobra3d](docs/source/_static/cobra3d.png)

compiler/build.gradle

@@ -1,41 +1,29 @@
-buildscript {
-    dependencies {
-        classpath "org.jetbrains.kotlin:kotlin-gradle-plugin:1.4.10"
-    }
-}
-
 plugins {
-    // id "org.jetbrains.kotlin.jvm" version "1.4.10"
-    id 'application'
-    id 'org.jetbrains.dokka' version "0.9.18"
-    id 'com.github.johnrengelman.shadow' version '5.2.0'
     id 'java'
+    id 'application'
+    id "org.jetbrains.kotlin.jvm" version "1.4.30"
+    id 'org.jetbrains.dokka' version "0.9.18"
+    id 'com.github.johnrengelman.shadow' version '6.1.0'
 }
 
-apply plugin: "kotlin"
-apply plugin: "java"
-
-targetCompatibility = 1.8
-sourceCompatibility = 1.8
+targetCompatibility = 11
+sourceCompatibility = 11
 
 repositories {
     mavenLocal()
     mavenCentral()
-    jcenter()
-    maven { url "https://dl.bintray.com/orangy/maven/" }
+    maven { url "https://kotlin.bintray.com/kotlinx" }
 }
 
 def prog8version = rootProject.file('compiler/res/version.txt').text.trim()
 
 dependencies {
-    implementation project(':parser')
+    implementation project(':compilerAst')
     implementation "org.jetbrains.kotlin:kotlin-stdlib-jdk8"
     // implementation "org.jetbrains.kotlin:kotlin-reflect"
-    implementation 'org.antlr:antlr4-runtime:4.8'
-    implementation 'org.jetbrains.kotlinx:kotlinx-cli-jvm:0.1.0-dev-5'
-    // implementation 'net.razorvine:ksim65:1.6'
-    // implementation "com.github.hypfvieh:dbus-java:3.2.0"
-    implementation project(':parser')
+    implementation 'org.jetbrains.kotlinx:kotlinx-cli:0.3.1'
+    // implementation 'net.razorvine:ksim65:1.8'
+    // implementation "com.github.hypfvieh:dbus-java:3.2.4"
 
     testImplementation "org.jetbrains.kotlin:kotlin-test-junit5"
     testImplementation 'org.junit.jupiter:junit-jupiter-api:5.3.2'
@@ -45,7 +33,8 @@ dependencies {
 
 compileKotlin {
     kotlinOptions {
-        jvmTarget = "1.8"
+        jvmTarget = "11"
+        useIR = true
         // verbose = true
         // freeCompilerArgs += "-XXLanguage:+NewInference"
     }
@@ -53,7 +42,8 @@ compileKotlin {
 
 compileTestKotlin {
     kotlinOptions {
-        jvmTarget = "1.8"
+        jvmTarget = "11"
+        useIR = true
     }
 }
 
@@ -76,7 +66,8 @@ sourceSets {
 startScripts.enabled = true
 
 application {
-    mainClassName = 'prog8.CompilerMainKt'
+    mainClass = 'prog8.CompilerMainKt'
+    mainClassName = 'prog8.CompilerMainKt'  // deprecated
     applicationName = 'p8compile'
 }
 
@@ -112,5 +103,5 @@ dokka {
 }
 
 task wrapper(type: Wrapper) {
-    gradleVersion = '6.1.1'
+    gradleVersion = '6.7'
 }

compiler/compiler.iml

@@ -1,5 +1,10 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <module type="JAVA_MODULE" version="4">
+  <component name="FacetManager">
+    <facet type="Python" name="Python">
+      <configuration sdkName="Python 3.9" />
+    </facet>
+  </component>
   <component name="NewModuleRootManager" inherit-compiler-output="true">
     <exclude-output />
     <content url="file://$MODULE_DIR$">
@@ -11,9 +16,9 @@
     <orderEntry type="jdk" jdkName="11" jdkType="JavaSDK" />
     <orderEntry type="sourceFolder" forTests="false" />
     <orderEntry type="library" name="KotlinJavaRuntime" level="project" />
-    <orderEntry type="module" module-name="parser" />
     <orderEntry type="library" name="unittest-libs" level="project" />
-    <orderEntry type="library" name="kotlinx-cli-jvm-0.1.0-dev-5" level="project" />
-    <orderEntry type="library" name="antlr-runtime-4.8" level="project" />
+    <orderEntry type="library" name="kotlinx-cli-jvm" level="project" />
+    <orderEntry type="module" module-name="compilerAst" />
+    <orderEntry type="library" name="Python 3.9 interpreter library" level="application" />
   </component>
 </module>

compiler/res/.editorconfig

@@ -0,0 +1,20 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+indent_size = 4
+indent_style = space
+insert_final_newline = true
+max_line_length = 120
+tab_width = 8
+trim_trailing_whitespace = true
+ij_smart_tabs = true
+
+[*.p8]
+indent_size = 4
+indent_style = space
+
+[*.asm]
+indent_size = 8
+indent_style = tab

compiler/res/prog8lib/c64/floats.asm

@@ -2,6 +2,8 @@
 
 FL_ONE_const	.byte  129     			; 1.0
 FL_ZERO_const	.byte  0,0,0,0,0		; 0.0
+FL_LOG2_const	.byte  $80, $31, $72, $17, $f8	; log(2)
+
 
 floats_store_reg	.byte  0		; temp storage
 
@@ -89,17 +91,27 @@ cast_from_b	.proc
 
 cast_as_uw_into_ya	.proc               ; also used for float 2 ub
 		; -- cast float at A/Y to uword into Y/A
-		stx  P8ZP_SCRATCH_REG
 		jsr  MOVFM
+		jmp  cast_FAC1_as_uw_into_ya
+		.pend
+
+cast_as_w_into_ay	.proc               ; also used for float 2 b
+		; -- cast float at A/Y to word into A/Y
+		jsr  MOVFM
+		jmp  cast_FAC1_as_w_into_ay
+		.pend
+
+cast_FAC1_as_uw_into_ya	.proc               ; also used for float 2 ub
+		; -- cast fac1 to uword into Y/A
+		stx  P8ZP_SCRATCH_REG
 		jsr  GETADR     ; into Y/A
 		ldx  P8ZP_SCRATCH_REG
 		rts
 		.pend
 
-cast_as_w_into_ay	.proc               ; also used for float 2 b
-		; -- cast float at A/Y to word into A/Y
+cast_FAC1_as_w_into_ay	.proc               ; also used for float 2 b
+		; -- cast fac1 to word into A/Y
 		stx  P8ZP_SCRATCH_REG
-		jsr  MOVFM
 		jsr  AYINT
 		ldy  $64
 		lda  $65
@@ -380,6 +392,71 @@ neg_f		.proc
 		rts
 		.pend
 
+var_fac1_less_f	.proc
+		; -- is the float in FAC1 < the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #255
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_lesseq_f	.proc
+		; -- is the float in FAC1 <= the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #0
+		beq  +
+		cmp  #255
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_greater_f	.proc
+		; -- is the float in FAC1 > the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #1
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_greatereq_f	.proc
+		; -- is the float in FAC1 >= the variable AY?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		cmp  #0
+		beq  +
+		cmp  #1
+		beq  +
+		lda  #0
+		rts
++		lda  #1
+		rts
+		.pend
+
+var_fac1_notequal_f	.proc
+		; -- are the floats numbers in FAC1 and the variable AY *not* identical?
+		stx  P8ZP_SCRATCH_REG
+		jsr  FCOMP
+		ldx  P8ZP_SCRATCH_REG
+		and  #1
+		rts
+		.pend
+
 vars_equal_f	.proc
 		; -- are the mflpt5 numbers in P8ZP_SCRATCH_W1 and AY identical?
 		sta  P8ZP_SCRATCH_W2

compiler/res/prog8lib/c64/floats.p8

@@ -19,23 +19,6 @@ floats {
 		; note: the fac1 and fac2 are working registers and take 6 bytes each,
 		; floats in memory  (and rom) are stored in 5-byte MFLPT packed format.
 
-		; constants in five-byte "mflpt" format in the BASIC ROM
-		&float  FL_PIVAL	= $aea8  ; 3.1415926...
-		&float  FL_N32768	= $b1a5  ; -32768
-		&float  FL_FONE		= $b9bc  ; 1
-		&float  FL_SQRHLF	= $b9d6  ; SQR(2) / 2
-		&float  FL_SQRTWO	= $b9db  ; SQR(2)
-		&float  FL_NEGHLF	= $b9e0  ; -.5
-		&float  FL_LOG2		= $b9e5  ; LOG(2)
-		&float  FL_TENC		= $baf9  ; 10
-		&float  FL_NZMIL	= $bdbd  ; 1e9 (1 billion)
-		&float  FL_FHALF	= $bf11  ; .5
-		&float  FL_LOGEB2	= $bfbf  ; 1 / LOG(2)
-		&float  FL_PIHALF	= $e2e0  ; PI / 2
-		&float  FL_TWOPI	= $e2e5  ; 2 * PI
-		&float  FL_FR4		= $e2ea  ; .25
-		; oddly enough, 0.0 isn't available in the kernel.
-
 
 ; note: fac1/2 might get clobbered even if not mentioned in the function's name.
 ; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.

compiler/res/prog8lib/c64/floats_funcs.asm

@@ -175,8 +175,8 @@ func_log2_fac1	.proc
 		stx  P8ZP_SCRATCH_REG
 		jsr  LOG
 		jsr  MOVEF
-		lda  #<FL_LOG2
-		ldy  #>FL_LOG2
+		lda  #<FL_LOG2_const
+		ldy  #>FL_LOG2_const
 		jsr  MOVFM
 		jsr  FDIVT
 		ldx  P8ZP_SCRATCH_REG

compiler/res/prog8lib/c64/graphics.p8

@@ -2,7 +2,7 @@
 %import textio
 
 ; bitmap pixel graphics module for the C64
-; only black/white monchrome 320x200 for now
+; only black/white monochrome 320x200 for now
 ; assumes bitmap screen memory is $2000-$3fff
 
 graphics {
@@ -12,7 +12,7 @@ graphics {
 
     sub enable_bitmap_mode() {
         ; enable bitmap screen, erase it and set colors to black/white.
-        c64.SCROLY = %00111000
+        c64.SCROLY = %00111011
         c64.SCROLX = %00001000
         c64.VMCSB = (c64.VMCSB & %11110000) | %00001000   ; $2000-$3fff
         clear_screen(1, 0)
@@ -20,36 +20,47 @@ graphics {
 
     sub disable_bitmap_mode() {
         ; enables text mode, erase the text screen, color white
-        c64.SCROLY = %00011000
+        c64.SCROLY = %00011011
         c64.SCROLX = %00001000
         c64.VMCSB = (c64.VMCSB & %11110000) | %00000100   ; $1000-$2fff
         txt.fill_screen(' ', 1)
     }
 
     sub clear_screen(ubyte pixelcolor, ubyte bgcolor) {
-        memset(BITMAP_ADDRESS, 320*200/8, 0)
+        sys.memset(BITMAP_ADDRESS, 320*200/8, 0)
         txt.fill_screen(pixelcolor << 4 | bgcolor, 0)
     }
 
     sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
         ; Bresenham algorithm.
-        ; This code special cases various quadrant loops to allow simple ++ and -- operations.
-        ; TODO rewrite this in optimized assembly
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
         if y1>y2 {
-            ; make sure dy is always positive to avoid 8 instead of just 4 special cases
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
             swap(x1, x2)
             swap(y1, y2)
         }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as ubyte +1)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1)
+            return
+        }
+
         word @zp d = 0
         ubyte positive_ix = true
-        word @zp dx = x2-x1 as word
-        word @zp dy = y2-y1
         if dx < 0 {
             dx = -dx
             positive_ix = false
         }
-        dx *= 2
-        dy *= 2
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
         internal_plotx = x1
 
         if dx >= dy {
@@ -59,10 +70,10 @@ graphics {
                     if internal_plotx==x2
                         return
                     internal_plotx++
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             } else {
@@ -71,10 +82,10 @@ graphics {
                     if internal_plotx==x2
                         return
                     internal_plotx--
-                    d += dy
+                    d += dy2
                     if d > dx {
                         y1++
-                        d -= dx
+                        d -= dx2
                     }
                 }
             }
@@ -86,10 +97,10 @@ graphics {
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         internal_plotx++
-                        d -= dy
+                        d -= dy2
                     }
                 }
             } else {
@@ -98,88 +109,183 @@ graphics {
                     if y1 == y2
                         return
                     y1++
-                    d += dx
+                    d += dx2
                     if d > dy {
                         internal_plotx--
-                        d -= dy
+                        d -= dy2
                     }
                 }
             }
         }
     }
 
+    sub rect(uword x, ubyte y, uword width, ubyte height) {
+        if width==0 or height==0
+            return
+        horizontal_line(x, y, width)
+        if height==1
+            return
+        horizontal_line(x, y+height-1, width)
+        vertical_line(x, y+1, height-2)
+        if width==1
+            return
+        vertical_line(x+width-1, y+1, height-2)
+    }
+
+    sub fillrect(uword x, ubyte y, uword width, ubyte height) {
+        if width==0
+            return
+        repeat height {
+            horizontal_line(x, y, width)
+            y++
+        }
+    }
+
+    sub horizontal_line(uword x, ubyte y, uword length) {
+        if length<8 {
+            internal_plotx=x
+            repeat lsb(length) {
+                internal_plot(y)
+                internal_plotx++
+            }
+            return
+        }
+
+        ubyte separate_pixels = lsb(x) & 7
+        uword addr = get_y_lookup(y) + (x&$fff8)
+
+        if separate_pixels {
+            %asm {{
+                lda  addr
+                sta  P8ZP_SCRATCH_W1
+                lda  addr+1
+                sta  P8ZP_SCRATCH_W1+1
+                ldy  separate_pixels
+                lda  _filled_right,y
+                eor  #255
+                ldy  #0
+                ora  (P8ZP_SCRATCH_W1),y
+                sta  (P8ZP_SCRATCH_W1),y
+            }}
+            addr += 8
+            length += separate_pixels
+            length -= 8
+        }
+
+        if length {
+            %asm {{
+                lda  length
+                and  #7
+                sta  separate_pixels
+                stx  P8ZP_SCRATCH_REG
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lsr  length+1
+                ror  length
+                lda  addr
+                sta  _modified+1
+                lda  addr+1
+                sta  _modified+2
+                lda  length
+                ora  length+1
+                beq  _zero
+                ldy  length
+                ldx  #$ff
+_modified       stx  $ffff      ; modified
+                lda  _modified+1
+                clc
+                adc  #8
+                sta  _modified+1
+                bcc  +
+                inc  _modified+2
++               dey
+                bne  _modified
+_zero           ldx  P8ZP_SCRATCH_REG
+
+                ldy  separate_pixels
+                beq  _zero2
+                lda  _modified+1
+                sta  P8ZP_SCRATCH_W1
+                lda  _modified+2
+                sta  P8ZP_SCRATCH_W1+1
+                lda  _filled_right,y
+                ldy  #0
+                ora  (P8ZP_SCRATCH_W1),y
+                sta  (P8ZP_SCRATCH_W1),y
+                jmp  _zero2
+_filled_right   .byte  0, %10000000, %11000000, %11100000, %11110000, %11111000, %11111100, %11111110
+_zero2
+            }}
+        }
+    }
+
+    sub vertical_line(uword x, ubyte y, ubyte height) {
+        internal_plotx = x
+        repeat height {
+            internal_plot(y)
+            y++
+        }
+    }
+
     sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
         ; Midpoint algorithm
+        if radius==0
+            return
         ubyte @zp ploty
-        ubyte @zp xx = radius
         ubyte @zp yy = 0
-        byte @zp decisionOver2 = 1-xx as byte
+        word @zp decisionOver2 = (1 as word)-radius
 
-        while xx>=yy {
-            internal_plotx = xcenter + xx
+        while radius>=yy {
+            internal_plotx = xcenter + radius
             ploty = ycenter + yy
             internal_plot(ploty)
-            internal_plotx = xcenter - xx
+            internal_plotx = xcenter - radius
             internal_plot(ploty)
-            internal_plotx = xcenter + xx
+            internal_plotx = xcenter + radius
             ploty = ycenter - yy
             internal_plot(ploty)
-            internal_plotx = xcenter - xx
+            internal_plotx = xcenter - radius
             internal_plot(ploty)
             internal_plotx = xcenter + yy
-            ploty = ycenter + xx
+            ploty = ycenter + radius
             internal_plot(ploty)
             internal_plotx = xcenter - yy
             internal_plot(ploty)
             internal_plotx = xcenter + yy
-            ploty = ycenter - xx
+            ploty = ycenter - radius
             internal_plot(ploty)
             internal_plotx = xcenter - yy
             internal_plot(ploty)
             yy++
             if decisionOver2<=0
-                decisionOver2 += 2*yy+1
+                decisionOver2 += (yy as word)*2+1
             else {
-                xx--
-                decisionOver2 += 2*(yy-xx)+1
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
             }
         }
     }
 
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
         ; Midpoint algorithm, filled
-        ubyte xx = radius
-        ubyte yy = 0
-        byte decisionOver2 = 1-xx as byte
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word decisionOver2 = (1 as word)-radius
 
-        while xx>=yy {
-            ubyte ycenter_plus_yy = ycenter + yy
-            ubyte ycenter_min_yy = ycenter - yy
-            ubyte ycenter_plus_xx = ycenter + xx
-            ubyte ycenter_min_xx = ycenter - xx
-
-            for internal_plotx in xcenter to xcenter+xx {
-                internal_plot(ycenter_plus_yy)
-                internal_plot(ycenter_min_yy)
-            }
-            for internal_plotx in xcenter-xx to xcenter-1 {
-                internal_plot(ycenter_plus_yy)
-                internal_plot(ycenter_min_yy)
-            }
-            for internal_plotx in xcenter to xcenter+yy {
-                internal_plot(ycenter_plus_xx)
-                internal_plot(ycenter_min_xx)
-            }
-            for internal_plotx in xcenter-yy to xcenter {
-                internal_plot(ycenter_plus_xx)
-                internal_plot(ycenter_min_xx)
-            }
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
             yy++
             if decisionOver2<=0
-                decisionOver2 += 2*yy+1
+                decisionOver2 += (yy as word)*2+1
             else {
-                xx--
-                decisionOver2 += 2*(yy-xx)+1
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
             }
         }
     }
@@ -192,11 +298,11 @@ graphics {
 ;        @(addr) |= ormask[lsb(px) & 7]
 ;    }
 
-    asmsub  plot(uword plotx @XY, ubyte ploty @A) clobbers (A, X, Y) {
+    inline asmsub  plot(uword plotx @XY, ubyte ploty @A) clobbers (A, X, Y) {
         %asm {{
-            stx  internal_plotx
-            sty  internal_plotx+1
-            jmp  internal_plot
+            stx  graphics.internal_plotx
+            sty  graphics.internal_plotx+1
+            jsr  graphics.internal_plot
         }}
     }
 
@@ -248,6 +354,17 @@ _y_lookup_hi    .byte  >_plot_y_values
         }}
     }
 
+    asmsub get_y_lookup(ubyte y @Y) -> uword @AY {
+        %asm {{
+            lda  internal_plot._y_lookup_lo,y
+            pha
+            lda  internal_plot._y_lookup_hi,y
+            tay
+            pla
+            rts
+        }}
+    }
+
 }
 
 

compiler/res/prog8lib/c64/syslib.p8

@@ -11,7 +11,7 @@ c64 {
         &ubyte  TIME_HI         = $a0       ; software jiffy clock, hi byte
         &ubyte  TIME_MID        = $a1       ;  .. mid byte
         &ubyte  TIME_LO         = $a2       ;    .. lo byte. Updated by IRQ every 1/60 sec
-        &ubyte  STATUS          = $90       ; kernel status variable for I/O
+        &ubyte  STATUS          = $90       ; kernal status variable for I/O
         &ubyte  STKEY           = $91       ; various keyboard statuses (updated by IRQ)
         &ubyte  SFDX            = $cb       ; current key pressed (matrix value) (updated by IRQ)
 
@@ -178,7 +178,7 @@ c64 {
 
 ; ---- C64 ROM kernal routines ----
 
-romsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use c64scr.print instead)
+romsub $AB1E = STROUT(uword strptr @ AY) clobbers(A, X, Y)      ; print null-terminated string (use txt.print instead)
 romsub $E544 = CLEARSCR() clobbers(A,X,Y)                       ; clear the screen
 romsub $E566 = HOMECRSR() clobbers(A,X,Y)                       ; cursor to top left of screen
 romsub $EA31 = IRQDFRT() clobbers(A,X,Y)                        ; default IRQ routine
@@ -202,7 +202,7 @@ romsub $FFAE = UNLSN() clobbers(A)                              ; command serial
 romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
 romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
 romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word
-romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte address @ Y)   ; set logical file parameters
+romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
 romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
 romsub $FFC0 = OPEN() clobbers(X,Y) -> ubyte @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
 romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
@@ -211,8 +211,8 @@ romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320
 romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
 romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
 romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
-romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, ubyte @ X, ubyte @ Y     ; (via 816 ($330)) load from device
-romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A                    ; (via 818 ($332)) save to a device
+romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
 romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
 romsub $FFE1 = STOP() clobbers(X) -> ubyte @ Pz, ubyte @ A      ; (via 808 ($328)) check the STOP key (and some others in A)
@@ -225,6 +225,41 @@ romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base addr
 
 ; ---- end of C64 ROM kernal routines ----
 
+; ---- utilities -----
+
+asmsub STOP2() -> ubyte @A  {
+    ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
+    %asm {{
+        txa
+        pha
+        jsr  c64.STOP
+        beq  +
+        pla
+        tax
+        lda  #0
+        rts
++       pla
+        tax
+        lda  #1
+        rts
+    }}
+}
+
+asmsub RDTIM16() -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    %asm {{
+        stx  P8ZP_SCRATCH_REG
+        jsr  c64.RDTIM
+        pha
+        txa
+        tay
+        pla
+        ldx  P8ZP_SCRATCH_REG
+        rts
+    }}
+}
+
+
 
 ; ---- C64 specific system utility routines: ----
 
@@ -259,17 +294,13 @@ asmsub  init_system()  {
     }}
 }
 
-asmsub  reset_system()  {
-    ; Soft-reset the system back to Basic prompt.
+asmsub  init_system_phase2()  {
     %asm {{
-        sei
-        lda  #14
-        sta  $01        ; bank the kernal in
-        jmp  (c64.RESET_VEC)
+        rts     ; no phase 2 steps on the C64
     }}
 }
 
-asmsub  disable_runstop_and_charsetswitch() {
+asmsub  disable_runstop_and_charsetswitch() clobbers(A) {
     %asm {{
         lda  #$80
         sta  657    ; disable charset switching
@@ -279,27 +310,13 @@ asmsub  disable_runstop_and_charsetswitch() {
     }}
 }
 
-asmsub  set_irqvec_excl() clobbers(A)  {
-	%asm {{
-		sei
-		lda  #<_irq_handler
-		sta  c64.CINV
-		lda  #>_irq_handler
-		sta  c64.CINV+1
-		cli
-		rts
-_irq_handler	jsr  set_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		lda  c64.CIA1ICR		; acknowledge CIA1 interrupt
-		jmp  c64.IRQDFEND		; end irq processing - don't call kernel
-	}}
-}
-
-asmsub  set_irqvec() clobbers(A)  {
+asmsub  set_irq(uword handler @AY, ubyte useKernal @Pc) clobbers(A)  {
 	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  _use_kernal
 		sei
 		lda  #<_irq_handler
 		sta  c64.CINV
@@ -308,9 +325,23 @@ asmsub  set_irqvec() clobbers(A)  {
 		cli
 		rts
 _irq_handler    jsr  _irq_handler_init
-		jsr  irq.irq
+_modified	jsr  $ffff                      ; modified
 		jsr  _irq_handler_end
-		jmp  c64.IRQDFRT		; continue with normal kernel irq routine
+		lda  _use_kernal
+		bne  +
+		lda  #$ff
+		sta  c64.VICIRQ			; acknowledge raster irq
+		lda  c64.CIA1ICR		; acknowledge CIA1 interrupt
+		; end irq processing - don't use kernal's irq handling
+		pla
+		tay
+		pla
+		tax
+		pla
+		rti
++		jmp  c64.IRQDFRT		; continue with normal kernal irq routine
+
+_use_kernal     .byte  0
 
 _irq_handler_init
 		; save all zp scratch registers and the X register as these might be clobbered by the irq routine
@@ -363,7 +394,7 @@ IRQ_SCRATCH_ZPWORD2	.word  0
 		}}
 }
 
-asmsub  restore_irqvec() {
+asmsub  restore_irq() clobbers(A) {
 	%asm {{
 		sei
 		lda  #<c64.IRQDFRT
@@ -379,8 +410,15 @@ asmsub  restore_irqvec() {
 	}}
 }
 
-asmsub  set_rasterirq(uword rasterpos @ AY) clobbers(A) {
+asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0, ubyte useKernal @Pc) clobbers(A) {
 	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  set_irq._use_kernal
+		lda  cx16.r0
+		ldy  cx16.r0+1
 		sei
 		jsr  _setup_raster_irq
 		lda  #<_raster_irq_handler
@@ -391,12 +429,21 @@ asmsub  set_rasterirq(uword rasterpos @ AY) clobbers(A) {
 		rts
 
 _raster_irq_handler
-		jsr  set_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		jmp  c64.IRQDFRT
+		jsr  set_irq._irq_handler_init
+_modified	jsr  $ffff              ; modified
+		jsr  set_irq._irq_handler_end
+                lda  #$ff
+                sta  c64.VICIRQ			; acknowledge raster irq
+		lda  set_irq._use_kernal
+		bne  +
+		; end irq processing - don't use kernal's irq handling
+		pla
+		tay
+		pla
+		tax
+		pla
+		rti
++		jmp  c64.IRQDFRT                ; continue with kernal irq routine
 
 _setup_raster_irq
 		pha
@@ -420,29 +467,216 @@ _setup_raster_irq
 	}}
 }
 
-asmsub  set_rasterirq_excl(uword rasterpos @ AY) clobbers(A) {
-	%asm {{
-		sei
-		jsr  set_rasterirq._setup_raster_irq
-		lda  #<_raster_irq_handler
-		sta  c64.CINV
-		lda  #>_raster_irq_handler
-		sta  c64.CINV+1
-		cli
-		rts
-
-_raster_irq_handler
-		jsr  set_irqvec._irq_handler_init
-		jsr  irq.irq
-		jsr  set_irqvec._irq_handler_end
-		lda  #$ff
-		sta  c64.VICIRQ			; acknowledge raster irq
-		jmp  c64.IRQDFEND		; end irq processing - don't call kernel
-
-	}}
-}
-
 ; ---- end of C64 specific system utility routines ----
 
+}
+
+sys {
+    ; ------- lowlevel system routines --------
+
+    const ubyte target = 64         ;  compilation target specifier.  64 = C64,  16 = CommanderX16.
+
+
+    asmsub  reset_system()  {
+        ; Soft-reset the system back to Basic prompt.
+        %asm {{
+            sei
+            lda  #14
+            sta  $01        ; bank the kernal in
+            jmp  (c64.RESET_VEC)
+        }}
+    }
+
+    sub wait(uword jiffies) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds)
+        repeat jiffies {
+            ubyte jiff = lsb(c64.RDTIM16())
+            while jiff==lsb(c64.RDTIM16()) {
+                ; wait until 1 jiffy has passed
+            }
+        }
+    }
+
+    asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1        ; source in ZP
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2        ; target in ZP
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            cpy  #0
+            bne  _longcopy
+
+            ; copy <= 255 bytes
+            tay
+            bne  _copyshort
+            rts     ; nothing to copy
+
+_copyshort
+            ; decrease source and target pointers so we can simply index by Y
+            lda  P8ZP_SCRATCH_W1
+            bne  +
+            dec  P8ZP_SCRATCH_W1+1
++           dec  P8ZP_SCRATCH_W1
+            lda  P8ZP_SCRATCH_W2
+            bne  +
+            dec  P8ZP_SCRATCH_W2+1
++           dec  P8ZP_SCRATCH_W2
+
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            dey
+            bne  -
+            rts
+
+_longcopy
+            sta  P8ZP_SCRATCH_B1        ; lsb(count) = remainder in last page
+            tya
+            tax                         ; x = num pages (1+)
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  (P8ZP_SCRATCH_W2),y
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1
+            inc  P8ZP_SCRATCH_W2+1
+            dex
+            bne  -
+            ldy  P8ZP_SCRATCH_B1
+            bne  _copyshort
+            rts
+        }}
+    }
+
+    asmsub memset(uword mem @R0, uword numbytes @R1, ubyte value @A) clobbers(A,X,Y) {
+        %asm {{
+            ldy  cx16.r0
+            sty  P8ZP_SCRATCH_W1
+            ldy  cx16.r0+1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            ldy  cx16.r1+1
+            jmp  prog8_lib.memset
+        }}
+    }
+
+    asmsub memsetw(uword mem @R0, uword numwords @R1, uword value @AY) clobbers(A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            jmp  prog8_lib.memsetw
+        }}
+    }
+
+
+    inline asmsub rsave() {
+        ; save cpu status flag and all registers A, X, Y.
+        ; see http://6502.org/tutorials/register_preservation.html
+        %asm {{
+            php
+            sta  P8ZP_SCRATCH_REG
+            pha
+            txa
+            pha
+            tya
+            pha
+            lda  P8ZP_SCRATCH_REG
+        }}
+    }
+
+    inline asmsub rrestore() {
+        ; restore all registers and cpu status flag
+        %asm {{
+            pla
+            tay
+            pla
+            tax
+            pla
+            plp
+        }}
+    }
+
+    inline asmsub read_flags() -> ubyte @A {
+        %asm {{
+            php
+            pla
+        }}
+    }
+
+    inline asmsub clear_carry() {
+        %asm {{
+            clc
+        }}
+    }
+
+    inline asmsub set_carry() {
+        %asm {{
+            sec
+        }}
+    }
+
+    inline asmsub clear_irqd() {
+        %asm {{
+            cli
+        }}
+    }
+
+    inline asmsub set_irqd() {
+        %asm {{
+            sei
+        }}
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            jsr  c64.CLRCHN		; reset i/o channels
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            rts		; return to original caller
+        }}
+    }
+
+    inline asmsub progend() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_end
+            ldy  #>prog8_program_end
+        }}
+    }
+
+}
+
+cx16 {
+
+    ; the sixteen virtual 16-bit registers that the CX16 has defined in the zeropage
+    ; they are simulated on the C64 as well but their location in memory is different
+    ; (because there's no room for them in the zeropage)
+    ; they are allocated at the bottom of the eval-stack (should be ample space unless
+    ; you're doing insane nesting of expressions...)
+    &uword r0  = $cf00
+    &uword r1  = $cf02
+    &uword r2  = $cf04
+    &uword r3  = $cf06
+    &uword r4  = $cf08
+    &uword r5  = $cf0a
+    &uword r6  = $cf0c
+    &uword r7  = $cf0e
+    &uword r8  = $cf10
+    &uword r9  = $cf12
+    &uword r10 = $cf14
+    &uword r11 = $cf16
+    &uword r12 = $cf18
+    &uword r13 = $cf1a
+    &uword r14 = $cf1c
+    &uword r15 = $cf1e
 
 }

compiler/res/prog8lib/c64/textio.p8

@@ -16,7 +16,30 @@ const ubyte DEFAULT_HEIGHT = 25
 
 
 sub  clear_screen() {
-    clear_screenchars(' ')
+    txt.chrout(147)
+}
+
+sub home() {
+    txt.chrout(19)
+}
+
+sub nl() {
+    txt.chrout('\n')
+}
+
+sub spc() {
+    txt.chrout(' ')
+}
+
+asmsub column(ubyte col @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given column (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  c64.PLOT
+        tay
+        clc
+        jmp  c64.PLOT
+    }}
 }
 
 asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
@@ -38,13 +61,13 @@ asmsub  clear_screenchars (ubyte char @ A) clobbers(Y)  {
 	; ---- clear the character screen with the given fill character (leaves colors)
 	;      (assumes screen matrix is at the default address)
 	%asm {{
-		ldy  #0
-_loop		sta  c64.Screen,y
-		sta  c64.Screen+$0100,y
-		sta  c64.Screen+$0200,y
-		sta  c64.Screen+$02e8,y
-		iny
-		bne  _loop
+		ldy  #250
+-		sta  c64.Screen+250*0-1,y
+		sta  c64.Screen+250*1-1,y
+		sta  c64.Screen+250*2-1,y
+		sta  c64.Screen+250*3-1,y
+		dey
+		bne  -
 		rts
         }}
 }
@@ -53,13 +76,13 @@ asmsub  clear_screencolors (ubyte color @ A) clobbers(Y)  {
 	; ---- clear the character screen colors with the given color (leaves characters).
 	;      (assumes color matrix is at the default address)
 	%asm {{
-		ldy  #0
-_loop		sta  c64.Colors,y
-		sta  c64.Colors+$0100,y
-		sta  c64.Colors+$0200,y
-		sta  c64.Colors+$02e8,y
-		iny
-		bne  _loop
+		ldy  #250
+-		sta  c64.Colors+250*0-1,y
+		sta  c64.Colors+250*1-1,y
+		sta  c64.Colors+250*2-1,y
+		sta  c64.Colors+250*3-1,y
+		dey
+		bne  -
 		rts
         }}
 }
@@ -83,29 +106,31 @@ asmsub  scroll_left  (ubyte alsocolors @ Pc) clobbers(A, Y)  {
 
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
-		bcs  +
-		jmp  _scroll_screen
+		bcc _scroll_screen
 
-+               ; scroll the color memory
++               ; scroll the screen and the color memory
 		ldx  #0
 		ldy  #38
 -
-	.for row=0, row<=24, row+=1
-		lda  c64.Colors + 40*row + 1,x
-		sta  c64.Colors + 40*row,x
-	.next
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 1,x
+            sta  c64.Screen + 40*row + 0,x
+            lda  c64.Colors + 40*row + 1,x
+            sta  c64.Colors + 40*row + 0,x
+        .next
 		inx
 		dey
 		bpl  -
+		rts
 
-_scroll_screen  ; scroll the screen memory
+_scroll_screen  ; scroll only the screen memory
 		ldx  #0
 		ldy  #38
 -
-	.for row=0, row<=24, row+=1
-		lda  c64.Screen + 40*row + 1,x
-		sta  c64.Screen + 40*row,x
-	.next
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 1,x
+            sta  c64.Screen + 40*row + 0,x
+        .next
 		inx
 		dey
 		bpl  -
@@ -121,26 +146,28 @@ asmsub  scroll_right  (ubyte alsocolors @ Pc) clobbers(A)  {
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
-		bcs  +
-		jmp  _scroll_screen
+		bcc  _scroll_screen
 
-+               ; scroll the color memory
++               ; scroll the screen and the color memory
 		ldx  #38
 -
-	.for row=0, row<=24, row+=1
-		lda  c64.Colors + 40*row + 0,x
-		sta  c64.Colors + 40*row + 1,x
-	.next
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 0,x
+            sta  c64.Screen + 40*row + 1,x
+            lda  c64.Colors + 40*row + 0,x
+            sta  c64.Colors + 40*row + 1,x
+        .next
 		dex
 		bpl  -
+		rts
 
-_scroll_screen  ; scroll the screen memory
+_scroll_screen  ; scroll only the screen memory
 		ldx  #38
 -
-	.for row=0, row<=24, row+=1
-		lda  c64.Screen + 40*row + 0,x
-		sta  c64.Screen + 40*row + 1,x
-	.next
+        .for row=0, row<=24, row+=1
+            lda  c64.Screen + 40*row + 0,x
+            sta  c64.Screen + 40*row + 1,x
+        .next
 		dex
 		bpl  -
 
@@ -155,26 +182,28 @@ asmsub  scroll_up  (ubyte alsocolors @ Pc) clobbers(A)  {
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
-		bcs  +
-		jmp  _scroll_screen
+		bcc  _scroll_screen
 
-+               ; scroll the color memory
++               ; scroll the screen and the color memory
 		ldx #39
 -
-	.for row=1, row<=24, row+=1
-		lda  c64.Colors + 40*row,x
-		sta  c64.Colors + 40*(row-1),x
-	.next
+        .for row=1, row<=24, row+=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row-1),x
+            lda  c64.Colors + 40*row,x
+            sta  c64.Colors + 40*(row-1),x
+        .next
 		dex
 		bpl  -
+		rts
 
-_scroll_screen  ; scroll the screen memory
+_scroll_screen  ; scroll only the screen memory
 		ldx #39
 -
-	.for row=1, row<=24, row+=1
-		lda  c64.Screen + 40*row,x
-		sta  c64.Screen + 40*(row-1),x
-	.next
+        .for row=1, row<=24, row+=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row-1),x
+        .next
 		dex
 		bpl  -
 
@@ -189,26 +218,28 @@ asmsub  scroll_down  (ubyte alsocolors @ Pc) clobbers(A)  {
 	;      Carry flag determines if screen color data must be scrolled too
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
-		bcs  +
-		jmp  _scroll_screen
+		bcc  _scroll_screen
 
-+               ; scroll the color memory
++               ; scroll the screen and the color memory
 		ldx #39
 -
-	.for row=23, row>=0, row-=1
-		lda  c64.Colors + 40*row,x
-		sta  c64.Colors + 40*(row+1),x
-	.next
+        .for row=23, row>=0, row-=1
+            lda  c64.Colors + 40*row,x
+            sta  c64.Colors + 40*(row+1),x
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row+1),x
+        .next
 		dex
 		bpl  -
+		rts
 
-_scroll_screen  ; scroll the screen memory
+_scroll_screen  ; scroll only the screen memory
 		ldx #39
 -
-	.for row=23, row>=0, row-=1
-		lda  c64.Screen + 40*row,x
-		sta  c64.Screen + 40*(row+1),x
-	.next
+        .for row=23, row>=0, row-=1
+            lda  c64.Screen + 40*row,x
+            sta  c64.Screen + 40*(row+1),x
+        .next
 		dex
 		bpl  -
 
@@ -555,7 +586,7 @@ _colormod	sta  $ffff		; modified
 }
 
 asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernel routine, to save the X register.
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
 	%asm  {{
 		stx  P8ZP_SCRATCH_REG
 		tax

compiler/res/prog8lib/conv.p8

@@ -1,30 +1,518 @@
-; Prog8 definitions for number conversions routines.
+; Number conversions routines.
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
 
 
 conv {
 
-; ----- number conversions to decimal strings
+; ----- number conversions to decimal strings ----
 
-asmsub  ubyte2decimal  (ubyte value @ A) -> ubyte @ Y, ubyte @ A, ubyte @ X  {
-	; ---- A to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
-    %asm {{
-        ldy  #uword2decimal.ASCII_0_OFFSET
-        bne  uword2decimal.hex_try200
+    str  string_out = "????????????????"       ; result buffer for the string conversion routines
+
+asmsub  str_ub0  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in decimal string form, with left padding 0s (3 positions total)
+	%asm {{
+            phx
+            jsr  conv.ubyte2decimal
+            sty  string_out
+            sta  string_out+1
+            stx  string_out+2
+            lda  #0
+            sta  string_out+3
+            plx
+            rts
+	}}
+}
+
+asmsub  str_ub  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in decimal string form, without left padding 0s
+	%asm {{
+		phx
+		ldy  #0
+		sty  P8ZP_SCRATCH_B1
+		jsr  conv.ubyte2decimal
+_output_byte_digits
+                ; hundreds?
+		cpy  #'0'
+		beq  +
+		pha
+		tya
+		ldy  P8ZP_SCRATCH_B1
+		sta  string_out,y
+		pla
+		inc  P8ZP_SCRATCH_B1
+		; tens?
++		ldy  P8ZP_SCRATCH_B1
+                cmp  #'0'
+		beq  +
+		sta  string_out,y
+		iny
++               ; ones.
+                txa
+                sta  string_out,y
+                iny
+                lda  #0
+                sta  string_out,y
+                plx
+                rts
+	}}
+}
+
+asmsub  str_b  (byte value @ A) clobbers(A,Y)  {
+	; ---- convert the byte in A in decimal string form, without left padding 0s
+	%asm {{
+            phx
+            ldy  #0
+            sty  P8ZP_SCRATCH_B1
+            cmp  #0
+            bpl  +
+            pha
+            lda  #'-'
+            sta  string_out
+            inc  P8ZP_SCRATCH_B1
+            pla
++	    jsr  conv.byte2decimal
+            bra  str_ub._output_byte_digits
+	}}
+}
+
+asmsub  str_ubhex  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in hex string form
+	%asm {{
+            jsr  conv.ubyte2hex
+            sta  string_out
+            sty  string_out+1
+            lda  #0
+            sta  string_out+2
+            rts
+	}}
+}
+
+asmsub  str_ubbin  (ubyte value @ A) clobbers(A,Y)  {
+	; ---- convert the ubyte in A in binary string form
+	%asm {{
+	    sta  P8ZP_SCRATCH_B1
+	    ldy  #0
+	    sty  string_out+8
+	    ldy  #7
+-	    lsr  P8ZP_SCRATCH_B1
+            bcc  +
+            lda  #'1'
+            bne  _digit
++           lda  #'0'
+_digit      sta  string_out,y
+            dey
+	    bpl  -
+	    rts
+	}}
+}
+
+asmsub  str_uwbin  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in binary string form
+	%asm {{
+	    sta  P8ZP_SCRATCH_REG
+	    tya
+	    jsr  str_ubbin
+	    ldy  #0
+	    sty  string_out+16
+	    ldy  #7
+-	    lsr  P8ZP_SCRATCH_REG
+            bcc  +
+            lda  #'1'
+            bne  _digit
++           lda  #'0'
+_digit      sta  string_out+8,y
+            dey
+	    bpl  -
+	    rts
+	}}
+}
+
+asmsub  str_uwhex  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in hexadecimal string form (4 digits)
+	%asm {{
+            pha
+            tya
+            jsr  conv.ubyte2hex
+            sta  string_out
+            sty  string_out+1
+            pla
+            jsr  conv.ubyte2hex
+            sta  string_out+2
+            sty  string_out+3
+            lda  #0
+            sta  string_out+4
+            rts
+	}}
+}
+
+asmsub  str_uw0  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
+	%asm {{
+	    phx
+	    jsr  conv.uword2decimal
+	    ldy  #0
+-           lda  conv.uword2decimal.decTenThousands,y
+            sta  string_out,y
+            beq  +
+            iny
+            bne  -
++           plx
+	    rts
+	}}
+}
+
+asmsub  str_uw  (uword value @ AY) clobbers(A,Y)  {
+	; ---- convert the uword in A/Y in decimal string form, without left padding 0s
+	%asm {{
+	    phx
+	    jsr  conv.uword2decimal
+	    ldx  #0
+_output_digits
+	    ldy  #0
+-           lda  conv.uword2decimal.decTenThousands,y
+            beq  _allzero
+            cmp  #'0'
+            bne  _gotdigit
+            iny
+            bne  -
+_gotdigit   sta  string_out,x
+            inx
+            iny
+            lda  conv.uword2decimal.decTenThousands,y
+            bne  _gotdigit
+_end        lda  #0
+            sta  string_out,x
+            plx
+            rts
+
+_allzero    lda  #'0'
+            sta  string_out,x
+            inx
+            bne  _end
+	}}
+}
+
+asmsub  str_w  (word value @ AY) clobbers(A,Y)  {
+	; ---- convert the (signed) word in A/Y in decimal string form, without left padding 0's
+	%asm {{
+	    cpy  #0
+	    bpl  str_uw
+	    phx
+	    pha
+	    lda  #'-'
+	    sta  string_out
+            tya
+            eor  #255
+            tay
+            pla
+            eor  #255
+            clc
+            adc  #1
+            bcc  +
+            iny
++	    jsr  conv.uword2decimal
+	    ldx  #1
+	    bne  str_uw._output_digits
+	}}
+}
+
+
+; ---- string conversion to numbers -----
+
+asmsub  any2uword(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- parses a string into a 16 bit unsigned number. String may be in decimal, hex or binary format.
+	;    (the latter two require a $ or % prefix to be recognised)
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    returns amount of processed characters in A, and the parsed number will be in cx16.r15.
+	;    if the string was invalid, 0 will be returned in A.
+	%asm {{
+	pha
+	sta  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	ldy  #0
+	lda  (P8ZP_SCRATCH_W1),y
+	ldy  P8ZP_SCRATCH_W1+1
+	cmp  #'$'
+	beq  _hex
+	cmp  #'%'
+	beq  _bin
+	pla
+	jsr  str2uword
+	jmp  _result
+_hex	pla
+	jsr  hex2uword
+	jmp  _result
+_bin	pla
+	jsr  bin2uword
+_result
+        pha
+        lda  cx16.r15
+        sta  P8ZP_SCRATCH_B1        ; result value
+        pla
+        sta  cx16.r15
+        sty  cx16.r15+1
+        lda  P8ZP_SCRATCH_B1
         rts
 	}}
 }
 
-asmsub  uword2decimal  (uword value @ AY) -> ubyte @Y, ubyte @A, ubyte @X  {
-    ;  ---- convert 16 bit uword in A/Y to decimal
-    ;  output in uword2decimal.decTenThousands, decThousands, decHundreds, decTens, decOnes
-    ;  (these are terminated by a zero byte so they can be easily printed)
-    ;  also returns Y = 100's, A = 10's, X = 1's
+inline asmsub  str2ubyte(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- returns in A the unsigned byte value of the string number argument in AY
+	;    the number may NOT be preceded by a + sign and may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in A,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+		jsr  conv.str2uword
+	}}
+}
 
-    %asm {{
+inline asmsub  str2byte(str string @AY) clobbers(Y) -> ubyte @A {
+	; -- returns in A the signed byte value of the string number argument in AY
+	;    the number may be preceded by a + or - sign but may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in A,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+		jsr  conv.str2word
+	}}
+}
+
+asmsub  str2uword(str string @AY) -> uword @AY {
+	; -- returns the unsigned word value of the string number argument in AY
+	;    the number may NOT be preceded by a + sign and may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+_result = P8ZP_SCRATCH_W1
+        	sta  P8ZP_SCRATCH_W2
+        	sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		sty  _result
+		sty  _result+1
+		sty  cx16.r15+1
+_loop
+		lda  (P8ZP_SCRATCH_W2),y
+		sec
+		sbc  #48
+		bpl  _digit
+_done
+		sty  cx16.r15
+		lda  _result
+		ldy  _result+1
+		rts
+_digit
+		cmp  #10
+		bcs  _done
+		; add digit to result
+		pha
+		jsr  _result_times_10
+		pla
+		clc
+		adc  _result
+		sta  _result
+		bcc  +
+		inc  _result+1
++		iny
+		bne  _loop
+		; never reached
+
+_result_times_10     ; (W*4 + W)*2
+		lda  _result+1
+		sta  P8ZP_SCRATCH_REG
+		lda  _result
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		clc
+		adc  _result
+		sta  _result
+		lda  P8ZP_SCRATCH_REG
+		adc  _result+1
+		asl  _result
+		rol  a
+		sta  _result+1
+		rts
+	}}
+}
+
+asmsub  str2word(str string @AY) -> word @AY {
+	; -- returns the signed word value of the string number argument in AY
+	;    the number may be preceded by a + or - sign but may NOT contain spaces
+	;    (any non-digit character will terminate the number string that is parsed)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+_result = P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		sty  _result
+		sty  _result+1
+		sty  _negative
+		sty  cx16.r15+1
+		lda  (P8ZP_SCRATCH_W2),y
+		cmp  #'+'
+		bne  +
+		iny
++		cmp  #'-'
+		bne  _parse
+		inc  _negative
+		iny
+_parse		lda  (P8ZP_SCRATCH_W2),y
+		sec
+		sbc  #48
+		bpl  _digit
+_done
+		sty  cx16.r15
+		lda  _negative
+		beq  +
+		sec
+		lda  #0
+		sbc  _result
+		sta  _result
+		lda  #0
+		sbc  _result+1
+		sta  _result+1
++		lda  _result
+		ldy  _result+1
+		rts
+_digit
+		cmp  #10
+		bcs  _done
+		; add digit to result
+		pha
+		jsr  str2uword._result_times_10
+		pla
+		clc
+		adc  _result
+		sta  _result
+		bcc  +
+		inc  _result+1
++		iny
+		bne  _parse
+		; never reached
+_negative	.byte  0
+	}}
+}
+
+asmsub  hex2uword(str string @AY) -> uword @AY {
+	; -- hexadecimal string (with or without '$') to uword.
+	;    string may be in petscii or c64-screencode encoding.
+	;    stops parsing at the first character that's not a hex digit (except leading $)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+	sta  P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_W2+1
+	ldy  #0
+	sty  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	sty  cx16.r15+1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #'$'
+	bne  _loop
+	iny
+_loop
+	lda  #0
+	sta  P8ZP_SCRATCH_B1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #7                 ; screencode letters A-F are 1-6
+	bcc  _add_letter
+	cmp  #'g'
+	bcs  _stop
+	cmp  #'a'
+	bcs  _add_letter
+	cmp  #'0'
+	bcc  _stop
+	cmp  #'9'+1
+	bcs  _stop
+_calc
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	and  #$0f
+	clc
+	adc  P8ZP_SCRATCH_B1
+	ora  P8ZP_SCRATCH_W1
+	sta  P8ZP_SCRATCH_W1
+	iny
+	bne  _loop
+_stop
+	sty  cx16.r15
+	lda  P8ZP_SCRATCH_W1
+	ldy  P8ZP_SCRATCH_W1+1
+	rts
+_add_letter
+	pha
+	lda  #9
+	sta  P8ZP_SCRATCH_B1
+	pla
+	jmp  _calc
+	}}
+}
+
+asmsub  bin2uword(str string @AY) -> uword @AY {
+	; -- binary string (with or without '%') to uword.
+	;    stops parsing at the first character that's not a 0 or 1. (except leading %)
+	;    result in AY,  number of characters processed also remains in cx16.r15 if you want to use it!! (0 = error)
+	%asm {{
+	sta  P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_W2+1
+	ldy  #0
+	sty  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	sty  cx16.r15+1
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #'%'
+	bne  _loop
+	iny
+_loop
+	lda  (P8ZP_SCRATCH_W2),y
+	cmp  #'0'
+	bcc  _stop
+	cmp  #'2'
+	bcs  _stop
+_first  asl  P8ZP_SCRATCH_W1
+	rol  P8ZP_SCRATCH_W1+1
+	and  #1
+	ora  P8ZP_SCRATCH_W1
+	sta  P8ZP_SCRATCH_W1
+	iny
+	bne  _loop
+_stop
+	sty  cx16.r15
+	lda  P8ZP_SCRATCH_W1
+	ldy  P8ZP_SCRATCH_W1+1
+	rts
+	}}
+}
+
+
+; ----- low level number conversions to decimal strings ----
+
+asmsub  ubyte2decimal  (ubyte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
+	; ---- A to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
+	%asm {{
+		ldy  #uword2decimal.ASCII_0_OFFSET
+		bne  uword2decimal.hex_try200
+		rts
+	}}
+}
+
+asmsub  uword2decimal  (uword value @AY) -> ubyte @Y, ubyte @A, ubyte @X  {
+	;  ---- convert 16 bit uword in A/Y to decimal
+	;  output in uword2decimal.decTenThousands, decThousands, decHundreds, decTens, decOnes
+	;  (these are terminated by a zero byte so they can be easily printed)
+	;  also returns Y = 100's, A = 10's, X = 1's
+
+	%asm {{
 
 ;Convert 16 bit Hex to Decimal (0-65535) Rev 2
 ;By Omegamatrix    Further optimizations by tepples
@@ -193,11 +681,7 @@ decOnes   		.byte  0
     }}
 }
 
-
-; ----- utility functions ----
-
-
-asmsub  byte2decimal  (byte value @ A) -> ubyte @ Y, ubyte @ A, ubyte @ X  {
+asmsub  byte2decimal  (byte value @A) -> ubyte @Y, ubyte @A, ubyte @X  {
 	; ---- A (signed byte) to decimal string in Y/A/X  (100s in Y, 10s in A, 1s in X)
 	;      note: if the number is negative, you have to deal with the '-' yourself!
 	%asm {{
@@ -210,7 +694,7 @@ asmsub  byte2decimal  (byte value @ A) -> ubyte @ Y, ubyte @ A, ubyte @ X  {
 	}}
 }
 
-asmsub  ubyte2hex  (ubyte value @ A) -> ubyte @ A, ubyte @ Y  {
+asmsub  ubyte2hex  (ubyte value @A) -> ubyte @A, ubyte @Y  {
 	; ---- A to hex petscii string in AY (first hex char in A, second hex char in Y)
 	%asm {{
 		stx  P8ZP_SCRATCH_REG
@@ -232,7 +716,7 @@ _hex_digits	.text "0123456789abcdef"	; can probably be reused for other stuff as
 	}}
 }
 
-asmsub  uword2hex  (uword value @ AY) clobbers(A,Y)  {
+asmsub  uword2hex  (uword value @AY) clobbers(A,Y)  {
 	; ---- convert 16 bit uword in A/Y into 4-character hexadecimal string 'uword2hex.output' (0-terminated)
 	%asm {{
 		sta  P8ZP_SCRATCH_REG
@@ -245,209 +729,8 @@ asmsub  uword2hex  (uword value @ AY) clobbers(A,Y)  {
 		sta  output+2
 		sty  output+3
 		rts
-output	.text  "0000", $00      ; 0-terminated output buffer (to make printing easier)
+output		.text  "0000", $00      ; 0-terminated output buffer (to make printing easier)
 	}}
 }
 
-
-asmsub  str2ubyte(str string @ AY) clobbers(Y) -> ubyte @A {
-	; -- returns the unsigned byte value of the string number argument in AY
-	;    the number may NOT be preceded by a + sign and may NOT contain spaces
-	;    (any non-digit character will terminate the number string that is parsed)
-	%asm {{
-    	jmp  str2uword
-	}}
-}
-
-asmsub  str2byte(str string @ AY) clobbers(Y) -> ubyte @A {
-	; -- returns the signed byte value of the string number argument in AY
-	;    the number may be preceded by a + or - sign but may NOT contain spaces
-	;    (any non-digit character will terminate the number string that is parsed)
-	%asm {{
-    	jmp  str2word
-	}}
-}
-
-asmsub  str2uword(str string @ AY) -> uword @ AY {
-	; -- returns the unsigned word value of the string number argument in AY
-	;    the number may NOT be preceded by a + sign and may NOT contain spaces
-	;    (any non-digit character will terminate the number string that is parsed)
-	%asm {{
-_result = P8ZP_SCRATCH_W2
-		sta  _mod+1
-		sty  _mod+2
-		ldy  #0
-		sty  _result
-		sty  _result+1
-_mod		lda  $ffff,y		; modified
-		sec
-		sbc  #48
-		bpl  +
-_done		; return result
-		lda  _result
-		ldy  _result+1
-		rts
-+		cmp  #10
-		bcs  _done
-		; add digit to result
-		pha
-		jsr  _result_times_10
-		pla
-		clc
-		adc  _result
-		sta  _result
-		bcc  +
-		inc  _result+1
-+		iny
-		bne  _mod
-		; never reached
-
-_result_times_10     ; (W*4 + W)*2
-		lda  _result+1
-		sta  P8ZP_SCRATCH_REG
-		lda  _result
-		asl  a
-		rol  P8ZP_SCRATCH_REG
-		asl  a
-		rol  P8ZP_SCRATCH_REG
-		clc
-		adc  _result
-		sta  _result
-		lda  P8ZP_SCRATCH_REG
-		adc  _result+1
-		asl  _result
-		rol  a
-		sta  _result+1
-		rts
-	}}
-}
-
-asmsub  str2word(str string @ AY) -> word @ AY {
-	; -- returns the signed word value of the string number argument in AY
-	;    the number may be preceded by a + or - sign but may NOT contain spaces
-	;    (any non-digit character will terminate the number string that is parsed)
-	%asm {{
-_result = P8ZP_SCRATCH_W2
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		ldy  #0
-		sty  _result
-		sty  _result+1
-		sty  _negative
-		lda  (P8ZP_SCRATCH_W1),y
-		cmp  #'+'
-		bne  +
-		iny
-+		cmp  #'-'
-		bne  _parse
-		inc  _negative
-		iny
-_parse		lda  (P8ZP_SCRATCH_W1),y
-		sec
-		sbc  #48
-		bpl  _digit
-_done		; return result
-		lda  _negative
-		beq  +
-		sec
-		lda  #0
-		sbc  _result
-		sta  _result
-		lda  #0
-		sbc  _result+1
-		sta  _result+1
-+		lda  _result
-		ldy  _result+1
-		rts
-_digit		cmp  #10
-		bcs  _done
-		; add digit to result
-		pha
-		jsr  str2uword._result_times_10
-		pla
-		clc
-		adc  _result
-		sta  _result
-		bcc  +
-		inc  _result+1
-+		iny
-		bne  _parse
-		; never reached
-_negative	.byte  0
-	}}
-}
-
-asmsub  hex2uword(str string @ AY) -> uword @AY {
-    ; -- hexadecimal string with or without '$' to uword.
-    ;    string may be in petscii or c64-screencode encoding.
-    %asm {{
-        sta  P8ZP_SCRATCH_W2
-        sty  P8ZP_SCRATCH_W2+1
-        ldy  #0
-        sty  P8ZP_SCRATCH_W1
-        sty  P8ZP_SCRATCH_W1+1
-_loop       ldy  #0
-        sty  P8ZP_SCRATCH_B1
-        lda  (P8ZP_SCRATCH_W2),y
-        beq  _stop
-        cmp  #'$'
-        beq  _skip
-        cmp  #7
-        bcc  _add_nine
-        cmp  #'9'
-        beq  _calc
-        bcs  _add_nine
-_calc       asl  P8ZP_SCRATCH_W1
-        rol  P8ZP_SCRATCH_W1+1
-        asl  P8ZP_SCRATCH_W1
-        rol  P8ZP_SCRATCH_W1+1
-        asl  P8ZP_SCRATCH_W1
-        rol  P8ZP_SCRATCH_W1+1
-        asl  P8ZP_SCRATCH_W1
-        rol  P8ZP_SCRATCH_W1+1
-        and  #$0f
-        clc
-        adc  P8ZP_SCRATCH_B1
-        ora  P8ZP_SCRATCH_W1
-        sta  P8ZP_SCRATCH_W1
-_skip       inc  P8ZP_SCRATCH_W2
-        bne  _loop
-        inc  P8ZP_SCRATCH_W2+1
-        bne  _loop
-_stop       lda  P8ZP_SCRATCH_W1
-        ldy  P8ZP_SCRATCH_W1+1
-        rts
-_add_nine   ldy  #9
-        sty  P8ZP_SCRATCH_B1
-        bne  _calc
-    }}
-}
-
-asmsub  bin2uword(str string @ AY) -> uword @AY {
-    ; -- binary string with or without '%' to uword.
-    %asm {{
-        sta  P8ZP_SCRATCH_W2
-        sty  P8ZP_SCRATCH_W2+1
-        ldy  #0
-        sty  P8ZP_SCRATCH_W1
-        sty  P8ZP_SCRATCH_W1+1
-_loop       lda  (P8ZP_SCRATCH_W2),y
-        beq  _stop
-        cmp  #'%'
-        beq  +
-        asl  P8ZP_SCRATCH_W1
-        rol  P8ZP_SCRATCH_W1+1
-        and  #1
-        ora  P8ZP_SCRATCH_W1
-        sta  P8ZP_SCRATCH_W1
-+           inc  P8ZP_SCRATCH_W2
-        bne  _loop
-        inc  P8ZP_SCRATCH_W2+1
-        bne  _loop
-_stop       lda  P8ZP_SCRATCH_W1
-        ldy  P8ZP_SCRATCH_W1+1
-        rts
-    }}
-}
-
 }

compiler/res/prog8lib/cx16/floats.p8

@@ -64,6 +64,7 @@ romsub $fe7e = FINLOG(byte value @A) clobbers (A, X, Y)           ; fac1 += sign
 romsub $fe81 = FOUT() clobbers(X) -> uword @ AY             ; fac1 -> string, address returned in AY
 romsub $fe8a = SQR() clobbers(A,X,Y)                        ; fac1 = SQRT(fac1)
 romsub $fe8d = FPWRT() clobbers(A,X,Y)                      ; fac1 = fac2 ** fac1
+; note: there is no FPWR() on the Cx16
 romsub $fe93 = NEGOP() clobbers(A)                          ; switch the sign of fac1
 romsub $fe96 = EXP() clobbers(A,X,Y)                        ; fac1 = EXP(fac1)  (e ** fac1)
 romsub $fe9f = RND2(byte value @A) clobbers(A,X,Y)                ; fac1 = RND(A) float random number generator
@@ -78,10 +79,10 @@ asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- unsigned 16 bit word in A/Y (lo/hi) to fac1
 	%asm {{
         phx
-        sta  P8ZP_SCRATCH_W2
+        sta  _tmp
         sty  P8ZP_SCRATCH_B1
         tya
-        ldy  P8ZP_SCRATCH_W2
+        ldy  _tmp
         jsr  GIVAYF                 ; load it as signed... correct afterwards
         lda  P8ZP_SCRATCH_B1
 	    bpl  +
@@ -90,6 +91,7 @@ asmsub  GIVUAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	    jsr  FADD
 +	    plx
         rts
+_tmp        .byte 0
 _flt65536    .byte 145,0,0,0,0       ; 65536.0
 	}}
 }
@@ -98,9 +100,9 @@ _flt65536    .byte 145,0,0,0,0       ; 65536.0
 asmsub  GIVAYFAY  (uword value @ AY) clobbers(A,X,Y)  {
 	; ---- signed 16 bit word in A/Y (lo/hi) to float in fac1
 	%asm {{
-		sta  P8ZP_SCRATCH_W2
+		sta  P8ZP_SCRATCH_B1
 		tya
-		ldy  P8ZP_SCRATCH_W2
+		ldy  P8ZP_SCRATCH_B1
 		jmp  GIVAYF		; this uses the inverse order, Y/A
 	}}
 }
@@ -109,9 +111,9 @@ asmsub  FTOSWRDAY  () clobbers(X) -> uword @ AY  {
 	; ---- fac1 to signed word in A/Y
 	%asm {{
 		jsr  FTOSWORDYA	; note the inverse Y/A order
-		sta  P8ZP_SCRATCH_REG
+		sta  P8ZP_SCRATCH_B1
 		tya
-		ldy  P8ZP_SCRATCH_REG
+		ldy  P8ZP_SCRATCH_B1
 		rts
 	}}
 }
@@ -127,6 +129,14 @@ asmsub  GETADRAY  () clobbers(X) -> uword @ AY  {
 	}}
 }
 
+asmsub  FREADUY (ubyte value @Y) {
+    ; -- 8 bit unsigned Y -> float in fac1
+    %asm {{
+        lda  #0
+        jmp  GIVAYF
+    }}
+}
+
 sub  print_f  (float value) {
 	; ---- prints the floating point value (without a newline).
 	%asm {{

compiler/res/prog8lib/cx16/gfx2.p8

@@ -0,0 +1,974 @@
+%target cx16
+
+; Bitmap pixel graphics routines for the CommanderX16
+; Custom routines to use the full-screen 640x480 and 320x240 screen modes.
+; (These modes are not supported by the documented GRAPH_xxxx kernal routines)
+;
+; No text layer is currently shown, text can be drawn as part of the bitmap itself.
+; Note: for similar graphics routines that also work on the C-64, use the "graphics" module instead.
+; Note: for color palette manipulation, use the "palette" module or write Vera registers yourself.
+; Note: this library implements code for various resolutions and color depths. This takes up memory.
+;       If you're memory constrained you should probably not use this built-in library,
+;       but make a copy in your project only containing the code for the required resolution.
+;
+;
+; SCREEN MODE LIST:
+;   mode 0 = reset back to default text mode
+;   mode 1 = bitmap 320 x 240 monochrome
+;   mode 2 = bitmap 320 x 240 x 4c (TODO not yet implemented)
+;   mode 3 = bitmap 320 x 240 x 16c (TODO not yet implemented)
+;   mode 4 = bitmap 320 x 240 x 256c
+;   mode 5 = bitmap 640 x 480 monochrome
+;   mode 6 = bitmap 640 x 480 x 4c
+;   higher color dephts in highres are not supported due to lack of VRAM
+
+; TODO can we make a FB vector table and emulation routines for the Cx16s' GRAPH_init() call? to replace the builtin 320x200 fb driver?
+
+gfx2 {
+
+    ; read-only control variables:
+    ubyte active_mode = 0
+    uword width = 0
+    uword height = 0
+    ubyte bpp = 0
+    ubyte monochrome_dont_stipple_flag = false            ; set to false to enable stippling mode in monochrome displaymodes
+
+    sub screen_mode(ubyte mode) {
+        when mode {
+            1 -> {
+                ; lores monochrome
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 64
+                cx16.VERA_DC_VSCALE = 64
+                cx16.VERA_L1_CONFIG = %00000100
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = 0
+                width = 320
+                height = 240
+                bpp = 1
+            }
+            ; TODO modes 2, 3 not yet implemented
+            4 -> {
+                ; lores 256c
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 64
+                cx16.VERA_DC_VSCALE = 64
+                cx16.VERA_L1_CONFIG = %00000111
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = 0
+                width = 320
+                height = 240
+                bpp = 8
+            }
+            5 -> {
+                ; highres monochrome
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 128
+                cx16.VERA_DC_VSCALE = 128
+                cx16.VERA_L1_CONFIG = %00000100
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = %00000001
+                width = 640
+                height = 480
+                bpp = 1
+            }
+            6 -> {
+                ; highres 4c
+                cx16.VERA_DC_VIDEO = (cx16.VERA_DC_VIDEO & %11001111) | %00100000      ; enable only layer 1
+                cx16.VERA_DC_HSCALE = 128
+                cx16.VERA_DC_VSCALE = 128
+                cx16.VERA_L1_CONFIG = %00000101
+                cx16.VERA_L1_MAPBASE = 0
+                cx16.VERA_L1_TILEBASE = %00000001
+                width = 640
+                height = 480
+                bpp = 2
+            }
+            else -> {
+                ; back to default text mode and colors
+                cx16.VERA_CTRL = %10000000      ; reset VERA and palette
+                c64.CINT()      ; back to text mode
+                width = 0
+                height = 0
+                bpp = 0
+                mode = 0
+            }
+        }
+
+        active_mode = mode
+        if bpp
+            clear_screen()
+    }
+
+    sub clear_screen() {
+        monochrome_stipple(false)
+        position(0, 0)
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                repeat 240/2/8
+                    cs_innerloop640()
+            }
+            ; TODO mode 2, 3
+            4 -> {
+                ; lores 256c
+                repeat 240/2
+                    cs_innerloop640()
+            }
+            5 -> {
+                ; highres monochrome
+                repeat 480/8
+                    cs_innerloop640()
+            }
+            6 -> {
+                ; highres 4c
+                repeat 480/4
+                    cs_innerloop640()
+            }
+            ; modes 7 and 8 not supported due to lack of VRAM
+        }
+        position(0, 0)
+    }
+
+    sub monochrome_stipple(ubyte enable) {
+        monochrome_dont_stipple_flag = not enable
+    }
+
+    sub rect(uword x, uword y, uword width, uword height, ubyte color) {
+        if width==0 or height==0
+            return
+        horizontal_line(x, y, width, color)
+        if height==1
+            return
+        horizontal_line(x, y+height-1, width, color)
+        vertical_line(x, y+1, height-2, color)
+        if width==1
+            return
+        vertical_line(x+width-1, y+1, height-2, color)
+    }
+
+    sub fillrect(uword x, uword y, uword width, uword height, ubyte color) {
+        if width==0
+            return
+        repeat height {
+            horizontal_line(x, y, width, color)
+            y++
+        }
+    }
+
+    sub horizontal_line(uword x, uword y, uword length, ubyte color) {
+        if length==0
+            return
+        when active_mode {
+            1, 5 -> {
+                ; monochrome modes, either resolution
+                ubyte separate_pixels = (8-lsb(x)) & 7
+                if separate_pixels as uword > length
+                    separate_pixels = lsb(length)
+                repeat separate_pixels {
+                    ; TODO optimize this by writing a masked byte in 1 go
+                    plot(x, y, color)
+                    x++
+                }
+                length -= separate_pixels
+                if length {
+                    position(x, y)
+                    separate_pixels = lsb(length) & 7
+                    x += length & $fff8
+                    %asm {{
+                        lsr  length+1
+                        ror  length
+                        lsr  length+1
+                        ror  length
+                        lsr  length+1
+                        ror  length
+                        lda  color
+                        bne  +
+                        ldy  #0     ; black
+                        bra  _loop
++                       lda  monochrome_dont_stipple_flag
+                        beq  _stipple
+                        ldy  #255       ; don't stipple
+                        bra  _loop
+_stipple                lda  y
+                        and  #1         ; determine stipple pattern to use
+                        bne  +
+                        ldy  #%01010101
+                        bra  _loop
++                       ldy  #%10101010
+_loop                   lda  length
+                        ora  length+1
+                        beq  _done
+                        sty  cx16.VERA_DATA0
+                        lda  length
+                        bne  +
+                        dec  length+1
++                       dec  length
+                        bra  _loop
+_done
+                    }}
+                    repeat separate_pixels {
+                        ; TODO optimize this by writing a masked byte in 1 go
+                        plot(x, y, color)
+                        x++
+                    }
+                }
+                cx16.VERA_ADDR_H &= %00000111   ; vera auto-increment off again
+            }
+            4 -> {
+                ; lores 256c
+                position(x, y)
+                %asm {{
+                    lda  color
+                    phx
+                    ldx  length+1
+                    beq  +
+                    ldy  #0
+-                   sta  cx16.VERA_DATA0
+                    iny
+                    bne  -
+                    dex
+                    bne  -
++                   ldy  length     ; remaining
+                    beq  +
+-                   sta  cx16.VERA_DATA0
+                    dey
+                    bne  -
++                   plx
+                }}
+            }
+            6 -> {
+                ; highres 4c
+                ; TODO also mostly usable for lores 4c?
+                color &= 3
+                ubyte[4] colorbits
+                ubyte ii
+                for ii in 3 downto 0 {
+                    colorbits[ii] = color
+                    color <<= 2
+                }
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                %asm {{
+                    lda  cx16.VERA_ADDR_H
+                    and  #%00000111         ; no auto advance
+                    sta  cx16.VERA_ADDR_H
+                    stz  cx16.VERA_CTRL     ; setup vera addr 0
+                    lda  cx16.r1
+                    and  #1
+                    sta  cx16.VERA_ADDR_H
+                    lda  cx16.r0
+                    sta  cx16.VERA_ADDR_L
+                    lda  cx16.r0+1
+                    sta  cx16.VERA_ADDR_M
+                    phx
+                    ldx  x
+                }}
+
+                repeat length {
+                    %asm {{
+                        txa
+                        and  #3
+                        tay
+                        lda  cx16.VERA_DATA0
+                        and  gfx2.plot.mask4c,y
+                        ora  colorbits,y
+                        sta  cx16.VERA_DATA0
+                        cpy  #%00000011         ; next vera byte?
+                        bne  +
+                        inc  cx16.VERA_ADDR_L
+                        bne  +
+                        inc  cx16.VERA_ADDR_M
++                       bne  +
+                        inc  cx16.VERA_ADDR_H
++                       inx                     ; next pixel
+                    }}
+                }
+
+                %asm {{
+                    plx
+                }}
+            }
+        }
+    }
+
+    sub vertical_line(uword x, uword y, uword height, ubyte color) {
+        position(x,y)
+        when active_mode {
+            1, 5 -> {
+                ; monochrome, either resolution
+                ; note for the 1 bpp modes we can't use vera's auto increment mode because we have to 'or' the pixel data in place.
+                ; TODO use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment IS possible
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                cx16.r15 = gfx2.plot.bits[x as ubyte & 7]           ; bitmask
+                if active_mode>=5
+                    cx16.r14 = 640/8
+                else
+                    cx16.r14 = 320/8
+                if color {
+                    if monochrome_dont_stipple_flag {
+                        repeat height {
+                            %asm {{
+                                lda  cx16.VERA_DATA0
+                                ora  cx16.r15
+                                sta  cx16.VERA_DATA0
+                                lda  cx16.VERA_ADDR_L
+                                clc
+                                adc  cx16.r14                 ; advance vera ptr to go to the next line
+                                sta  cx16.VERA_ADDR_L
+                                lda  cx16.VERA_ADDR_M
+                                adc  #0
+                                sta  cx16.VERA_ADDR_M
+                                ; lda  cx16.VERA_ADDR_H     ; the bitmap size is small enough to not have to deal with the _H part.
+                                ; adc  #0
+                                ; sta  cx16.VERA_ADDR_H
+                            }}
+                        }
+                    } else {
+                        ; stippling.
+                        height = (height+1)/2       ; TODO is the line sometimes 1 pixel too long now because of rounding?
+                        %asm {{
+                            lda x
+                            eor y
+                            and #1
+                            bne +
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14                ; advance vera ptr to go to the next line for correct stipple pattern
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+        +
+                            asl  cx16.r14
+                            ldy  height
+                            beq  +
+        -                   lda  cx16.VERA_DATA0
+                            ora  cx16.r15
+                            sta  cx16.VERA_DATA0
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14               ; advance vera data ptr to go to the next-next line
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+                            ; lda  cx16.VERA_ADDR_H      ; the bitmap size is small enough to not have to deal with the _H part.
+                            ; adc  #0
+                            ; sta  cx16.VERA_ADDR_H
+                            dey
+                            bne  -
+        +
+                        }}
+                    }
+                } else {
+                    cx16.r15 = ~cx16.r15
+                    repeat height {
+                        %asm {{
+                            lda  cx16.VERA_DATA0
+                            and  cx16.r15
+                            sta  cx16.VERA_DATA0
+                            lda  cx16.VERA_ADDR_L
+                            clc
+                            adc  cx16.r14             ; advance vera data ptr to go to the next line
+                            sta  cx16.VERA_ADDR_L
+                            lda  cx16.VERA_ADDR_M
+                            adc  #0
+                            sta  cx16.VERA_ADDR_M
+                            ; the bitmap size is small enough to not have to deal with the _H part:
+                            ; lda  cx16.VERA_ADDR_H
+                            ; adc  #0
+                            ; sta  cx16.VERA_ADDR_H
+                        }}
+                    }
+                }
+            }
+            4 -> {
+                ; lores 256c
+                ; set vera auto-increment to 320 pixel increment (=next line)
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | (14<<4)
+                %asm {{
+                    ldy  height
+                    beq  +
+                    lda  color
+-                   sta  cx16.VERA_DATA0
+                    dey
+                    bne  -
++
+                }}
+            }
+            6 -> {
+                ; highres 4c
+                ; note for this mode we can't use vera's auto increment mode because we have to 'or' the pixel data in place.
+                ; TODO use TWO vera adress pointers simultaneously one for reading, one for writing, so auto-increment IS possible
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                ; TODO also mostly usable for lores 4c?
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+
+                ; TODO optimize this vertical line loop in pure assembly
+                color &= 3
+                color <<= gfx2.plot.shift4c[lsb(x) & 3]
+                ubyte mask = gfx2.plot.mask4c[lsb(x) & 3]
+                repeat height {
+                    ubyte value = cx16.vpeek(lsb(cx16.r1), cx16.r0) & mask | color
+                    cx16.vpoke(lsb(cx16.r1), cx16.r0, value)
+                    %asm {{
+                        ; 24 bits add 160 (640/4)
+                        clc
+                        lda  cx16.r0
+                        adc  #640/4
+                        sta  cx16.r0
+                        lda  cx16.r0+1
+                        adc  #0
+                        sta  cx16.r0+1
+                        bcc  +
+                        inc  cx16.r1
++
+                    }}
+                }
+            }
+        }
+
+    }
+
+    sub line(uword @zp x1, uword @zp y1, uword @zp x2, uword @zp y2, ubyte color) {
+        ; Bresenham algorithm.
+        ; This code special-cases various quadrant loops to allow simple ++ and -- operations.
+        if y1>y2 {
+            ; make sure dy is always positive to have only 4 instead of 8 special cases
+            swap(x1, x2)
+            swap(y1, y2)
+        }
+        word @zp dx = (x2 as word)-x1
+        word @zp dy = (y2 as word)-y1
+
+        if dx==0 {
+            vertical_line(x1, y1, abs(dy) as uword +1, color)
+            return
+        }
+        if dy==0 {
+            if x1>x2
+                x1=x2
+            horizontal_line(x1, y1, abs(dx) as uword +1, color)
+            return
+        }
+
+        word @zp d = 0
+        cx16.r13 = true      ; 'positive_ix'
+        if dx < 0 {
+            dx = -dx
+            cx16.r13 = false
+        }
+        word @zp dx2 = dx*2
+        word @zp dy2 = dy*2
+        cx16.r14 = x1       ; internal plot X
+
+        if dx >= dy {
+            if cx16.r13 {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14++
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if cx16.r14==x2
+                        return
+                    cx16.r14--
+                    d += dy2
+                    if d > dx {
+                        y1++
+                        d -= dx2
+                    }
+                }
+            }
+        }
+        else {
+            if cx16.r13 {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14++
+                        d -= dy2
+                    }
+                }
+            } else {
+                repeat {
+                    plot(cx16.r14, y1, color)
+                    if y1 == y2
+                        return
+                    y1++
+                    d += dx2
+                    if d > dy {
+                        cx16.r14--
+                        d -= dy2
+                    }
+                }
+            }
+        }
+    }
+
+    sub circle(uword @zp xcenter, uword @zp ycenter, ubyte radius, ubyte color) {
+        ; Midpoint algorithm.
+        if radius==0
+            return
+
+        ubyte @zp xx = radius
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-xx
+        ; R14 = internal plot X
+        ; R15 = internal plot Y
+
+        while xx>=yy {
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter + yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + xx
+            cx16.r15 = ycenter - yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter + xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - yy
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter + yy
+            cx16.r15 = ycenter - xx
+            plot(cx16.r14, cx16.r15, color)
+            cx16.r14 = xcenter - yy
+            plot(cx16.r14, cx16.r15, color)
+
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                xx--
+                decisionOver2 += (yy as word -xx)*2+1
+            }
+        }
+    }
+
+    sub disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
+        ; Midpoint algorithm, filled
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word @zp decisionOver2 = (1 as word)-radius
+
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
+            yy++
+            if decisionOver2<=0
+                decisionOver2 += (yy as word)*2+1
+            else {
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
+            }
+        }
+    }
+
+    sub plot(uword @zp x, uword y, ubyte color) {
+        ubyte[8] bits = [128, 64, 32, 16, 8, 4, 2, 1]
+        ubyte[4] mask4c = [%00111111, %11001111, %11110011, %11111100]
+        ubyte[4] shift4c = [6,4,2,0]
+        uword addr
+        ubyte value
+
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    addr = x/8 + y*(320/8)
+                    value = bits[lsb(x)&7]
+                    if color
+                        cx16.vpoke_or(0, addr, value)
+                    else {
+                        value = ~value
+                        cx16.vpoke_and(0, addr, value)
+                    }
+                }
+            }
+            4 -> {
+                ; lores 256c
+                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                cx16.vpoke(lsb(cx16.r1), cx16.r0, color)
+                ; activate vera auto-increment mode so next_pixel() can be used after this
+                cx16.VERA_ADDR_H = cx16.VERA_ADDR_H & %00000111 | %00010000
+                color = cx16.VERA_DATA0
+            }
+            5 -> {
+                ; highres monochrome
+                %asm {{
+                    lda  x
+                    eor  y
+                    ora  monochrome_dont_stipple_flag
+                    and  #1
+                }}
+                if_nz {
+                    addr = x/8 + y*(640/8)
+                    value = bits[lsb(x)&7]
+                    if color
+                        cx16.vpoke_or(0, addr, value)
+                    else {
+                        value = ~value
+                        cx16.vpoke_and(0, addr, value)
+                    }
+                }
+            }
+            6 -> {
+                ; highres 4c
+                ; TODO also mostly usable for lores 4c?
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                color &= 3
+                color <<= shift4c[lsb(x) & 3]
+                ; TODO optimize the vera memory manipulation in pure assembly
+                cx16.VERA_ADDR_H &= %00000111   ; no auto advance
+                value = cx16.vpeek(lsb(cx16.r1), cx16.r0) & mask4c[lsb(x) & 3] | color
+                cx16.vpoke(lsb(cx16.r1), cx16.r0, value)
+            }
+        }
+    }
+
+    sub position(uword @zp x, uword y) {
+        ubyte bank
+        when active_mode {
+            1 -> {
+                ; lores monochrome
+                cx16.r0 = y*(320/8) + x/8
+                cx16.vaddr(0, cx16.r0, 0, 1)
+            }
+            ; TODO modes 2,3
+            4 -> {
+                ; lores 256c
+                void addr_mul_24_for_lores_256c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                bank = lsb(cx16.r1)
+                cx16.vaddr(bank, cx16.r0, 0, 1)
+            }
+            5 -> {
+                ; highres monochrome
+                cx16.r0 = y*(640/8) + x/8
+                cx16.vaddr(0, cx16.r0, 0, 1)
+            }
+            6 -> {
+                ; highres 4c
+                void addr_mul_24_for_highres_4c(y, x)      ; 24 bits result is in r0 and r1L (highest byte)
+                bank = lsb(cx16.r1)
+                cx16.vaddr(bank, cx16.r0, 0, 1)
+            }
+        }
+    }
+
+    inline asmsub next_pixel(ubyte color @A) {
+        ; -- sets the next pixel byte to the graphics chip.
+        ;    for 8 bpp screens this will plot 1 pixel.
+        ;    for 1 bpp screens it will plot 8 pixels at once (color = bit pattern).
+        ;    for 2 bpp screens it will plot 4 pixels at once (color = bit pattern).
+        %asm {{
+            sta  cx16.VERA_DATA0
+        }}
+    }
+
+    asmsub next_pixels(uword pixels @AY, uword amount @R0) clobbers(A, Y)  {
+        ; -- sets the next bunch of pixels from a prepared array of bytes.
+        ;    for 8 bpp screens this will plot 1 pixel per byte.
+        ;    for 1 bpp screens it will plot 8 pixels at once (colors are the bit patterns per byte).
+        ;    for 2 bpp screens it will plot 4 pixels at once (colors are the bit patterns per byte).
+        %asm {{
+            phx
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r0+1
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            bne  -
+            inc  P8ZP_SCRATCH_W1+1       ; next page of 256 pixels
+            dex
+            bne  -
+
++           ldx  cx16.r0           ; remaining pixels
+            beq  +
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            dex
+            bne  -
++           plx
+        }}
+    }
+
+    asmsub set_8_pixels_from_bits(ubyte bits @R0, ubyte oncolor @A, ubyte offcolor @Y) {
+        ; this is only useful in 256 color mode where one pixel equals one byte value.
+        %asm {{
+            phx
+            ldx  #8
+-           asl  cx16.r0
+            bcc  +
+            sta  cx16.VERA_DATA0
+            bra  ++
++           sty  cx16.VERA_DATA0
++           dex
+            bne  -
+            plx
+            rts
+        }}
+    }
+
+    const ubyte charset_orig_bank = $0
+    const uword charset_orig_addr = $f800        ; in bank 0, so $0f800
+    const ubyte charset_bank = $1
+    const uword charset_addr = $f000       ; in bank 1, so $1f000
+
+    sub text_charset(ubyte charset) {
+        ; -- make a copy of the selected character set to use with text()
+        ;    the charset number is the same as for the cx16.screen_set_charset() ROM function.
+        ;    1 = ISO charset, 2 = PETSCII uppercase+graphs, 3= PETSCII uppercase+lowercase.
+        cx16.screen_set_charset(charset, 0)
+        cx16.vaddr(charset_orig_bank, charset_orig_addr, 0, 1)
+        cx16.vaddr(charset_bank, charset_addr, 1, 1)
+        repeat 256*8 {
+            cx16.VERA_DATA1 = cx16.VERA_DATA0
+        }
+    }
+
+    sub text(uword @zp x, uword y, ubyte color, uword sctextptr) {
+        ; -- Write some text at the given pixel position. The text string must be in screencode encoding (not petscii!).
+        ;    You must also have called text_charset() first to select and prepare the character set to use.
+        ;    NOTE: in monochrome (1bpp) screen modes, x position is currently constrained to mulitples of 8 !  TODO allow per-pixel horizontal positioning
+        uword chardataptr
+        when active_mode {
+            1, 5 -> {
+                ; monochrome mode, either resolution
+                cx16.r2 = 40
+                if active_mode>=5
+                    cx16.r2 = 80
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
+                    position(x,y)
+                    %asm {{
+                        lda  cx16.VERA_ADDR_H
+                        and  #%111              ; don't auto-increment, we have to do that manually because of the ora
+                        sta  cx16.VERA_ADDR_H
+                        lda  color
+                        sta  P8ZP_SCRATCH_B1
+                        ldy  #8
+-                       lda  P8ZP_SCRATCH_B1
+                        bne  +                  ; white color, plot normally
+                        lda  cx16.VERA_DATA1
+                        eor  #255               ; black color, keep only the other pixels
+                        and  cx16.VERA_DATA0
+                        bra  ++
++                       lda  cx16.VERA_DATA0
+                        ora  cx16.VERA_DATA1
++                       sta  cx16.VERA_DATA0
+                        lda  cx16.VERA_ADDR_L
+                        clc
+                        adc  cx16.r2
+                        sta  cx16.VERA_ADDR_L
+                        bcc  +
+                        inc  cx16.VERA_ADDR_M
++                       lda  x
+                        clc
+                        adc  #1
+                        sta  x
+                        bcc  +
+                        inc  x+1
++                       dey
+                        bne  -
+                    }}
+                    sctextptr++
+                }
+            }
+            4 -> {
+                ; lores 256c
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    cx16.vaddr(charset_bank, chardataptr, 1, 1)
+                    repeat 8 {
+                        ; TODO rewrite this inner loop fully in assembly
+                        position(x,y)
+                        y++
+                        %asm {{
+                            phx
+                            ldx  #1
+                            lda  cx16.VERA_DATA1
+                            sta  P8ZP_SCRATCH_B1
+                            ldy  #8
+-                           asl  P8ZP_SCRATCH_B1
+                            bcc  +
+                            stx  cx16.VERA_DATA0    ; write a pixel
+                            bra  ++
++                           lda  cx16.VERA_DATA0    ; don't write a pixel, but do advance to the next address
++                           dey
+                            bne  -
+                            plx
+                        }}
+                    }
+                    x+=8
+                    y-=8
+                    sctextptr++
+                }
+            }
+            6 -> {
+                ; hires 4c
+                while @(sctextptr) {
+                    chardataptr = charset_addr + (@(sctextptr) as uword)*8
+                    repeat 8 {
+                        ; TODO rewrite this inner loop fully in assembly
+                        ubyte charbits = cx16.vpeek(charset_bank, chardataptr)
+                        repeat 8 {
+                            charbits <<= 1
+                            if_cs
+                                plot(x, y, color)
+                            x++
+                        }
+                        x-=8
+                        chardataptr++
+                        y++
+                    }
+                    x+=8
+                    y-=8
+                    sctextptr++
+                }
+            }
+        }
+    }
+
+    asmsub cs_innerloop640() clobbers(Y) {
+        %asm {{
+            ldy  #80
+-           stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            stz  cx16.VERA_DATA0
+            dey
+            bne  -
+            rts
+        }}
+    }
+
+    asmsub addr_mul_24_for_highres_4c(uword yy @R2, uword xx @R3)  clobbers(A, Y) -> uword @R0, uword @R1 {
+        ; yy * 160 + xx/4  (24 bits calculation)
+        ; 24 bits result is in r0 and r1L (highest byte)
+        %asm {{
+            ldy  #5
+-           asl  cx16.r2
+            rol  cx16.r2+1
+            dey
+            bne  -
+            lda  cx16.r2
+            sta  cx16.r0
+            lda  cx16.r2+1
+            sta  cx16.r0+1
+            asl  cx16.r0
+            rol  cx16.r0+1
+            asl  cx16.r0
+            rol  cx16.r0+1
+
+            ; xx >>= 2  (xx=R3)
+            lsr  cx16.r3+1
+            ror  cx16.r3
+            lsr  cx16.r3+1
+            ror  cx16.r3
+
+            ; add r2 and xx (r3) to r0 (24-bits)
+            stz  cx16.r1
+            clc
+            lda  cx16.r0
+            adc  cx16.r2
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  cx16.r2+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           clc
+            lda  cx16.r0
+            adc  cx16.r3
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  cx16.r3+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++
+            rts
+        }}
+    }
+
+    asmsub addr_mul_24_for_lores_256c(uword yy @R0, uword xx @AY) clobbers(A) -> uword @R0, ubyte @R1  {
+        ; yy * 320 + xx (24 bits calculation)
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            lda  cx16.r0
+            sta  P8ZP_SCRATCH_B1
+            lda  cx16.r0+1
+            sta  cx16.r1
+            sta  P8ZP_SCRATCH_REG
+            lda  cx16.r0
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            asl  a
+            rol  P8ZP_SCRATCH_REG
+            sta  cx16.r0
+            lda  P8ZP_SCRATCH_B1
+            clc
+            adc  P8ZP_SCRATCH_REG
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           ; now add the value to this 24-bits number
+            lda  cx16.r0
+            clc
+            adc  P8ZP_SCRATCH_W1
+            sta  cx16.r0
+            lda  cx16.r0+1
+            adc  P8ZP_SCRATCH_W1+1
+            sta  cx16.r0+1
+            bcc  +
+            inc  cx16.r1
++           lda  cx16.r1
+            rts
+        }}
+    }
+
+}

compiler/res/prog8lib/cx16/graphics.p8

@@ -2,9 +2,12 @@
 %import syslib
 %import textio
 
-; bitmap pixel graphics module for the CommanderX16
+; Bitmap pixel graphics module for the CommanderX16
 ; wraps the graphics functions that are in ROM.
-; only black/white monchrome 320x200 for now.
+; only black/white monochrome 320x200 for now. (i.e. truncated at the bottom)
+; For full-screen 640x480 or 320x240 graphics, use the "gfx2" module instead. (but that is Cx16-specific)
+; Note: there is no color palette manipulation here, you have to do that yourself or use the "palette" module.
+
 
 graphics {
     const uword WIDTH = 320
@@ -13,15 +16,14 @@ graphics {
     sub enable_bitmap_mode() {
         ; enable bitmap screen, erase it and set colors to black/white.
         void cx16.screen_set_mode($80)
-        cx16.r0 = 0
-        cx16.GRAPH_init()
+        cx16.GRAPH_init(0)
         clear_screen(1, 0)
     }
 
     sub disable_bitmap_mode() {
         ; enables text mode, erase the text screen, color white
         void cx16.screen_set_mode(2)
-        txt.fill_screen(' ', 1)     ; TODO doesn't seem to fully clear the text screen after returning from gfx mode
+        txt.fill_screen(' ', 1)     ; doesn't seem to fully clear the text screen after returning from gfx mode
     }
 
 
@@ -31,11 +33,25 @@ graphics {
     }
 
     sub line(uword @zp x1, ubyte @zp y1, uword @zp x2, ubyte @zp y2) {
-        cx16.r0 = x1
-        cx16.r1 = y1
-        cx16.r2 = x2
-        cx16.r3 = y2
-        cx16.GRAPH_draw_line()
+        cx16.GRAPH_draw_line(x1, y1, x2, y2)
+    }
+
+    sub fillrect(uword x, uword y, uword width, uword height) {
+        cx16.GRAPH_draw_rect(x, y, width, height, 0, 1)
+    }
+
+    sub rect(uword x, uword y, uword width, uword height) {
+        cx16.GRAPH_draw_rect(x, y, width, height, 0, 0)
+    }
+
+    sub horizontal_line(uword x, uword y, uword length) {
+        if length
+            cx16.GRAPH_draw_line(x, y, x+length-1, y)
+    }
+
+    sub vertical_line(uword x, uword y, uword height) {
+        if height
+            cx16.GRAPH_draw_line(x, y, x, y+height-1)
     }
 
     sub circle(uword xcenter, ubyte ycenter, ubyte radius) {
@@ -43,122 +59,80 @@ graphics {
         ;cx16.r1 = ycenter - radius/2
         ;cx16.r2 = radius*2
         ;cx16.r3 = radius*2
-        ;cx16.GRAPH_draw_oval(false)          ; TODO  currently is not implemented on cx16, does a BRK
+        ;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
-        byte @zp decisionOver2 = 1-xx as byte
+        word @zp decisionOver2 = (1 as word)-xx
 
         while xx>=yy {
             cx16.r0 = xcenter + xx
             cx16.r1 = ycenter + yy
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter - xx
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter + xx
             cx16.r1 = ycenter - yy
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter - xx
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter + yy
             cx16.r1 = ycenter + xx
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter - yy
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter + yy
             cx16.r1 = ycenter - xx
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             cx16.r0 = xcenter - yy
-            cx16.FB_cursor_position()
+            cx16.FB_cursor_position2()
             cx16.FB_set_pixel(1)
             yy++
             if decisionOver2<=0 {
-                decisionOver2 += 2*yy+1
+                decisionOver2 += (yy as word)*2+1
             } else {
                 xx--
-                decisionOver2 += 2*(yy-xx)+1
+                decisionOver2 += (yy as word -xx)*2+1
             }
         }
     }
 
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
-;        cx16.r0 = xcenter - radius/2
-;        cx16.r1 = ycenter - radius/2
-;        cx16.r2 = radius*2
-;        cx16.r3 = radius*2
-;        cx16.GRAPH_draw_oval(true)          ; TODO  currently is not implemented on cx16, does a BRK
+        if radius==0
+            return
+        ubyte @zp yy = 0
+        word decisionOver2 = (1 as word)-radius
 
-        ubyte xx = radius
-        ubyte yy = 0
-        byte decisionOver2 = 1-xx as byte
-
-        while xx>=yy {
-            ubyte ycenter_plus_yy = ycenter + yy
-            ubyte ycenter_min_yy = ycenter - yy
-            ubyte ycenter_plus_xx = ycenter + xx
-            ubyte ycenter_min_xx = ycenter - xx
-            uword @zp plotx
-
-            for plotx in xcenter to xcenter+xx {
-                cx16.r0 = plotx
-                cx16.r1 = ycenter_plus_yy
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-                cx16.r1 = ycenter_min_yy
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-            }
-            for plotx in xcenter-xx to xcenter-1 {
-                cx16.r0 = plotx
-                cx16.r1 = ycenter_plus_yy
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-                cx16.r1 = ycenter_min_yy
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-            }
-            for plotx in xcenter to xcenter+yy {
-                cx16.r0 = plotx
-                cx16.r1 = ycenter_plus_xx
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-                cx16.r1 = ycenter_min_xx
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-            }
-            for plotx in xcenter-yy to xcenter {
-                cx16.r0 = plotx
-                cx16.r1 = ycenter_plus_xx
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-                cx16.r1 = ycenter_min_xx
-                cx16.FB_cursor_position()
-                cx16.FB_set_pixel(1)
-            }
+        while radius>=yy {
+            horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
+            horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
+            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
             yy++
             if decisionOver2<=0
-                decisionOver2 += 2*yy+1
+                decisionOver2 += (yy as word)*2+1
             else {
-                xx--
-                decisionOver2 += 2*(yy-xx)+1
+                radius--
+                decisionOver2 += (yy as word -radius)*2+1
             }
         }
     }
 
-    sub  plot(uword plotx, ubyte ploty) {
-        cx16.r0 = plotx
-        cx16.r1 = ploty
-        cx16.FB_cursor_position()
-        cx16.FB_set_pixel(1)
+    inline asmsub  plot(uword plotx @R0, uword ploty @R1) clobbers(A, X, Y) {
+        %asm {{
+            jsr  cx16.FB_cursor_position
+            lda  #1
+            jsr  cx16.FB_set_pixel
+        }}
     }
 }
-
-

compiler/res/prog8lib/cx16/palette.p8

@@ -0,0 +1,183 @@
+%target cx16
+
+; Manipulate the Commander X16's display color palette.
+; Should you want to restore the default palette, you have to reinitialize the Vera yourself.
+
+palette {
+
+    uword vera_palette_ptr
+    ubyte c
+
+    sub set_color(ubyte index, uword color) {
+        vera_palette_ptr = $fa00+index*2
+        cx16.vpoke(1, vera_palette_ptr, lsb(color))
+        vera_palette_ptr++
+        cx16.vpoke(1, vera_palette_ptr, msb(color))
+    }
+
+    sub set_rgb4(uword palette_bytes_ptr, uword num_colors) {
+        ; 2 bytes per color entry, the Vera uses this, but the R/GB bytes order is swapped
+        vera_palette_ptr = $fa00
+        repeat num_colors {
+            cx16.vpoke(1, vera_palette_ptr+1, @(palette_bytes_ptr))
+            palette_bytes_ptr++
+            cx16.vpoke(1, vera_palette_ptr, @(palette_bytes_ptr))
+            palette_bytes_ptr++
+            vera_palette_ptr+=2
+        }
+    }
+
+    sub set_rgb(uword palette_words_ptr, uword num_colors) {
+        ; 1 word per color entry (in little endian format so $gb0r)
+        vera_palette_ptr = $fa00
+        repeat num_colors*2 {
+            cx16.vpoke(1, vera_palette_ptr, @(palette_words_ptr))
+            palette_words_ptr++
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_rgb8(uword palette_bytes_ptr, uword num_colors) {
+        ; 3 bytes per color entry, adjust color depth from 8 to 4 bits per channel.
+        vera_palette_ptr = $fa00
+        ubyte red
+        ubyte greenblue
+        repeat num_colors {
+            red = @(palette_bytes_ptr) >> 4
+            palette_bytes_ptr++
+            greenblue = @(palette_bytes_ptr) & %11110000
+            palette_bytes_ptr++
+            greenblue |= @(palette_bytes_ptr) >> 4    ; add Blue
+            palette_bytes_ptr++
+            cx16.vpoke(1, vera_palette_ptr, greenblue)
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, red)
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_monochrome(uword screencolorRGB, uword drawcolorRGB) {
+        vera_palette_ptr = $fa00
+        cx16.vpoke(1, vera_palette_ptr, lsb(screencolorRGB))   ; G,B
+        vera_palette_ptr++
+        cx16.vpoke(1, vera_palette_ptr, msb(screencolorRGB))   ; R
+        vera_palette_ptr++
+        repeat 255 {
+            cx16.vpoke(1, vera_palette_ptr, lsb(drawcolorRGB)) ; G,B
+            vera_palette_ptr++
+            cx16.vpoke(1, vera_palette_ptr, msb(drawcolorRGB)) ; R
+            vera_palette_ptr++
+        }
+    }
+
+    sub set_grayscale() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            c=0
+            repeat 16 {
+                cx16.vpoke(1, vera_palette_ptr, c)
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, c)
+                vera_palette_ptr++
+                c += $11
+            }
+        }
+    }
+
+    uword[] C64_colorpalette_dark = [   ; this is a darker palette with more contrast
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $632,  ; 2 = red
+        $7AB,  ; 3 = cyan
+        $638,  ; 4 = purple
+        $584,  ; 5 = green
+        $327,  ; 6 = blue
+        $BC6,  ; 7 = yellow
+        $642,  ; 8 = orange
+        $430,  ; 9 = brown
+        $965,  ; 10 = light red
+        $444,  ; 11 = dark grey
+        $666,  ; 12 = medium grey
+        $9D8,  ; 13 = light green
+        $65B,  ; 14 = light blue
+        $999   ; 15 = light grey
+    ]
+
+    uword[] C64_colorpalette_pepto = [  ; # this is Pepto's Commodore-64 palette  http://www.pepto.de/projects/colorvic/
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $833,  ; 2 = red
+        $7cc,  ; 3 = cyan
+        $839,  ; 4 = purple
+        $5a4,  ; 5 = green
+        $229,  ; 6 = blue
+        $ef7,  ; 7 = yellow
+        $852,  ; 8 = orange
+        $530,  ; 9 = brown
+        $c67,  ; 10 = light red
+        $444,  ; 11 = dark grey
+        $777,  ; 12 = medium grey
+        $af9,  ; 13 = light green
+        $76e,  ; 14 = light blue
+        $bbb   ; 15 = light grey
+    ]
+
+    uword[] C64_colorpalette_light = [  ; this is a lighter palette
+        $000,  ; 0 = black
+        $FFF,  ; 1 = white
+        $944,  ; 2 = red
+        $7CC,  ; 3 = cyan
+        $95A,  ; 4 = purple
+        $6A5,  ; 5 = green
+        $549,  ; 6 = blue
+        $CD8,  ; 7 = yellow
+        $963,  ; 8 = orange
+        $650,  ; 9 = brown
+        $C77,  ; 10 = light red
+        $666,  ; 11 = dark grey
+        $888,  ; 12 = medium grey
+        $AE9,  ; 13 = light green
+        $87C,  ; 14 = light blue
+        $AAA   ; 15 = light grey
+    ]
+
+    sub set_c64pepto() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_pepto[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+    sub set_c64light() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_light[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+    sub set_c64dark() {
+        vera_palette_ptr = $fa00
+        repeat 16 {
+            for c in 0 to 15 {
+                uword cc = C64_colorpalette_dark[c]
+                cx16.vpoke(1, vera_palette_ptr, lsb(cc))     ; G, B
+                vera_palette_ptr++
+                cx16.vpoke(1, vera_palette_ptr, msb(cc))     ; R
+                vera_palette_ptr++
+            }
+        }
+    }
+
+}

compiler/res/prog8lib/cx16/syslib.p8

@@ -24,7 +24,7 @@ romsub $FF8D = VECTOR(uword userptr @ XY, ubyte dir @ Pc) clobbers(A,Y)     ; re
 romsub $FF90 = SETMSG(ubyte value @ A)                          ; set Kernal message control flag
 romsub $FF93 = SECOND(ubyte address @ A) clobbers(A)            ; (alias: LSTNSA) send secondary address after LISTEN
 romsub $FF96 = TKSA(ubyte address @ A) clobbers(A)              ; (alias: TALKSA) send secondary address after TALK
-romsub $FF99 = MEMTOP(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer
+romsub $FF99 = MEMTOP(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set top of memory  pointer.   NOTE: as a Cx16 extension, also returns the number of RAM memory banks in register A !  See MEMTOP2
 romsub $FF9C = MEMBOT(uword address @ XY, ubyte dir @ Pc) -> uword @ XY     ; read/set bottom of memory  pointer
 romsub $FF9F = SCNKEY() clobbers(A,X,Y)                         ; scan the keyboard
 romsub $FFA2 = SETTMO(ubyte timeout @ A)                        ; set time-out flag for IEEE bus
@@ -35,7 +35,7 @@ romsub $FFAE = UNLSN() clobbers(A)                              ; command serial
 romsub $FFB1 = LISTEN(ubyte device @ A) clobbers(A)             ; command serial bus device to LISTEN
 romsub $FFB4 = TALK(ubyte device @ A) clobbers(A)               ; command serial bus device to TALK
 romsub $FFB7 = READST() -> ubyte @ A                            ; read I/O status word
-romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte address @ Y)   ; set logical file parameters
+romsub $FFBA = SETLFS(ubyte logical @ A, ubyte device @ X, ubyte secondary @ Y)   ; set logical file parameters
 romsub $FFBD = SETNAM(ubyte namelen @ A, str filename @ XY)     ; set filename parameters
 romsub $FFC0 = OPEN() clobbers(X,Y) -> ubyte @Pc, ubyte @A      ; (via 794 ($31A)) open a logical file
 romsub $FFC3 = CLOSE(ubyte logical @ A) clobbers(A,X,Y)         ; (via 796 ($31C)) close a logical file
@@ -44,8 +44,8 @@ romsub $FFC9 = CHKOUT(ubyte logical @ X) clobbers(A,X)          ; (via 800 ($320
 romsub $FFCC = CLRCHN() clobbers(A,X)                           ; (via 802 ($322)) restore default devices
 romsub $FFCF = CHRIN() clobbers(X, Y) -> ubyte @ A   ; (via 804 ($324)) input a character (for keyboard, read a whole line from the screen) A=byte read.
 romsub $FFD2 = CHROUT(ubyte char @ A)                           ; (via 806 ($326)) output a character
-romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, ubyte @ X, ubyte @ Y     ; (via 816 ($330)) load from device
-romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A                    ; (via 818 ($332)) save to a device
+romsub $FFD5 = LOAD(ubyte verify @ A, uword address @ XY) -> ubyte @Pc, ubyte @ A, uword @ XY     ; (via 816 ($330)) load from device
+romsub $FFD8 = SAVE(ubyte zp_startaddr @ A, uword endaddr @ XY) -> ubyte @ Pc, ubyte @ A          ; (via 818 ($332)) save to a device
 romsub $FFDB = SETTIM(ubyte low @ A, ubyte middle @ X, ubyte high @ Y)      ; set the software clock
 romsub $FFDE = RDTIM() -> ubyte @ A, ubyte @ X, ubyte @ Y       ; read the software clock
 romsub $FFE1 = STOP() clobbers(X) -> ubyte @ Pz, ubyte @ A      ; (via 808 ($328)) check the STOP key (and some others in A)
@@ -56,14 +56,57 @@ romsub $FFED = SCREEN() -> ubyte @ X, ubyte @ Y                 ; read number of
 romsub $FFF0 = PLOT(ubyte col @ Y, ubyte row @ X, ubyte dir @ Pc) -> ubyte @ X, ubyte @ Y       ; read/set position of cursor on screen.  Use txt.plot for a 'safe' wrapper that preserves X.
 romsub $FFF3 = IOBASE() -> uword @ XY                           ; read base address of I/O devices
 
+; ---- utility
+
+asmsub STOP2() -> ubyte @A  {
+    ; -- check if STOP key was pressed, returns true if so.  More convenient to use than STOP() because that only sets the carry status flag.
+    %asm {{
+        phx
+        jsr  c64.STOP
+        beq  +
+        plx
+        lda  #0
+        rts
++       plx
+        lda  #1
+        rts
+    }}
+}
+
+asmsub RDTIM16() -> uword @AY {
+    ; --  like RDTIM() but only returning the lower 16 bits for convenience
+    %asm {{
+        phx
+        jsr  c64.RDTIM
+        pha
+        txa
+        tay
+        pla
+        plx
+        rts
+    }}
+}
+
+asmsub MEMTOP2() -> ubyte @A {
+    ; -- uses MEMTOP's cx16 extension to query the number of available RAM banks.
+    %asm {{
+        phx
+        sec
+        jsr  c64.MEMTOP
+        plx
+        rts
+    }}
+}
+
 }
 
 cx16 {
 
-; 65c02 hardware vectors:
-    &uword  NMI_VEC         = $FFFA     ; 6502 nmi vector, determined by the kernal if banked in
-    &uword  RESET_VEC       = $FFFC     ; 6502 reset vector, determined by the kernal if banked in
-    &uword  IRQ_VEC         = $FFFE     ; 6502 interrupt vector, determined by the kernal if banked in
+; irq and hardware vectors:
+    &uword  CINV            = $0314     ; IRQ vector (in ram)
+    &uword  NMI_VEC         = $FFFA     ; 65c02 nmi vector, determined by the kernal if banked in
+    &uword  RESET_VEC       = $FFFC     ; 65c02 reset vector, determined by the kernal if banked in
+    &uword  IRQ_VEC         = $FFFE     ; 65c02 interrupt vector, determined by the kernal if banked in
 
 
 ; the sixteen virtual 16-bit registers
@@ -86,50 +129,50 @@ cx16 {
 
 ; VERA registers
 
-    const uword VERA_BASE = $9F20
-    &ubyte  VERA_ADDR_L   	  = VERA_BASE + $0000
-    &ubyte  VERA_ADDR_M   	  = VERA_BASE + $0001
-    &ubyte  VERA_ADDR_H   	  = VERA_BASE + $0002
-    &ubyte  VERA_DATA0        = VERA_BASE + $0003
-    &ubyte  VERA_DATA1        = VERA_BASE + $0004
-    &ubyte  VERA_CTRL         = VERA_BASE + $0005
-    &ubyte  VERA_IEN          = VERA_BASE + $0006
-    &ubyte  VERA_ISR          = VERA_BASE + $0007
-    &ubyte  VERA_IRQ_LINE_L   = VERA_BASE + $0008
-    &ubyte  VERA_DC_VIDEO     = VERA_BASE + $0009
-    &ubyte  VERA_DC_HSCALE    = VERA_BASE + $000A
-    &ubyte  VERA_DC_VSCALE    = VERA_BASE + $000B
-    &ubyte  VERA_DC_BORDER    = VERA_BASE + $000C
-    &ubyte  VERA_DC_HSTART    = VERA_BASE + $0009
-    &ubyte  VERA_DC_HSTOP     = VERA_BASE + $000A
-    &ubyte  VERA_DC_VSTART    = VERA_BASE + $000B
-    &ubyte  VERA_DC_VSTOP     = VERA_BASE + $000C
-    &ubyte  VERA_L0_CONFIG    = VERA_BASE + $000D
-    &ubyte  VERA_L0_MAPBASE   = VERA_BASE + $000E
-    &ubyte  VERA_L0_TILEBASE  = VERA_BASE + $000F
-    &ubyte  VERA_L0_HSCROLL_L = VERA_BASE + $0010
-    &ubyte  VERA_L0_HSCROLL_H = VERA_BASE + $0011
-    &ubyte  VERA_L0_VSCROLL_L = VERA_BASE + $0012
-    &ubyte  VERA_L0_VSCROLL_H = VERA_BASE + $0013
-    &ubyte  VERA_L1_CONFIG    = VERA_BASE + $0014
-    &ubyte  VERA_L1_MAPBASE   = VERA_BASE + $0015
-    &ubyte  VERA_L1_TILEBASE  = VERA_BASE + $0016
-    &ubyte  VERA_L1_HSCROLL_L = VERA_BASE + $0017
-    &ubyte  VERA_L1_HSCROLL_H = VERA_BASE + $0018
-    &ubyte  VERA_L1_VSCROLL_L = VERA_BASE + $0019
-    &ubyte  VERA_L1_VSCROLL_H = VERA_BASE + $001A
-    &ubyte  VERA_AUDIO_CTRL   = VERA_BASE + $001B
-    &ubyte  VERA_AUDIO_RATE   = VERA_BASE + $001C
-    &ubyte  VERA_AUDIO_DATA   = VERA_BASE + $001D
-    &ubyte  VERA_SPI_DATA     = VERA_BASE + $001E
-    &ubyte  VERA_SPI_CTRL     = VERA_BASE + $001F
+    const uword VERA_BASE       = $9F20
+    &ubyte  VERA_ADDR_L         = VERA_BASE + $0000
+    &ubyte  VERA_ADDR_M         = VERA_BASE + $0001
+    &ubyte  VERA_ADDR_H         = VERA_BASE + $0002
+    &ubyte  VERA_DATA0          = VERA_BASE + $0003
+    &ubyte  VERA_DATA1          = VERA_BASE + $0004
+    &ubyte  VERA_CTRL           = VERA_BASE + $0005
+    &ubyte  VERA_IEN            = VERA_BASE + $0006
+    &ubyte  VERA_ISR            = VERA_BASE + $0007
+    &ubyte  VERA_IRQ_LINE_L     = VERA_BASE + $0008
+    &ubyte  VERA_DC_VIDEO       = VERA_BASE + $0009
+    &ubyte  VERA_DC_HSCALE      = VERA_BASE + $000A
+    &ubyte  VERA_DC_VSCALE      = VERA_BASE + $000B
+    &ubyte  VERA_DC_BORDER      = VERA_BASE + $000C
+    &ubyte  VERA_DC_HSTART      = VERA_BASE + $0009
+    &ubyte  VERA_DC_HSTOP       = VERA_BASE + $000A
+    &ubyte  VERA_DC_VSTART      = VERA_BASE + $000B
+    &ubyte  VERA_DC_VSTOP       = VERA_BASE + $000C
+    &ubyte  VERA_L0_CONFIG      = VERA_BASE + $000D
+    &ubyte  VERA_L0_MAPBASE     = VERA_BASE + $000E
+    &ubyte  VERA_L0_TILEBASE    = VERA_BASE + $000F
+    &ubyte  VERA_L0_HSCROLL_L   = VERA_BASE + $0010
+    &ubyte  VERA_L0_HSCROLL_H   = VERA_BASE + $0011
+    &ubyte  VERA_L0_VSCROLL_L   = VERA_BASE + $0012
+    &ubyte  VERA_L0_VSCROLL_H   = VERA_BASE + $0013
+    &ubyte  VERA_L1_CONFIG      = VERA_BASE + $0014
+    &ubyte  VERA_L1_MAPBASE     = VERA_BASE + $0015
+    &ubyte  VERA_L1_TILEBASE    = VERA_BASE + $0016
+    &ubyte  VERA_L1_HSCROLL_L   = VERA_BASE + $0017
+    &ubyte  VERA_L1_HSCROLL_H   = VERA_BASE + $0018
+    &ubyte  VERA_L1_VSCROLL_L   = VERA_BASE + $0019
+    &ubyte  VERA_L1_VSCROLL_H   = VERA_BASE + $001A
+    &ubyte  VERA_AUDIO_CTRL     = VERA_BASE + $001B
+    &ubyte  VERA_AUDIO_RATE     = VERA_BASE + $001C
+    &ubyte  VERA_AUDIO_DATA     = VERA_BASE + $001D
+    &ubyte  VERA_SPI_DATA       = VERA_BASE + $001E
+    &ubyte  VERA_SPI_CTRL       = VERA_BASE + $001F
 ; VERA_PSG_BASE     = $1F9C0
 ; VERA_PALETTE_BASE = $1FA00
 ; VERA_SPRITES_BASE = $1FC00
 
 ; I/O
 
-    const uword  via1  = $9f60                  ;VIA 6522 #1
+    const uword  via1   = $9f60                  ;VIA 6522 #1
     &ubyte  d1prb	= via1+0
     &ubyte  d1pra	= via1+1
     &ubyte  d1ddrb	= via1+2
@@ -147,23 +190,23 @@ cx16 {
     &ubyte  d1ier	= via1+14
     &ubyte  d1ora	= via1+15
 
-    const uword  via2  = $9f70                  ;VIA 6522 #2
-    &ubyte  d2prb	=via2+0
-    &ubyte  d2pra	=via2+1
-    &ubyte  d2ddrb	=via2+2
-    &ubyte  d2ddra	=via2+3
-    &ubyte  d2t1l	=via2+4
-    &ubyte  d2t1h	=via2+5
-    &ubyte  d2t1ll	=via2+6
-    &ubyte  d2t1lh	=via2+7
-    &ubyte  d2t2l	=via2+8
-    &ubyte  d2t2h	=via2+9
-    &ubyte  d2sr	=via2+10
-    &ubyte  d2acr	=via2+11
-    &ubyte  d2pcr	=via2+12
-    &ubyte  d2ifr	=via2+13
-    &ubyte  d2ier	=via2+14
-    &ubyte  d2ora	=via2+15
+    const uword  via2   = $9f70                  ;VIA 6522 #2
+    &ubyte  d2prb	= via2+0
+    &ubyte  d2pra	= via2+1
+    &ubyte  d2ddrb	= via2+2
+    &ubyte  d2ddra	= via2+3
+    &ubyte  d2t1l	= via2+4
+    &ubyte  d2t1h	= via2+5
+    &ubyte  d2t1ll	= via2+6
+    &ubyte  d2t1lh	= via2+7
+    &ubyte  d2t2l	= via2+8
+    &ubyte  d2t2h	= via2+9
+    &ubyte  d2sr	= via2+10
+    &ubyte  d2acr	= via2+11
+    &ubyte  d2pcr	= via2+12
+    &ubyte  d2ifr	= via2+13
+    &ubyte  d2ier	= via2+14
+    &ubyte  d2ora	= via2+15
 
 
 ; ---- Commander X-16 additions on top of C64 kernal routines ----
@@ -190,62 +233,268 @@ romsub $ff6b = mouse_get(ubyte zpdataptr @X)  clobbers(A)
 romsub $ff71 = mouse_scan()  clobbers(A, X, Y)
 romsub $ff53 = joystick_scan()  clobbers(A, X, Y)
 romsub $ff56 = joystick_get(ubyte joynr @A) -> ubyte @A, ubyte @X, ubyte @Y
-romsub $ff4d = clock_set_date_time()  clobbers(A, X, Y)      ; args: r0, r1, r2, r3L
-romsub $ff50 = clock_get_date_time()  clobbers(A, X, Y)      ; outout args: r0, r1, r2, r3L
+romsub $ff4d = clock_set_date_time(uword yearmonth @R0, uword dayhours @R1, uword minsecs @R2, ubyte jiffies @R3)  clobbers(A, X, Y)
+romsub $ff50 = clock_get_date_time()  clobbers(A, X, Y)  -> uword @R0, uword @R1, uword @R2, ubyte @R3   ; result registers see clock_set_date_time()
 
 ; TODO specify the correct clobbers for alle these functions below, we now assume all 3 regs are clobbered
 
 ; high level graphics & fonts
-romsub $ff20 = GRAPH_init()  clobbers(A,X,Y)           ; uses vectors=r0
+romsub $ff20 = GRAPH_init(uword vectors @R0)  clobbers(A,X,Y)
 romsub $ff23 = GRAPH_clear()  clobbers(A,X,Y)
-romsub $ff26 = GRAPH_set_window()  clobbers(A,X,Y)      ; uses x=r0, y=r1, width=r2, height=r3
+romsub $ff26 = GRAPH_set_window(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
 romsub $ff29 = GRAPH_set_colors(ubyte stroke @A, ubyte fill @X, ubyte background @Y)  clobbers (A,X,Y)
-romsub $ff2c = GRAPH_draw_line()  clobbers(A,X,Y)       ; uses x1=r0, y1=r1, x2=r2, y2=r3
-romsub $ff2f = GRAPH_draw_rect(ubyte fill @Pc)  clobbers(A,X,Y)   ; uses x=r0, y=r1, width=r2, height=r3, cornerradius=r4
-romsub $ff32 = GRAPH_move_rect()  clobbers(A,X,Y)       ; uses sx=r0, sy=r1, tx=r2, ty=r3, width=r4, height=r5
-romsub $ff35 = GRAPH_draw_oval(ubyte fill @Pc)  clobbers(A,X,Y)       ; uses x=r0, y=r1, width=r2, height=r3
-romsub $ff38 = GRAPH_draw_image()  clobbers(A,X,Y)      ; uses x=r0, y=r1, ptr=r2, width=r3, height=r4
-romsub $ff3b = GRAPH_set_font()  clobbers(A,X,Y)         ; uses ptr=r0
+romsub $ff2c = GRAPH_draw_line(uword x1 @R0, uword y1 @R1, uword x2 @R2, uword y2 @R3)  clobbers(A,X,Y)
+romsub $ff2f = GRAPH_draw_rect(uword x @R0, uword y @R1, uword width @R2, uword height @R3, uword cornerradius @R4, ubyte fill @Pc)  clobbers(A,X,Y)
+romsub $ff32 = GRAPH_move_rect(uword sx @R0, uword sy @R1, uword tx @R2, uword ty @R3, uword width @R4, uword height @R5)  clobbers(A,X,Y)
+romsub $ff35 = GRAPH_draw_oval(uword x @R0, uword y @R1, uword width @R2, uword height @R3, ubyte fill @Pc)  clobbers(A,X,Y)
+romsub $ff38 = GRAPH_draw_image(uword x @R0, uword y @R1, uword ptr @R2, uword width @R3, uword height @R4)  clobbers(A,X,Y)
+romsub $ff3b = GRAPH_set_font(uword fontptr @R0)  clobbers(A,X,Y)
 romsub $ff3e = GRAPH_get_char_size(ubyte baseline @A, ubyte width @X, ubyte height_or_style @Y, ubyte is_control @Pc)  clobbers(A,X,Y)
-romsub $ff41 = GRAPH_put_char(ubyte char @A)  clobbers(A,X,Y)   ; uses x=r0, y=r1
+romsub $ff41 = GRAPH_put_char(uword x @R0, uword y @R1, ubyte char @A)  clobbers(A,X,Y)
+romsub $ff41 = GRAPH_put_next_char(ubyte char @A)  clobbers(A,X,Y)     ; alias for the routine above that doesn't reset the position of the initial character
 
 ; framebuffer
 romsub $fef6 = FB_init()  clobbers(A,X,Y)
-romsub $fef9 = FB_get_info()  clobbers(X,Y) -> byte @A    ; also outputs width=r0, height=r1
-romsub $fefc = FB_set_palette(ubyte index @A, ubyte bytecount @X)  clobbers(A,X,Y)      ; also uses pointer=r0
-romsub $feff = FB_cursor_position()  clobbers(A,X,Y)    ; uses x=r0, y=r1
-romsub $ff02 = FB_cursor_next_line()  clobbers(A,X,Y)   ; uses x=r0
+romsub $fef9 = FB_get_info()  clobbers(X,Y) -> byte @A, uword @R0, uword @R1    ; width=r0, height=r1
+romsub $fefc = FB_set_palette(uword pointer @R0, ubyte index @A, ubyte bytecount @X)  clobbers(A,X,Y)
+romsub $feff = FB_cursor_position(uword x @R0, uword y @R1)  clobbers(A,X,Y)
+romsub $feff = FB_cursor_position2()  clobbers(A,X,Y)           ;  alias for the previous routine, but avoiding having to respecify both x and y every time
+romsub $ff02 = FB_cursor_next_line(uword x @R0)  clobbers(A,X,Y)
 romsub $ff05 = FB_get_pixel()  clobbers(X,Y) -> ubyte @A
-romsub $ff08 = FB_get_pixels()  clobbers(A,X,Y)         ; uses ptr=r0, count=r1
+romsub $ff08 = FB_get_pixels(uword pointer @R0, uword count @R1)  clobbers(A,X,Y)
 romsub $ff0b = FB_set_pixel(ubyte color @A)  clobbers(A,X,Y)
-romsub $ff0e = FB_set_pixels()  clobbers(A,X,Y)         ; uses ptr=r0, count=r1
+romsub $ff0e = FB_set_pixels(uword pointer @R0, uword count @R1)  clobbers(A,X,Y)
 romsub $ff11 = FB_set_8_pixels(ubyte pattern @A, ubyte color @X)  clobbers(A,X,Y)
-romsub $ff14 = FB_set_8_pixels_opaque(ubyte pattern @A, ubyte color1 @X, ubyte color2 @Y)  clobbers(A,X,Y)  ; also uses mask=r0L
-romsub $ff17 = FB_fill_pixels(ubyte color @A)  clobbers(A,X,Y)   ; also uses count=r0, step=r1
-romsub $ff1a = FB_filter_pixels()  clobbers(A,X,Y)      ; uses ptr=r0, count=r1
-romsub $ff1d = FB_move_pixels()  clobbers(A,X,Y)        ; uses sx=r0, sy=r1, tx=r2, ty=r3, count=r4
+romsub $ff14 = FB_set_8_pixels_opaque(ubyte pattern @R0, ubyte mask @A, ubyte color1 @X, ubyte color2 @Y)  clobbers(A,X,Y)
+romsub $ff17 = FB_fill_pixels(uword count @R0, uword pstep @R1, ubyte color @A)  clobbers(A,X,Y)
+romsub $ff1a = FB_filter_pixels(uword pointer @ R0, uword count @R1)  clobbers(A,X,Y)
+romsub $ff1d = FB_move_pixels(uword sx @R0, uword sy @R1, uword tx @R2, uword ty @R3, uword count @R4)  clobbers(A,X,Y)
 
 ; misc
-romsub $fef0 = sprite_set_image(ubyte number @A, ubyte width @X, ubyte height @Y, ubyte apply_mask @Pc)  clobbers(A,X,Y) -> ubyte @Pc  ; also uses pixels=r0, mask=r1, bpp=r2L
-romsub $fef3 = sprite_set_position(ubyte number @A)  clobbers(A,X,Y)  ; also uses x=r0 and y=r1
-romsub $fee4 = memory_fill(ubyte value @A)  clobbers(A,X,Y)      ; uses address=r0, num_bytes=r1
-romsub $fee7 = memory_copy()  clobbers(A,X,Y)                    ; uses source=r0, target=r1, num_bytes=r2
-romsub $feea = memory_crc()  clobbers(A,X,Y)                     ; uses address=r0, num_bytes=r1     result->r2
-romsub $feed = memory_decompress()  clobbers(A,X,Y)              ; uses input=r0, output=r1     result->r1
-romsub $fedb = console_init()  clobbers(A,X,Y)           ; uses x=r0, y=r1, width=r2, height=r3
+romsub $fef0 = sprite_set_image(uword pixels @R0, uword mask @R1, ubyte bpp @R2, ubyte number @A, ubyte width @X, ubyte height @Y, ubyte apply_mask @Pc)  clobbers(A,X,Y) -> ubyte @Pc
+romsub $fef3 = sprite_set_position(uword x @R0, uword y @R1, ubyte number @A)  clobbers(A,X,Y)
+romsub $fee4 = memory_fill(uword address @R0, uword num_bytes @R1, ubyte value @A)  clobbers(A,X,Y)
+romsub $fee7 = memory_copy(uword source @R0, uword target @R1, uword num_bytes @R2)  clobbers(A,X,Y)
+romsub $feea = memory_crc(uword address @R0, uword num_bytes @R1)  clobbers(A,X,Y) -> uword @R2
+romsub $feed = memory_decompress(uword input @R0, uword output @R1)  clobbers(A,X,Y) -> uword @R1       ; last address +1 is result in R1
+romsub $fedb = console_init(uword x @R0, uword y @R1, uword width @R2, uword height @R3)  clobbers(A,X,Y)
 romsub $fede = console_put_char(ubyte char @A, ubyte wrapping @Pc)  clobbers(A,X,Y)
 romsub $fee1 = console_get_char()  clobbers(X,Y) -> ubyte @A
-romsub $fed8 = console_put_image()  clobbers(A,X,Y)      ; uses ptr=r0, width=r1, height=r2
-romsub $fed5 = console_set_paging_message()  clobbers(A,X,Y)     ; uses messageptr=r0
+romsub $fed8 = console_put_image(uword pointer @R0, uword width @R1, uword height @R2)  clobbers(A,X,Y)
+romsub $fed5 = console_set_paging_message(uword msgptr @R0)  clobbers(A,X,Y)
 romsub $fed2 = kbdbuf_put(ubyte key @A)  clobbers(A,X,Y)
 romsub $fecf = entropy_get() -> ubyte @A, ubyte @X, ubyte @Y
 romsub $fecc = monitor()  clobbers(A,X,Y)
 
-
-
 ; ---- end of kernal routines ----
 
-asmsub init_system()  {
+
+; ---- utilities -----
+
+inline asmsub rombank(ubyte rombank @A) {
+    ; -- set the rom banks
+    %asm {{
+        sta  $01            ; rom bank register (new)
+        sta  cx16.d1prb     ; rom bank register (old)
+    }}
+}
+
+inline asmsub rambank(ubyte rambank @A) {
+    ; -- set the ram bank
+    %asm {{
+        sta  $00            ; ram bank register (new)
+        sta  cx16.d1pra     ; ram bank register (old)
+    }}
+}
+
+asmsub vpeek(ubyte bank @A, uword address @XY) -> ubyte @A {
+        ; -- get a byte from VERA's video memory
+        ;    note: inefficient when reading multiple sequential bytes!
+        %asm {{
+                pha
+                lda  #1
+                sta  cx16.VERA_CTRL
+                pla
+                and  #1
+                sta  cx16.VERA_ADDR_H
+                sty  cx16.VERA_ADDR_M
+                stx  cx16.VERA_ADDR_L
+                lda  cx16.VERA_DATA1
+                rts
+            }}
+}
+
+asmsub vaddr(ubyte bank @A, uword address @R0, ubyte addrsel @R1, byte autoIncrOrDecrByOne @Y) clobbers(A) {
+        ; -- setup the VERA's data address register 0 or 1
+        %asm {{
+            and  #1
+            pha
+            lda  cx16.r1
+            and  #1
+            sta  cx16.VERA_CTRL
+            lda  cx16.r0
+            sta  cx16.VERA_ADDR_L
+            lda  cx16.r0+1
+            sta  cx16.VERA_ADDR_M
+            pla
+            cpy  #0
+            bmi  ++
+            beq  +
+            ora  #%00010000
++           sta  cx16.VERA_ADDR_H
+            rts
++           ora  #%00011000
+            sta  cx16.VERA_ADDR_H
+            rts
+        }}
+}
+
+asmsub vpoke(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers(A) {
+    ; -- write a single byte to VERA's video memory
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        sty  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_or(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers (A) {
+    ; -- or a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        ora  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_and(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers(A) {
+    ; -- and a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        and  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vpoke_xor(ubyte bank @A, uword address @R0, ubyte value @Y) clobbers (A) {
+    ; -- xor a single byte to the value already in the VERA's video memory at that location
+    ;    note: inefficient when writing multiple sequential bytes!
+    %asm {{
+        stz  cx16.VERA_CTRL
+        and  #1
+        sta  cx16.VERA_ADDR_H
+        lda  cx16.r0
+        sta  cx16.VERA_ADDR_L
+        lda  cx16.r0+1
+        sta  cx16.VERA_ADDR_M
+        tya
+        eor  cx16.VERA_DATA0
+        sta  cx16.VERA_DATA0
+        rts
+    }}
+}
+
+asmsub vload(str name @R0, ubyte device @Y, ubyte bank @A, uword address @R1) -> ubyte @A {
+    ; -- like the basic command VLOAD "filename",device,bank,address
+    ;    loads a file into video memory in the given bank:address, returns success in A
+    ;    !! NOTE !! the V38 ROMs contain a bug in the LOAD code that makes the load address not work correctly,
+    ;               it works fine when loading from local filesystem
+    %asm {{
+        ; -- load a file into video ram
+        phx
+        pha
+        tya
+        tax
+        lda  #1
+        ldy  #0
+        jsr  c64.SETLFS
+        lda  cx16.r0
+        ldy  cx16.r0+1
+        jsr  prog8_lib.strlen
+        tya
+        ldx  cx16.r0
+        ldy  cx16.r0+1
+        jsr  c64.SETNAM
+        pla
+        clc
+        adc  #2
+        ldx  cx16.r1
+        ldy  cx16.r1+1
+        stz  P8ZP_SCRATCH_B1
+        jsr  c64.LOAD
+        bcs  +
+        inc  P8ZP_SCRATCH_B1
++       jsr  c64.CLRCHN
+        lda  #1
+        jsr  c64.CLOSE
+        plx
+        lda  P8ZP_SCRATCH_B1
+        rts
+    }}
+}
+
+
+sub FB_set_pixels_from_buf(uword buffer, uword count) {
+    %asm {{
+            ; -- This is replacement code for the normal FB_set_pixels subroutine in ROM
+            ;    However that routine contains a bug in the current v38 ROM that makes it crash when count > 255.
+            ;    So the code below replaces that. Once the ROM is patched this routine is no longer necessary.
+            ;    See https://github.com/commanderx16/x16-rom/issues/179
+            phx
+            lda  buffer
+            ldy  buffer+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            jsr  _pixels
+            plx
+            rts
+
+_pixels     lda  count+1
+            beq  +
+            ldx  #0
+-           jsr  _loop
+            inc  P8ZP_SCRATCH_W1+1
+            dec  count+1
+            bne  -
+
++           ldx  count
+_loop       ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            sta  cx16.VERA_DATA0
+            iny
+            dex
+            bne  -
+            rts
+        }}
+}
+
+; ---- system stuff -----
+asmsub  init_system()  {
     ; Initializes the machine to a sane starting state.
     ; Called automatically by the loader program logic.
     %asm {{
@@ -253,7 +502,7 @@ asmsub init_system()  {
         cld
         ;stz  $00
         ;stz  $01
-        ;stz  d1prb      ; select rom bank 0
+        ;stz  d1prb      ; select rom bank 0 (enable kernal)
         lda  #$80
         sta  VERA_CTRL
         jsr  c64.IOINIT
@@ -277,15 +526,297 @@ asmsub init_system()  {
     }}
 }
 
-asmsub  reset_system()  {
-    ; Soft-reset the system back to Basic prompt.
+asmsub  init_system_phase2()  {
     %asm {{
         sei
-        lda  #14
-        sta  $01
-        stz  cx16.d1prb         ; bank the kernal in
-        jmp  (cx16.RESET_VEC)
+        lda  cx16.CINV
+        sta  restore_irq._orig_irqvec
+        lda  cx16.CINV+1
+        sta  restore_irq._orig_irqvec+1
+        cli
+        rts
+    }}
+}
+
+asmsub  set_irq(uword handler @AY, ubyte useKernal @Pc) clobbers(A)  {
+	%asm {{
+	        sta  _modified+1
+	        sty  _modified+2
+	        lda  #0
+	        adc  #0
+	        sta  _use_kernal
+		sei
+		lda  #<_irq_handler
+		sta  cx16.CINV
+		lda  #>_irq_handler
+		sta  cx16.CINV+1
+                lda  cx16.VERA_IEN
+                ora  #%00000001     ; enable the vsync irq
+                sta  cx16.VERA_IEN
+		cli
+		rts
+
+_irq_handler    jsr  _irq_handler_init
+_modified	jsr  $ffff                      ; modified
+		jsr  _irq_handler_end
+		lda  _use_kernal
+		bne  +
+		; end irq processing - don't use kernal's irq handling
+		lda  cx16.VERA_ISR
+		ora  #1
+		sta  cx16.VERA_ISR      ; clear Vera Vsync irq status
+		ply
+		plx
+		pla
+		rti
++		jmp  (restore_irq._orig_irqvec)   ; continue with normal kernal irq routine
+
+_use_kernal     .byte  0
+
+_irq_handler_init
+		; save all zp scratch registers and the X register as these might be clobbered by the irq routine
+		stx  IRQ_X_REG
+		lda  P8ZP_SCRATCH_B1
+		sta  IRQ_SCRATCH_ZPB1
+		lda  P8ZP_SCRATCH_REG
+		sta  IRQ_SCRATCH_ZPREG
+		lda  P8ZP_SCRATCH_W1
+		sta  IRQ_SCRATCH_ZPWORD1
+		lda  P8ZP_SCRATCH_W1+1
+		sta  IRQ_SCRATCH_ZPWORD1+1
+		lda  P8ZP_SCRATCH_W2
+		sta  IRQ_SCRATCH_ZPWORD2
+		lda  P8ZP_SCRATCH_W2+1
+		sta  IRQ_SCRATCH_ZPWORD2+1
+		; stack protector; make sure we don't clobber the top of the evaluation stack
+		dex
+		dex
+		dex
+		dex
+		dex
+		dex
+		cld
+		rts
+
+_irq_handler_end
+		; restore all zp scratch registers and the X register
+		lda  IRQ_SCRATCH_ZPB1
+		sta  P8ZP_SCRATCH_B1
+		lda  IRQ_SCRATCH_ZPREG
+		sta  P8ZP_SCRATCH_REG
+		lda  IRQ_SCRATCH_ZPWORD1
+		sta  P8ZP_SCRATCH_W1
+		lda  IRQ_SCRATCH_ZPWORD1+1
+		sta  P8ZP_SCRATCH_W1+1
+		lda  IRQ_SCRATCH_ZPWORD2
+		sta  P8ZP_SCRATCH_W2
+		lda  IRQ_SCRATCH_ZPWORD2+1
+		sta  P8ZP_SCRATCH_W2+1
+		ldx  IRQ_X_REG
+		rts
+
+IRQ_X_REG		.byte  0
+IRQ_SCRATCH_ZPB1	.byte  0
+IRQ_SCRATCH_ZPREG	.byte  0
+IRQ_SCRATCH_ZPWORD1	.word  0
+IRQ_SCRATCH_ZPWORD2	.word  0
+
+		}}
+}
+
+
+asmsub  restore_irq() clobbers(A) {
+	%asm {{
+	    sei
+	    lda  _orig_irqvec
+	    sta  cx16.CINV
+	    lda  _orig_irqvec+1
+	    sta  cx16.CINV+1
+	    lda  cx16.VERA_IEN
+	    and  #%11110000     ; disable all Vera IRQs
+	    ora  #%00000001     ; enable only the vsync Irq
+	    sta  cx16.VERA_IEN
+	    cli
+	    rts
+_orig_irqvec    .word  0
+        }}
+}
+
+asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
+	%asm {{
+            sta  _modified+1
+            sty  _modified+2
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            sei
+            lda  cx16.VERA_IEN
+            and  #%11110000     ; clear other IRQs
+            ora  #%00000010     ; enable the line (raster) irq
+            sta  cx16.VERA_IEN
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            jsr  set_rasterline
+            lda  #<_raster_irq_handler
+            sta  cx16.CINV
+            lda  #>_raster_irq_handler
+            sta  cx16.CINV+1
+            cli
+            rts
+
+_raster_irq_handler
+            jsr  set_irq._irq_handler_init
+_modified   jsr  $ffff                      ; modified
+            jsr  set_irq._irq_handler_end
+            ; end irq processing - don't use kernal's irq handling
+            lda  cx16.VERA_ISR
+            ora  #%00000010
+            sta  cx16.VERA_ISR      ; clear Vera line irq status
+            ply
+            plx
+            pla
+            rti
+        }}
+}
+
+asmsub  set_rasterline(uword line @AY) {
+    %asm {{
+        sta  cx16.VERA_IRQ_LINE_L
+        lda  cx16.VERA_IEN
+        and  #%01111111
+        sta  cx16.VERA_IEN
+        tya
+        lsr  a
+        ror  a
+        and  #%10000000
+        ora  cx16.VERA_IEN
+        sta  cx16.VERA_IEN
+        rts
     }}
 }
 
 }
+
+
+sys {
+    ; ------- lowlevel system routines --------
+
+    const ubyte target = 16         ;  compilation target specifier.  64 = C64,  16 = CommanderX16.
+
+
+    asmsub reset_system() {
+        ; Soft-reset the system back to Basic prompt.
+        %asm {{
+            sei
+            stz  $01            ; bank the kernal in (new rom bank register)
+            stz  cx16.d1prb     ; bank the kernal in (old rom bank register)
+            jmp  (cx16.RESET_VEC)
+        }}
+    }
+
+    sub wait(uword jiffies) {
+        ; --- wait approximately the given number of jiffies (1/60th seconds)
+        repeat jiffies {
+            ubyte jiff = lsb(c64.RDTIM16())
+            while jiff==lsb(c64.RDTIM16()) {
+                ; wait until 1 jiffy has passed
+            }
+        }
+    }
+
+    inline asmsub memcopy(uword source @R0, uword target @R1, uword count @AY) clobbers(A,X,Y) {
+        %asm {{
+            sta  cx16.r2
+            sty  cx16.r2+1
+            jsr  cx16.memory_copy
+        }}
+    }
+
+    inline asmsub memset(uword mem @R0, uword numbytes @R1, ubyte value @A) clobbers(A,X,Y) {
+        %asm {{
+            jsr  cx16.memory_fill
+        }}
+    }
+
+    asmsub memsetw(uword mem @R0, uword numwords @R1, uword value @AY) clobbers (A,X,Y) {
+        %asm {{
+            ldx  cx16.r0
+            stx  P8ZP_SCRATCH_W1
+            ldx  cx16.r0+1
+            stx  P8ZP_SCRATCH_W1+1
+            ldx  cx16.r1
+            stx  P8ZP_SCRATCH_W2
+            ldx  cx16.r1+1
+            stx  P8ZP_SCRATCH_W2+1
+            jmp  prog8_lib.memsetw
+        }}
+    }
+
+    inline asmsub rsave() {
+        ; save cpu status flag and all registers A, X, Y.
+        ; see http://6502.org/tutorials/register_preservation.html
+        %asm {{
+            php
+            pha
+            phy
+            phx
+        }}
+    }
+
+    inline asmsub rrestore() {
+        ; restore all registers and cpu status flag
+        %asm {{
+            plx
+            ply
+            pla
+            plp
+        }}
+    }
+
+    inline asmsub read_flags() -> ubyte @A {
+        %asm {{
+            php
+            pla
+        }}
+    }
+
+    inline asmsub clear_carry() {
+        %asm {{
+            clc
+        }}
+    }
+
+    inline asmsub set_carry() {
+        %asm {{
+            sec
+        }}
+    }
+
+    inline asmsub clear_irqd() {
+        %asm {{
+            cli
+        }}
+    }
+
+    inline asmsub set_irqd() {
+        %asm {{
+            sei
+        }}
+    }
+
+    inline asmsub exit(ubyte returnvalue @A) {
+        ; -- immediately exit the program with a return code in the A register
+        %asm {{
+            jsr  c64.CLRCHN		; reset i/o channels
+            ldx  prog8_lib.orig_stackpointer
+            txs
+            rts		; return to original caller
+        }}
+    }
+
+    inline asmsub progend() -> uword @AY {
+        %asm {{
+            lda  #<prog8_program_end
+            ldy  #>prog8_program_end
+        }}
+    }
+}

compiler/res/prog8lib/cx16/textio.p8

@@ -15,8 +15,31 @@ const ubyte DEFAULT_WIDTH = 80
 const ubyte DEFAULT_HEIGHT = 60
 
 
-sub  clear_screen() {
-    clear_screenchars(' ')
+sub clear_screen() {
+    txt.chrout(147)
+}
+
+sub home() {
+    txt.chrout(19)
+}
+
+sub nl() {
+    txt.chrout('\n')
+}
+
+sub spc() {
+    txt.chrout(' ')
+}
+
+asmsub column(ubyte col @A) clobbers(A, X, Y) {
+    ; ---- set the cursor on the given column (starting with 0) on the current line
+    %asm {{
+        sec
+        jsr  c64.PLOT
+        tay
+        clc
+        jmp  c64.PLOT
+    }}
 }
 
 asmsub  fill_screen (ubyte char @ A, ubyte color @ Y) clobbers(A)  {
@@ -154,10 +177,9 @@ sub uppercase() {
     cx16.screen_set_charset(2, 0)  ; uppercase charset
 }
 
-asmsub  scroll_left  (ubyte dummy @ Pc) clobbers(A, Y)  {
+asmsub  scroll_left() clobbers(A, Y)  {
 	; ---- scroll the whole screen 1 character to the left
 	;      contents of the rightmost column are unchanged, you should clear/refill this yourself
-	;      Carry flag is a dummy on the cx16
 	%asm {{
 	    phx
 	    jsr  c64.SCREEN
@@ -198,10 +220,9 @@ _lx     ldx  #0                ; modified
 	}}
 }
 
-asmsub  scroll_right  (ubyte dummy @ Pc) clobbers(A)  {
+asmsub  scroll_right() clobbers(A)  {
 	; ---- scroll the whole screen 1 character to the right
 	;      contents of the leftmost column are unchanged, you should clear/refill this yourself
-	;      Carry flag is a dummy on the cx16
 	%asm {{
 	    phx
 	    jsr  c64.SCREEN
@@ -250,10 +271,9 @@ _lx     ldx  #0                 ; modified
 	}}
 }
 
-asmsub  scroll_up  (ubyte dummy @ Pc) clobbers(A, Y)  {
+asmsub  scroll_up() clobbers(A, Y)  {
 	; ---- scroll the whole screen 1 character up
 	;      contents of the bottom row are unchanged, you should refill/clear this yourself
-	;      Carry flag is a dummy on the cx16
 	%asm {{
 	    phx
 	    jsr  c64.SCREEN
@@ -300,10 +320,9 @@ _nextline
 	}}
 }
 
-asmsub  scroll_down  (ubyte dummy @ Pc) clobbers(A, Y)  {
+asmsub  scroll_down() clobbers(A, Y)  {
 	; ---- scroll the whole screen 1 character down
 	;      contents of the top row are unchanged, you should refill/clear this yourself
-	;      Carry flag is a dummy on the cx16
 	%asm {{
 	    phx
 	    jsr  c64.SCREEN
@@ -401,7 +420,7 @@ _print_byte_digits
 		jsr  c64.CHROUT
 		pla
 		jsr  c64.CHROUT
-		jmp  _ones
+		bra  _ones
 +       pla
         cmp  #'0'
         beq  _ones
@@ -424,7 +443,7 @@ asmsub  print_b  (byte value @ A) clobbers(A,Y)  {
 		jsr  c64.CHROUT
 +		pla
 		jsr  conv.byte2decimal
-		jmp  print_ub._print_byte_digits
+		bra  print_ub._print_byte_digits
 	}}
 }
 
@@ -475,7 +494,7 @@ asmsub  print_uwbin  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
 		jsr  print_ubbin
 		pla
 		clc
-		jmp  print_ubbin
+		bra  print_ubbin
 	}}
 }
 
@@ -488,7 +507,7 @@ asmsub  print_uwhex  (uword value @ AY, ubyte prefix @ Pc) clobbers(A,Y)  {
 		jsr  print_ubhex
 		pla
 		clc
-		jmp  print_ubhex
+		bra  print_ubhex
 	}}
 }
 
@@ -551,7 +570,7 @@ asmsub  print_w  (word value @ AY) clobbers(A,Y)  {
 		adc  #1
 		bcc  +
 		iny
-+		jmp  print_uw
++		bra  print_uw
 	}}
 }
 
@@ -579,95 +598,126 @@ asmsub  input_chars  (uword buffer @ AY) clobbers(A) -> ubyte @ Y  {
 asmsub  setchr  (ubyte col @X, ubyte row @Y, ubyte character @A) clobbers(A)  {
 	; ---- sets the character in the screen matrix at the given position
 	%asm {{
-		pha
-		txa
-		asl  a
-        stz  cx16.VERA_CTRL
-		stz  cx16.VERA_ADDR_H
-		sta  cx16.VERA_ADDR_L
-		sty  cx16.VERA_ADDR_M
-		pla
-        sta  cx16.VERA_DATA0
-		rts
+            pha
+            txa
+            asl  a
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            pla
+            sta  cx16.VERA_DATA0
+            rts
 	}}
 }
 
 asmsub  getchr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
 	; ---- get the character in the screen matrix at the given location
 	%asm  {{
-		asl  a
-        stz  cx16.VERA_CTRL
-		stz  cx16.VERA_ADDR_H
-		sta  cx16.VERA_ADDR_L
-		sty  cx16.VERA_ADDR_M
-        lda  cx16.VERA_DATA0
-		rts
+            asl  a
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            rts
 	}}
 }
 
 asmsub  setclr  (ubyte col @X, ubyte row @Y, ubyte color @A) clobbers(A)  {
 	; ---- set the color in A on the screen matrix at the given position
+	;      note: on the CommanderX16 this allows you to set both Fg and Bg colors;
+	;            use the high nybble in A to set the Bg color!
 	%asm {{
-		pha
-		txa
-		asl  a
-		ina
-        stz  cx16.VERA_CTRL
-		stz  cx16.VERA_ADDR_H
-		sta  cx16.VERA_ADDR_L
-		sty  cx16.VERA_ADDR_M
-		pla
-        sta  cx16.VERA_DATA0
-		rts
+            pha
+            txa
+            asl  a
+            ina
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            pla
+            sta  cx16.VERA_DATA0
+            rts
 	}}
 }
 
 asmsub  getclr  (ubyte col @A, ubyte row @Y) -> ubyte @ A {
 	; ---- get the color in the screen color matrix at the given location
 	%asm  {{
-		asl  a
-		ina
-        stz  cx16.VERA_CTRL
-		stz  cx16.VERA_ADDR_H
-		sta  cx16.VERA_ADDR_L
-		sty  cx16.VERA_ADDR_M
-        lda  cx16.VERA_DATA0
-		rts
+            asl  a
+            ina
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            sta  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            rts
 	}}
 }
 
 sub  setcc  (ubyte column, ubyte row, ubyte char, ubyte charcolor)  {
 	; ---- set char+color at the given position on the screen
+	;      note: color handling is the same as on the C64: it only sets the foreground color.
+	;            use setcc2 if you want Cx-16 specific feature of setting both Bg+Fg colors.
 	%asm {{
-	    phx
-		lda  column
-		asl  a
-		tax
-		ldy  row
-		lda  charcolor
-		and  #$0f
-	    sta  P8ZP_SCRATCH_B1
-        stz  cx16.VERA_CTRL
-        stz  cx16.VERA_ADDR_H
-        stx  cx16.VERA_ADDR_L
-        sty  cx16.VERA_ADDR_M
-        lda  char
-        sta  cx16.VERA_DATA0
-        inx
-        stz  cx16.VERA_ADDR_H
-        stx  cx16.VERA_ADDR_L
-        sty  cx16.VERA_ADDR_M
-        lda  cx16.VERA_DATA0
-        and  #$f0
-        ora  P8ZP_SCRATCH_B1
-        sta  cx16.VERA_DATA0
-		plx
-		rts
+            phx
+            lda  column
+            asl  a
+            tax
+            ldy  row
+            lda  charcolor
+            and  #$0f
+            sta  P8ZP_SCRATCH_B1
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  char
+            sta  cx16.VERA_DATA0
+            inx
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  cx16.VERA_DATA0
+            and  #$f0
+            ora  P8ZP_SCRATCH_B1
+            sta  cx16.VERA_DATA0
+            plx
+            rts
+    }}
+}
+
+sub  setcc2  (ubyte column, ubyte row, ubyte char, ubyte colors)  {
+	; ---- set char+color at the given position on the screen
+	;      note: on the CommanderX16 this allows you to set both Fg and Bg colors;
+	;            use the high nybble in A to set the Bg color!
+	%asm {{
+            phx
+            lda  column
+            asl  a
+            tax
+            ldy  row
+            stz  cx16.VERA_CTRL
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  char
+            sta  cx16.VERA_DATA0
+            inx
+            stz  cx16.VERA_ADDR_H
+            stx  cx16.VERA_ADDR_L
+            sty  cx16.VERA_ADDR_M
+            lda  colors
+            sta  cx16.VERA_DATA0
+            plx
+            rts
     }}
 }
 
 asmsub  plot  (ubyte col @ Y, ubyte row @ A) clobbers(A) {
-	; ---- safe wrapper around PLOT kernel routine, to save the X register.
+	; ---- safe wrapper around PLOT kernal routine, to save the X register.
 	%asm  {{
 		phx
 		tax

compiler/res/prog8lib/cx16logo.p8

@@ -1,3 +1,5 @@
+; routine to draw the Commander X16's log in petscii.
+
 %import textio
 
 cx16logo {
@@ -14,7 +16,7 @@ cx16logo {
         uword strptr
         for strptr in logo_lines
             txt.print(strptr)
-        txt.chrout('\n')
+        txt.nl()
     }
 
     str[] logo_lines = [

compiler/res/prog8lib/diskio.p8

@@ -1,20 +1,22 @@
-%import textio
-%import syslib
+; C64 and Cx16 disk drive I/O routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
 
-; Note: this code is compatible with C64 and CX16.
+%import textio
+%import string
+%import syslib
 
 diskio {
 
-
-    sub directory(ubyte drivenumber) -> byte {
-        ; -- Shows the directory contents of disk drive 8-11 (provide as argument).
+    sub directory(ubyte drivenumber) -> ubyte {
+        ; -- Prints the directory contents of disk drive 8-11 to the screen. Returns success.
 
         c64.SETNAM(1, "$")
-        c64.SETLFS(1, drivenumber, 0)
-        void c64.OPEN()          ; open 1,8,0,"$"
+        c64.SETLFS(13, drivenumber, 0)
+        void c64.OPEN()          ; open 13,8,0,"$"
         if_cs
             goto io_error
-        void c64.CHKIN(1)        ; use #1 as input channel
+        void c64.CHKIN(13)        ; use #13 as input channel
         if_cs
             goto io_error
 
@@ -25,31 +27,34 @@ diskio {
         ; while not key pressed / EOF encountered, read data.
         ubyte status = c64.READST()
         while not status {
-            txt.print_uw(mkword(c64.CHRIN(), c64.CHRIN()))
-            txt.chrout(' ')
+            ubyte low = c64.CHRIN()
+            ubyte high = c64.CHRIN()
+            txt.print_uw(mkword(high, low))
+            txt.spc()
             ubyte @zp char
-            do {
+            repeat {
                 char = c64.CHRIN()
+                if char==0
+                    break
                 txt.chrout(char)
-            } until char==0
-            txt.chrout('\n')
+            }
+            txt.nl()
             void c64.CHRIN()     ; skip 2 bytes
             void c64.CHRIN()
             status = c64.READST()
-            void c64.STOP()
-            if_nz
+            if c64.STOP2()
                 break
         }
 
 io_error:
         status = c64.READST()
         c64.CLRCHN()        ; restore default i/o devices
-        c64.CLOSE(1)
+        c64.CLOSE(13)
 
-        if status and status != 64 {            ; 64=end of file
+        if status and status & $40 == 0 {            ; bit 6=end of file
             txt.print("\ni/o error, status: ")
             txt.print_ub(status)
-            txt.chrout('\n')
+            txt.nl()
             return false
         }
 
@@ -57,9 +62,284 @@ io_error:
     }
 
 
-    sub status(ubyte drivenumber) {
-        ; -- display the disk drive's current status message
-        c64.SETNAM(0, $0000)
+    ; internal variables for the iterative file lister / loader
+    ubyte list_skip_disk_name
+    uword list_pattern
+    uword list_blocks
+    ubyte iteration_in_progress = false
+    ubyte @zp first_byte
+    ubyte have_first_byte
+    str   list_filename = "?" * 32
+
+
+    ; ----- get a list of files (uses iteration functions internally) -----
+
+    sub list_files(ubyte drivenumber, uword pattern_ptr, uword name_ptrs, ubyte max_names) -> ubyte {
+        ; -- fill the array 'name_ptrs' with (pointers to) the names of the files requested.
+        uword names_buffer = memory("filenames", 512)
+        uword buffer_start = names_buffer
+        ubyte files_found = 0
+        if lf_start_list(drivenumber, pattern_ptr) {
+            while lf_next_entry() {
+                @(name_ptrs) = lsb(names_buffer)
+                name_ptrs++
+                @(name_ptrs) = msb(names_buffer)
+                name_ptrs++
+                names_buffer += string.copy(diskio.list_filename, names_buffer) + 1
+                files_found++
+                if names_buffer - buffer_start > 512-18
+                    break
+                if files_found == max_names
+                    break
+            }
+            lf_end_list()
+        }
+        return files_found
+    }
+
+    ; ----- iterative file lister functions (uses io channel 12) -----
+
+    sub lf_start_list(ubyte drivenumber, uword pattern_ptr) -> ubyte {
+        ; -- start an iterative file listing with optional pattern matching.
+        ;    note: only a single iteration loop can be active at a time!
+        lf_end_list()
+        list_pattern = pattern_ptr
+        list_skip_disk_name = true
+        iteration_in_progress = true
+
+        c64.SETNAM(1, "$")
+        c64.SETLFS(12, drivenumber, 0)
+        void c64.OPEN()          ; open 12,8,0,"$"
+        if_cs
+            goto io_error
+        void c64.CHKIN(12)        ; use #12 as input channel
+        if_cs
+            goto io_error
+
+        repeat 4 {
+            void c64.CHRIN()     ; skip the 4 prologue bytes
+        }
+
+        if c64.READST()==0
+            return true
+
+io_error:
+        lf_end_list()
+        return false
+    }
+
+    sub lf_next_entry() -> ubyte {
+        ; -- retrieve the next entry from an iterative file listing session.
+        ;    results will be found in list_blocks and list_filename.
+        ;    if it returns false though, there are no more entries (or an error occurred).
+
+        if not iteration_in_progress
+            return false
+
+        repeat {
+            void c64.CHKIN(12)        ; use #12 as input channel again
+
+            uword nameptr = &list_filename
+            ubyte blocks_lsb = c64.CHRIN()
+            ubyte blocks_msb = c64.CHRIN()
+
+            if c64.READST()
+                goto close_end
+
+            list_blocks = mkword(blocks_msb, blocks_lsb)
+
+            ; read until the filename starts after the first "
+            while c64.CHRIN()!='\"'  {
+                if c64.READST()
+                    goto close_end
+            }
+
+            ; read the filename
+            repeat {
+                ubyte char = c64.CHRIN()
+                if char==0
+                    break
+                if char=='\"'
+                    break
+                @(nameptr) = char
+                nameptr++
+            }
+
+            @(nameptr) = 0
+
+            while c64.CHRIN() {
+                ; read the rest of the entry until the end
+            }
+
+            void c64.CHRIN()     ; skip 2 bytes
+            void c64.CHRIN()
+
+            if not list_skip_disk_name {
+                if not list_pattern
+                    return true
+                if prog8_lib.pattern_match(list_filename, list_pattern)
+                    return true
+            }
+            list_skip_disk_name = false
+        }
+
+close_end:
+        lf_end_list()
+        return false
+    }
+
+    sub lf_end_list() {
+        ; -- end an iterative file listing session (close channels).
+        if iteration_in_progress {
+            c64.CLRCHN()
+            c64.CLOSE(12)
+            iteration_in_progress = false
+        }
+    }
+
+
+    ; ----- iterative file loader functions (uses io channel 11) -----
+
+    sub f_open(ubyte drivenumber, uword filenameptr) -> ubyte {
+        ; -- open a file for iterative reading with f_read
+        ;    note: only a single iteration loop can be active at a time!
+        f_close()
+
+        c64.SETNAM(string.length(filenameptr), filenameptr)
+        c64.SETLFS(11, drivenumber, 0)
+        void c64.OPEN()          ; open 11,8,0,"filename"
+        if_cc {
+            iteration_in_progress = true
+            have_first_byte = false
+            void c64.CHKIN(11)        ; use #11 as input channel
+            if_cc {
+                first_byte = c64.CHRIN()   ; read first byte to test for file not found
+                if not c64.READST() {
+                    have_first_byte = true
+                    return true
+                }
+            }
+        }
+        f_close()
+        return false
+    }
+
+    sub f_read(uword bufferpointer, uword num_bytes) -> uword {
+        ; -- read from the currently open file, up to the given number of bytes.
+        ;    returns the actual number of bytes read.  (checks for End-of-file and error conditions)
+        if not iteration_in_progress or not num_bytes
+            return 0
+
+        list_blocks = 0     ; we reuse this variable for the total number of bytes read
+        if have_first_byte {
+            have_first_byte=false
+            @(bufferpointer) = first_byte
+            bufferpointer++
+            list_blocks++
+            num_bytes--
+        }
+
+        void c64.CHKIN(11)        ; use #11 as input channel again
+        %asm {{
+            lda  bufferpointer
+            sta  _in_buffer+1
+            lda  bufferpointer+1
+            sta  _in_buffer+2
+        }}
+        repeat num_bytes {
+            %asm {{
+                jsr  c64.CHRIN
+                sta  cx16.r5
+_in_buffer      sta  $ffff
+                inc  _in_buffer+1
+                bne  +
+                inc  _in_buffer+2
++               inc  list_blocks
+                bne  +
+                inc  list_blocks+1
++
+            }}
+
+            if cx16.r5==$0d {   ; chance on I/o error status?
+                first_byte = c64.READST()
+                if first_byte & $40
+                    f_close()       ; end of file, close it
+                if first_byte
+                    return list_blocks
+            }
+        }
+        return list_blocks
+    }
+
+    sub f_read_all(uword bufferpointer) -> uword {
+        ; -- read the full contents of the file, returns number of bytes read.
+        if not iteration_in_progress
+            return 0
+
+        list_blocks = 0     ; we reuse this variable for the total number of bytes read
+        if have_first_byte {
+            have_first_byte=false
+            @(bufferpointer) = first_byte
+            bufferpointer++
+            list_blocks++
+        }
+
+        while not c64.READST() {
+            list_blocks += f_read(bufferpointer, 256)
+            bufferpointer += 256
+        }
+        return list_blocks
+    }
+
+    asmsub f_readline(uword bufptr @AY) clobbers(X) -> ubyte @Y {
+        ; Routine to read text lines from a text file. Lines must be less than 255 characters.
+        ; Reads characters from the input file UNTIL a newline or return character (or EOF).
+        ; The line read will be 0-terminated in the buffer (and not contain the end of line character).
+        ; The length of the line is returned in Y. Note that an empty line is okay and is length 0!
+        ; I/O error status should be checked by the caller itself via READST() routine.
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldx  #11
+            jsr  c64.CHKIN              ; use channel 11 again for input
+            ldy  #0
+            lda  have_first_byte
+            beq  _loop
+            lda  #0
+            sta  have_first_byte
+            lda  first_byte
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+_loop       jsr  c64.CHRIN
+            sta  (P8ZP_SCRATCH_W1),y
+            beq  _end
+            iny
+            cmp  #$0a
+            beq  _line_end
+            cmp  #$0d
+            bne  _loop
+_line_end   dey     ; get rid of the trailing end-of-line char
+            lda  #0
+            sta  (P8ZP_SCRATCH_W1),y
+_end        rts
+        }}
+    }
+
+
+    sub f_close() {
+        ; -- end an iterative file loading session (close channels).
+        if iteration_in_progress {
+            c64.CLRCHN()
+            c64.CLOSE(11)
+            iteration_in_progress = false
+        }
+    }
+
+
+    sub status(ubyte drivenumber) -> uword {
+        ; -- retrieve the disk drive's current status message
+        uword messageptr = &filename
+        c64.SETNAM(0, filename)
         c64.SETLFS(15, drivenumber, 15)
         void c64.OPEN()          ; open 15,8,15
         if_cs
@@ -68,17 +348,21 @@ io_error:
         if_cs
             goto io_error
 
-        while not c64.READST()
-            txt.chrout(c64.CHRIN())
+        while not c64.READST() {
+            @(messageptr) = c64.CHRIN()
+            messageptr++
+        }
 
 io_error:
+        @(messageptr) = 0
         c64.CLRCHN()        ; restore default i/o devices
         c64.CLOSE(15)
+        return filename
     }
 
 
-    sub save(ubyte drivenumber, uword filenameptr, uword address, uword size) -> byte {
-        c64.SETNAM(strlen(filenameptr), filenameptr)
+    sub save(ubyte drivenumber, uword filenameptr, uword address, uword size) -> ubyte {
+        c64.SETNAM(string.length(filenameptr), filenameptr)
         c64.SETLFS(1, drivenumber, 0)
         uword end_address = address + size
 
@@ -97,14 +381,18 @@ io_error:
             plp
         }}
 
+        first_byte = 0      ; result var reuse
         if_cc
-            return c64.READST()==0
+            first_byte = c64.READST()==0
 
-        return false
+        c64.CLRCHN()
+        c64.CLOSE(1)
+
+        return first_byte
     }
 
     sub load(ubyte drivenumber, uword filenameptr, uword address_override) -> uword {
-        c64.SETNAM(strlen(filenameptr), filenameptr)
+        c64.SETNAM(string.length(filenameptr), filenameptr)
         ubyte secondary = 1
         uword end_of_load = 0
         if address_override
@@ -122,6 +410,9 @@ io_error:
 +           ldx  P8ZP_SCRATCH_REG
         }}
 
+        c64.CLRCHN()
+        c64.CLOSE(1)
+
         if end_of_load
             return end_of_load - address_override
 
@@ -133,10 +424,9 @@ io_error:
 
     sub delete(ubyte drivenumber, uword filenameptr) {
         ; -- delete a file on the drive
-        ubyte flen = strlen(filenameptr)
         filename[0] = 's'
         filename[1] = ':'
-        memcopy(filenameptr, &filename+2, flen+1)
+        ubyte flen = string.copy(filenameptr, &filename+2)
         c64.SETNAM(flen+2, filename)
         c64.SETLFS(1, drivenumber, 15)
         void c64.OPEN()
@@ -146,13 +436,11 @@ io_error:
 
     sub rename(ubyte drivenumber, uword oldfileptr, uword newfileptr) {
         ; -- rename a file on the drive
-        ubyte flen_old = strlen(oldfileptr)
-        ubyte flen_new = strlen(newfileptr)
         filename[0] = 'r'
         filename[1] = ':'
-        memcopy(newfileptr, &filename+2, flen_new)
+        ubyte flen_new = string.copy(newfileptr, &filename+2)
         filename[flen_new+2] = '='
-        memcopy(oldfileptr, &filename+3+flen_new, flen_old+1)
+        ubyte flen_old = string.copy(oldfileptr, &filename+3+flen_new)
         c64.SETNAM(3+flen_new+flen_old, filename)
         c64.SETLFS(1, drivenumber, 15)
         void c64.OPEN()

compiler/res/prog8lib/math.asm

@@ -1,11 +1,8 @@
-; Prog8 internal Math library routines - always included by the compiler
+; Internal Math library routines - always included by the compiler
 ; Generic machine independent 6502 code.
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
 ;
-; indent format: TABS, size=8
-
-
 ;  some more interesting routines can be found here:
 ;	http://6502org.wikidot.com/software-math
 ;	http://codebase64.org/doku.php?id=base:6502_6510_maths
@@ -247,8 +244,8 @@ randseed	.proc
 		.pend
 
 
-randbyte	.proc
-	; -- 8-bit pseudo random number generator into A
+fast_randbyte	.proc
+	; -- fast but bad 8-bit pseudo random number generator into A
 		lda  _seed
 		beq  _eor
 		asl  a
@@ -266,6 +263,10 @@ _seed		.byte  $3a
 
 		.pend
 
+randbyte        .proc
+	; -- 8 bit pseudo random number generator into A (by just reusing randword)
+		jmp  randword
+		.pend
 
 randword	.proc
 	; -- 16 bit pseudo random number generator into AY
@@ -774,6 +775,31 @@ stack_mul_word_100	.proc
 		rts
 		.pend
 
+stack_mul_word_320	.proc
+		; stackW = stackLo * 256 + stackLo * 64	 (stackHi doesn't matter)
+		ldy  P8ESTACK_LO+1,x
+		lda  #0
+		sta  P8ESTACK_HI+1,x
+		tya
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		asl  a
+		rol  P8ESTACK_HI+1,x
+		sta  P8ESTACK_LO+1,x
+		tya
+		clc
+		adc  P8ESTACK_HI+1,x
+		sta  P8ESTACK_HI+1,x
+		rts
+		.pend
 
 ; ----------- optimized multiplications (in-place A (byte) and ?? (word)) : ---------
 mul_byte_3	.proc
@@ -1178,8 +1204,6 @@ mul_word_40	.proc
 		rol  a
 		asl  P8ZP_SCRATCH_W1
 		rol  a
-		asl  P8ZP_SCRATCH_W1
-		rol  a
 		tay
 		lda  P8ZP_SCRATCH_W1
 		rts
@@ -1241,6 +1265,32 @@ mul_word_100	.proc
 		rts
 		.pend
 
+mul_word_320	.proc
+		; AY = A * 256 + A * 64	 (msb in Y doesn't matter)
+		sta  P8ZP_SCRATCH_B1
+		ldy  #0
+		sty  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		asl  a
+		rol  P8ZP_SCRATCH_REG
+		pha
+		clc
+		lda  P8ZP_SCRATCH_B1
+		adc  P8ZP_SCRATCH_REG
+		tay
+		pla
+		rts
+		.pend
+
 ; ----------- end optimized multiplications -----------
 
 
@@ -1301,6 +1351,33 @@ shift_left_w_3	.proc
 		jmp  shift_left_w_7._shift3
 		.pend
 
+
+shift_left_w	.proc
+		; -- variable number of shifts left
+		inx
+		ldy  P8ESTACK_LO,x
+		bne  _shift
+		rts
+_shift		asl  P8ESTACK_LO+1,x
+		rol  P8ESTACK_HI+1,x
+		dey
+		bne  _shift
+		rts
+		.pend
+
+shift_right_uw	.proc
+		; -- uword variable number of shifts right
+		inx
+		ldy  P8ESTACK_LO,x
+		bne  _shift
+		rts
+_shift		lsr  P8ESTACK_HI+1,x
+		ror  P8ESTACK_LO+1,x
+		dey
+		bne  _shift
+		rts
+		.pend
+
 shift_right_uw_7	.proc
 		lda  P8ESTACK_LO+1,x
 		sta  P8ZP_SCRATCH_B1
@@ -1431,6 +1508,21 @@ shift_right_w_3	.proc
 		.pend
 
 
+shift_right_w	.proc
+		; -- signed word variable number of shifts right
+		inx
+		ldy  P8ESTACK_LO,x
+		bne  _shift
+		rts
+_shift		lda  P8ESTACK_HI+1,x
+		asl  a
+		ror  P8ESTACK_HI+1,x
+		ror  P8ESTACK_LO+1,x
+		dey
+		bne  _shift
+		rts
+		.pend
+
 
 ; support for bit shifting that is too large to be unrolled:
 
@@ -1449,3 +1541,71 @@ _negative	lsr  a
 		rts
 		.pend
 
+
+square          .proc
+; -- calculate square root of signed word in AY, result in AY
+; routine by Lee Davsion, source: http://6502.org/source/integers/square.htm
+; using this routine is about twice as fast as doing a regular multiplication.
+;
+; Calculates the 16 bit unsigned integer square of the signed 16 bit integer in
+; Numberl/Numberh.  The result is always in the range 0 to 65025 and is held in
+; Squarel/Squareh
+;
+; The maximum input range is only +/-255 and no checking is done to ensure that
+; this is so.
+;
+; This routine is useful if you are trying to draw circles as for any circle
+;
+; x^2+y^2=r^2 where x and y are the co-ordinates of any point on the circle and
+; r is the circle radius
+
+numberl = P8ZP_SCRATCH_W1       ; number to square low byte
+numberh = P8ZP_SCRATCH_W1+1     ; number to square high byte
+squarel = P8ZP_SCRATCH_W2       ; square low byte
+squareh = P8ZP_SCRATCH_W2+1     ; square high byte
+tempsq = P8ZP_SCRATCH_B1        ; temp byte for intermediate result
+
+	sta  numberl
+	sty  numberh
+	stx  P8ZP_SCRATCH_REG
+
+        lda     #$00        ; clear a
+        sta     squarel     ; clear square low byte
+                            ; (no need to clear the high byte, it gets shifted out)
+        lda	numberl     ; get number low byte
+	ldx	numberh     ; get number high  byte
+	bpl	_nonneg      ; if +ve don't negate it
+                            ; else do a two's complement
+	eor	#$ff        ; invert
+        sec	            ; +1
+	adc	#$00        ; and add it
+
+_nonneg:
+	sta	tempsq      ; save abs(number)
+	ldx	#$08        ; set bit count
+
+_nextr2bit:
+	asl	squarel     ; low byte *2
+	rol	squareh     ; high byte *2+carry from low
+	asl	a           ; shift number byte
+	bcc	_nosqadd     ; don't do add if c = 0
+	tay                 ; save a
+	clc                 ; clear carry for add
+	lda	tempsq      ; get number
+	adc	squarel     ; add number^2 low byte
+	sta	squarel     ; save number^2 low byte
+	lda	#$00        ; clear a
+	adc	squareh     ; add number^2 high byte
+	sta	squareh     ; save number^2 high byte
+	tya                 ; get a back
+
+_nosqadd:
+	dex                 ; decrement bit count
+	bne	_nextr2bit   ; go do next bit
+
+	lda  squarel
+	ldy  squareh
+	ldx  P8ZP_SCRATCH_REG
+	rts
+
+		.pend

compiler/res/prog8lib/math.p8

@@ -1,8 +1,6 @@
-; Prog8 internal Math library routines - always included by the compiler
+; Internal Math library routines - always included by the compiler
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
 
 math {
 	%asminclude "library:math.asm", ""

compiler/res/prog8lib/prog8_funcs.asm

@@ -387,6 +387,14 @@ func_sqrt16_into_A	.proc
 		rts
 		.pend
 
+func_fastrnd8_stack	.proc
+	; -- put a random ubyte on the estack (using fast but bad RNG)
+		jsr  math.fast_randbyte
+		sta  P8ESTACK_LO,x
+		dex
+		rts
+		.pend
+
 func_rnd_stack	.proc
 	; -- put a random ubyte on the estack
 		jsr  math.randbyte
@@ -432,6 +440,7 @@ func_min_ub_stack	.proc
 func_min_b_into_A	.proc
 		; -- min(barray) -> A.  (array in P8ZP_SCRATCH_W1, num elements in A)
 		tay
+		dey
 		lda  #127
 		sta  P8ZP_SCRATCH_B1
 -		lda  (P8ZP_SCRATCH_W1),y
@@ -548,6 +557,7 @@ func_min_w_stack	.proc
 func_max_ub_into_A	.proc
 		; -- max(ubarray) -> A  (array in P8ZP_SCRATCH_W1, num elements in A)
 		tay
+		dey
 		lda  #0
 		sta  P8ZP_SCRATCH_B1
 -		lda  (P8ZP_SCRATCH_W1),y
@@ -1078,271 +1088,28 @@ _loop_hi	ldy  _index_first
 		.pend
 
 
-func_exit	.proc
-		; -- immediately exit the program with a return code in the A register
-		ldx  orig_stackpointer
-		txs
-		rts		; return to original caller
-		.pend
+func_peekw   .proc
+	; -- read the word value on the address in AY
+	sta  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	ldy  #0
+	lda  (P8ZP_SCRATCH_W1),y
+	pha
+	iny
+	lda  (P8ZP_SCRATCH_W1),y
+	tay
+	pla
+	rts
+	.pend
 
 
-func_read_flags_stack	.proc
-		; -- put the processor status register on the stack
-		php
-		pla
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-
-func_memset	.proc
-	; note: clobbers A,Y
-		txa
-		pha
-		lda  _arg_address
-		sta  P8ZP_SCRATCH_W1
-		lda  _arg_address+1
-		sta  P8ZP_SCRATCH_W1+1
-		ldx  _arg_numbytes
-		ldy  _arg_numbytes+1
-		lda  _arg_bytevalue
-		jsr  memset
-		pla
-		tax
-		rts
-_arg_address	.word  0
-_arg_numbytes	.word  0
-_arg_bytevalue	.byte  0
-		.pend
-
-
-func_memsetw	.proc
-	; note: clobbers A,Y
-		txa
-		pha
-		lda  _arg_address
-		sta  P8ZP_SCRATCH_W1
-		lda  _arg_address+1
-		sta  P8ZP_SCRATCH_W1+1
-		lda  _arg_numwords
-		sta  P8ZP_SCRATCH_W2
-		lda  _arg_numwords+1
-		sta  P8ZP_SCRATCH_W2+1
-		lda  _arg_wordvalue
-		ldy  _arg_wordvalue+1
-		jsr  memsetw
-		pla
-		tax
-		rts
-_arg_address	.word  0
-_arg_numwords	.word  0
-_arg_wordvalue	.word  0
-		.pend
-
-
-func_memcopy	.proc
-	; memcopy of any number of bytes, note: clobbers A,Y
-		stx  P8ZP_SCRATCH_REG
-		lda  _arg_from
-		sta  P8ZP_SCRATCH_W1
-		lda  _arg_from+1
-		sta  P8ZP_SCRATCH_W1+1
-		lda  _arg_to
-		sta  P8ZP_SCRATCH_W2
-		lda  _arg_to+1
-		sta  P8ZP_SCRATCH_W2+1
-
-		ldy  #0
-		ldx  _arg_numbytes+1
-		beq  _remain
--		lda  (P8ZP_SCRATCH_W1),y	; move a page at a time
-		sta  (P8ZP_SCRATCH_W2),y
-		iny
-		bne  -
-		inc  P8ZP_SCRATCH_W1+1
-		inc  P8ZP_SCRATCH_W2+1
-		dex
-		bne  -
-_remain		ldx  _arg_numbytes
-		beq  _done
--		lda (P8ZP_SCRATCH_W1),y		; move the remaining bytes
-		sta (P8ZP_SCRATCH_W2),y
-		iny
-		dex
-		bne  -
-
-_done		ldx  P8ZP_SCRATCH_REG
-		rts
-
-_arg_from	.word  0
-_arg_to		.word  0
-_arg_numbytes	.word  0
-		.pend
-
-
-func_memcopy255	.proc
-	; fast memcopy of up to 255 bytes, note: clobbers A,Y
-	;  note: also uses the _arg variables from regular func_memcopy
-		stx  P8ZP_SCRATCH_REG
-		lda  func_memcopy._arg_from
-		sta  P8ZP_SCRATCH_W1
-		lda  func_memcopy._arg_from+1
-		sta  P8ZP_SCRATCH_W1+1
-		lda  func_memcopy._arg_to
-		sta  P8ZP_SCRATCH_W2
-		lda  func_memcopy._arg_to+1
-		sta  P8ZP_SCRATCH_W2+1
-		ldx  func_memcopy._arg_numbytes
-		ldy  #0
--		lda  (P8ZP_SCRATCH_W1), y
-		sta  (P8ZP_SCRATCH_W2), y
-		iny
-		dex
-		bne  -
-		ldx  P8ZP_SCRATCH_REG
-		rts
-		.pend
-
-
-func_leftstr	.proc
-		; leftstr(source, target, length)
-		lda  _arg_source
-		ldy  _arg_source+1
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		lda  _arg_target
-		ldy  _arg_target+1
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
-		ldy  _arg_length
-		lda  #0
-		sta  (P8ZP_SCRATCH_W2),y
-		cpy  #0
-		bne  _loop
-		rts
-_loop		dey
-		lda  (P8ZP_SCRATCH_W1),y
-		sta  (P8ZP_SCRATCH_W2),y
-		cpy  #0
-		bne  _loop
-+		rts
-
-_arg_source	.word  0
-_arg_target	.word  0
-_arg_length	.byte  0
-		.pend
-
-
-func_rightstr	.proc
-		; rightstr(source, target, length)
-		lda  _arg_source
-		ldy  _arg_source+1
-		jsr  func_strlen_into_A
-		sec
-		sbc  _arg_length
-		sta  P8ZP_SCRATCH_B1
-		lda  _arg_source
-		ldy  _arg_source+1
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		lda  _arg_target
-		ldy  _arg_target+1
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
-		ldy  #0
-		sty  P8ZP_SCRATCH_REG
--		ldy  P8ZP_SCRATCH_B1
-		lda  (P8ZP_SCRATCH_W1),y
-		inc  P8ZP_SCRATCH_B1
-		ldy  P8ZP_SCRATCH_REG
-		sta  (P8ZP_SCRATCH_W2),y
-		inc  P8ZP_SCRATCH_REG
-		cmp  #0
-		bne  -
-		rts
-
-_arg_source	.word  0
-_arg_target	.word  0
-_arg_length	.byte  0
-		.pend
-
-
-func_strlen_into_A		.proc
-	; -- put length of 0-terminated string in A/Y into A
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		ldy  #0
--		lda  (P8ZP_SCRATCH_W1),y
-		beq  +
-		iny
-		bne  -
-+		tya
-		rts
-		.pend
-
-func_strlen_stack	.proc
-		jsr  func_strlen_into_A
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
-
-
-func_substr	.proc
-		; substr(source, target, start, length)
-		lda  _arg_source
-		ldy  _arg_source+1
-		sta  P8ZP_SCRATCH_W1
-		sty  P8ZP_SCRATCH_W1+1
-		lda  _arg_target
-		ldy  _arg_target+1
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
-		ldy  _arg_length
-		lda  _arg_start
-		sta  P8ZP_SCRATCH_B1
-
-		; adjust src location
-		clc
-		lda  P8ZP_SCRATCH_W1
-		adc  P8ZP_SCRATCH_B1
-		sta  P8ZP_SCRATCH_W1
-		bcc  +
-		inc  P8ZP_SCRATCH_W1+1
-+		lda  #0
-		sta  (P8ZP_SCRATCH_W2),y
-		jmp  _startloop
--		lda  (P8ZP_SCRATCH_W1),y
-		sta  (P8ZP_SCRATCH_W2),y
-_startloop	dey
-		cpy  #$ff
-		bne  -
-		rts
-
-_arg_source	.word  0
-_arg_target	.word  0
-_arg_start	.byte  0
-_arg_length	.byte  0
-		.pend
-
-
-func_strcmp	.proc
-		; -- compare 2 strings -> A
-		lda  _arg_s2
-		ldy  _arg_s2+1
-		sta  P8ZP_SCRATCH_W2
-		sty  P8ZP_SCRATCH_W2+1
-		lda  _arg_s1
-		ldy  _arg_s1+1
-		jmp  strcmp_mem
-_arg_s1		.word  0
-_arg_s2		.word  0
-		.pend
-
-func_strcmp_stack	.proc
-		jsr	func_strcmp
-		sta  P8ESTACK_LO,x
-		dex
-		rts
-		.pend
+func_pokew   .proc
+	; -- store the word value in AY in the address in P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_REG
+	ldy  #0
+	sta  (P8ZP_SCRATCH_W1),y
+	iny
+	lda  P8ZP_SCRATCH_REG
+	sta  (P8ZP_SCRATCH_W1),y
+	rts
+	.pend

compiler/res/prog8lib/prog8_lib.asm

@@ -1,9 +1,7 @@
-; Prog8 internal library routines - always included by the compiler
+; Internal library routines - always included by the compiler
 ; Generic machine independent 6502 code.
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
 
 
 read_byte_from_address_on_stack	.proc
@@ -470,27 +468,19 @@ less_uw		.proc
 
 reg_less_w	.proc
 		; -- AY < P8ZP_SCRATCH_W2?
-		sta  P8ZP_SCRATCH_B1
+		cmp  P8ZP_SCRATCH_W2
 		tya
-		sec
 		sbc  P8ZP_SCRATCH_W2+1
 		bvc  +
 		eor  #$80
 +		bmi  _true
-		bvc  +
-		eor  #$80
-+		bne  _false
-		lda  P8ZP_SCRATCH_B1
-		sbc  P8ZP_SCRATCH_W2
-		bcs  _false
-_true		lda  #1
+		lda  #0
 		rts
-_false		lda  #0
+_true		lda  #1
 		rts
 		.pend
 
 less_w		.proc
-		; TODO is this word comparison < correct? reg_less_w is a lot larger...
 		lda  P8ESTACK_LO+2,x
 		cmp  P8ESTACK_LO+1,x
 		lda  P8ESTACK_HI+2,x
@@ -549,7 +539,6 @@ _true		lda  #1
 		.pend
 
 lesseq_uw	.proc
-		; TODO is this comparison uword <= correct????
 		lda  P8ESTACK_HI+1,x
 		cmp  P8ESTACK_HI+2,x
 		bcc  equal_b._equal_b_false
@@ -561,29 +550,20 @@ lesseq_uw	.proc
 		.pend
 
 reg_lesseq_w	.proc
-		; -- AY <= P8ZP_SCRATCH_W2?
-		sta  P8ZP_SCRATCH_B1
+		; -- P8ZP_SCRATCH_W2 <= AY ?   (note: order different from other routines)
+		cmp  P8ZP_SCRATCH_W2
 		tya
-		sec
 		sbc  P8ZP_SCRATCH_W2+1
 		bvc  +
 		eor  #$80
-+		bmi  _true
-		bvc  +
-		eor  #$80
-+		bne  _false
-		lda  P8ZP_SCRATCH_B1
-		sbc  P8ZP_SCRATCH_W2
-		beq  _true			; TODO is this correct word compare <= for all cases?
-		bcs  _false
-_true		lda  #1
++		bpl  +
+		lda  #0
 		rts
-_false		lda  #0
++		lda  #1
 		rts
 		.pend
 
 lesseq_w	.proc
-		; TODO is this word compare <= correct? it is different from reg_lesseq_w
 		lda  P8ESTACK_LO+1,x
 		cmp  P8ESTACK_LO+2,x
 		lda  P8ESTACK_HI+1,x
@@ -671,8 +651,8 @@ greatereq_w	.proc
 		sbc  P8ESTACK_HI+1,x
 		bvc  +
 		eor  #$80
-+		bpl  equal_b._equal_b_true
-		bmi  equal_b._equal_b_false
++		bmi  equal_b._equal_b_false
+		bpl  equal_b._equal_b_true
 		.pend
 
 
@@ -1011,6 +991,7 @@ _arg_index	.byte  0
 strcpy		.proc
 		; copy a string (must be 0-terminated) from A/Y to (P8ZP_SCRATCH_W1)
 		; it is assumed the target string is large enough.
+		; returns the length of the string that was copied in Y.
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1
 		ldy  #$ff
@@ -1038,30 +1019,67 @@ _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.
-            sta  P8ZP_SCRATCH_W1
-            sty  P8ZP_SCRATCH_W1+1
-_loop       ldy  #0
-            lda  (P8ZP_SCRATCH_W1),y
-            bne  +
-            lda  (P8ZP_SCRATCH_W2),y
-            bne  _return_minusone
-            beq  _return
-+           lda  (P8ZP_SCRATCH_W2),y
-            sec
-            sbc  (P8ZP_SCRATCH_W1),y
-            bmi  _return_one
-            bne  _return_minusone
-            inc  P8ZP_SCRATCH_W1
-            bne  +
-            inc  P8ZP_SCRATCH_W1+1
-+           inc  P8ZP_SCRATCH_W2
-            bne  _loop
-            inc  P8ZP_SCRATCH_W2+1
-            bne  _loop
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+_loop		ldy  #0
+		lda  (P8ZP_SCRATCH_W1),y
+		bne  +
+		lda  (P8ZP_SCRATCH_W2),y
+		bne  _return_minusone
+		beq  _return
++		lda  (P8ZP_SCRATCH_W2),y
+		sec
+		sbc  (P8ZP_SCRATCH_W1),y
+		bmi  _return_one
+		bne  _return_minusone
+		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
-            lda  #1
-_return     rts
+		lda  #1
+_return		rts
 _return_minusone
-            lda  #-1
-            rts
-            .pend
+		lda  #-1
+		rts
+		.pend
+
+
+sign_extend_stack_byte	.proc
+	; -- sign extend the (signed) byte on the stack to full 16 bits
+		lda  P8ESTACK_LO+1,x
+		ora  #$7f
+		bmi  +
+		lda  #0
++		sta  P8ESTACK_HI+1,x
+		rts
+		.pend
+
+
+sign_extend_AY_byte	.proc
+	; -- sign extend the (signed) byte in AY to full 16 bits
+		pha
+		and  #$80
+		beq  +
+		ldy  #$ff
+		pla
+		rts
++		ldy  #0
+		pla
+		rts
+		.pend
+
+strlen          .proc
+        ; -- returns the number of bytes in the string in AY, in Y.
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  +
+		iny
+		bne  -
++		rts
+		.pend

compiler/res/prog8lib/prog8_lib.p8

@@ -1,10 +1,84 @@
-; Prog8 internal library routines - always included by the compiler
+; Internal library routines - always included by the compiler
 ;
 ; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
-;
-; indent format: TABS, size=8
 
 prog8_lib {
 	%asminclude "library:prog8_lib.asm", ""
 	%asminclude "library:prog8_funcs.asm", ""
+
+	uword @zp retval_interm_uw      ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	word @zp retval_interm_w        ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	ubyte @zp retval_interm_ub      ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+	byte @zp retval_interm_b        ; to store intermediary expression results for return values (hopefully allocated on ZP to reduce code size)
+
+	asmsub pattern_match(str string @AY, str pattern @R0) clobbers(Y) -> ubyte @A {
+		%asm {{
+; pattern matching of a string.
+; Input:  cx16.r0:  A NUL-terminated, <255-length pattern
+;              AY:  A NUL-terminated, <255-length string
+;
+; Output: A = 1 if the string matches the pattern, A = 0 if not.
+;
+; Notes:  Clobbers A, X, Y. Each * in the pattern uses 4 bytes of stack.
+;
+; see http://6502.org/source/strings/patmatch.htm
+
+str = P8ZP_SCRATCH_W1
+
+	stx  P8ZP_SCRATCH_REG
+	sta  str
+	sty  str+1
+	lda  cx16.r0
+	sta  modify_pattern1+1
+	sta  modify_pattern2+1
+	lda  cx16.r0+1
+	sta  modify_pattern1+2
+	sta  modify_pattern2+2
+	jsr  _match
+	lda  #0
+	adc  #0
+	ldx  P8ZP_SCRATCH_REG
+	rts
+
+
+_match
+	ldx #$00        ; x is an index in the pattern
+	ldy #$ff        ; y is an index in the string
+modify_pattern1
+next    lda $ffff,x   ; look at next pattern character    MODIFIED
+	cmp #'*'     ; is it a star?
+	beq star        ; yes, do the complicated stuff
+	iny             ; no, let's look at the string
+	cmp #'?'     ; is the pattern caracter a ques?
+	bne reg         ; no, it's a regular character
+	lda (str),y     ; yes, so it will match anything
+	beq fail        ;  except the end of string
+reg     cmp (str),y     ; are both characters the same?
+	bne fail        ; no, so no match
+	inx             ; yes, keep checking
+	cmp #0          ; are we at end of string?
+	bne next        ; not yet, loop
+found   rts             ; success, return with c=1
+
+star    inx             ; skip star in pattern
+modify_pattern2
+	cmp $ffff,x   	; string of stars equals one star	MODIFIED
+	beq star        ;  so skip them also
+stloop  txa             ; we first try to match with * = ""
+	pha             ;  and grow it by 1 character every
+	tya             ;  time we loop
+	pha             ; save x and y on stack
+	jsr next        ; recursive call
+	pla             ; restore x and y
+	tay
+	pla
+	tax
+	bcs found       ; we found a match, return with c=1
+	iny             ; no match yet, try to grow * string
+	lda (str),y     ; are we at the end of string?
+	bne stloop      ; not yet, add a character
+fail    clc             ; yes, no match found, return with c=0
+	rts
+		}}
+	}
 }

compiler/res/prog8lib/string.p8

@@ -0,0 +1,235 @@
+; 0-terminated string manipulation routines.
+;
+; Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
+
+
+string {
+
+    asmsub length(uword string @AY) clobbers(A) -> ubyte @Y {
+        ; Returns the number of bytes in the string.
+        ; This value is determined during runtime and counts upto the first terminating 0 byte in the string,
+        ; regardless of the size of the string during compilation time. Don’t confuse this with len and sizeof!
+
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  +
+		iny
+		bne  -
++		rts
+        }}
+    }
+
+    asmsub left(uword source @R0, ubyte length @A, uword target @R1) clobbers(A, Y) {
+        ; Copies the left side of the source string of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that length is smaller or equal to the length of the source string.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+		ldy  cx16.r0
+		sty  P8ZP_SCRATCH_W1
+		ldy  cx16.r0+1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  cx16.r1
+		sty  P8ZP_SCRATCH_W2
+		ldy  cx16.r1+1
+		sty  P8ZP_SCRATCH_W2+1
+		tay
+		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
+		rts
+_loop		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
++		rts
+        }}
+;                asmgen.out("  jsr  prog8_lib.func_leftstr")
+    }
+
+    asmsub right(uword source @R0, ubyte length @A, uword target @R1) clobbers(A,Y) {
+        ; Copies the right side of the source string of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that length is smaller or equal to the length of the source string.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+                sta  P8ZP_SCRATCH_B1
+                lda  cx16.r0
+                ldy  cx16.r0+1
+                jsr  string.length
+                tya
+                sec
+                sbc  P8ZP_SCRATCH_B1
+                clc
+                adc  cx16.r0
+		sta  P8ZP_SCRATCH_W1
+		lda  cx16.r0+1
+		adc  #0
+		sta  P8ZP_SCRATCH_W1+1
+		ldy  cx16.r1
+		sty  P8ZP_SCRATCH_W2
+		ldy  cx16.r1+1
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  P8ZP_SCRATCH_B1
+		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
+		rts
+_loop		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		cpy  #0
+		bne  _loop
++		rts
+        }}
+    }
+
+    asmsub slice(uword source @R0, ubyte start @A, ubyte length @Y, uword target @R1) clobbers(A, Y) {
+        ; Copies a segment from the source string, starting at the given index,
+        ;  and of the given length to target string.
+        ; It is assumed the target string buffer is large enough to contain the result.
+        ; Also, you have to make sure yourself that start and length are within bounds of the strings.
+        ; Modifies in-place, doesn’t return a value (so can’t be used in an expression).
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+		; substr(source, target, start, length)
+		sta  P8ZP_SCRATCH_B1
+		lda  cx16.r0
+		sta  P8ZP_SCRATCH_W1
+		lda  cx16.r0+1
+		sta  P8ZP_SCRATCH_W1+1
+		lda  cx16.r1
+		sta  P8ZP_SCRATCH_W2
+		lda  cx16.r1+1
+		sta  P8ZP_SCRATCH_W2+1
+
+		; adjust src location
+		clc
+		lda  P8ZP_SCRATCH_W1
+		adc  P8ZP_SCRATCH_B1
+		sta  P8ZP_SCRATCH_W1
+		bcc  +
+		inc  P8ZP_SCRATCH_W1+1
++		lda  #0
+		sta  (P8ZP_SCRATCH_W2),y
+		beq  _startloop
+-		lda  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+_startloop	dey
+		cpy  #$ff
+		bne  -
+		rts
+        }}
+    }
+
+    asmsub find(uword string @R0, ubyte character @A) -> uword @AY {
+        ; Locates the first position of the given character in the string,
+        ;  returns the string starting with this character or $0000 if the character is not found.
+        %asm {{
+                ; need to copy the the cx16 virtual registers to zeropage to be compatible with C64...
+                sta  P8ZP_SCRATCH_B1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		ldy  #0
+-		lda  (P8ZP_SCRATCH_W1),y
+		beq  _notfound
+		cmp  P8ZP_SCRATCH_B1
+		beq  _found
+		iny
+		bne  -
+_notfound	lda  #0
+		ldy  #0
+		rts
+_found		sty  P8ZP_SCRATCH_B1
+		ldy  P8ZP_SCRATCH_W1+1
+		lda  P8ZP_SCRATCH_W1
+		clc
+		adc  P8ZP_SCRATCH_B1
+		bcc  +
+		iny
++		rts
+
+        }}
+    }
+
+    asmsub copy(uword source @R0, uword target @AY) clobbers(A) -> ubyte @Y {
+        ; Copy a string to another, overwriting that one.
+        ; Returns the length of the string that was copied.
+        ; Often you don’t have to call this explicitly and can just write string1 = string2
+        ; but this function is useful if you’re dealing with addresses for instance.
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcpy
+        }}
+    }
+
+    asmsub compare(uword string1 @R0, uword string2 @AY) clobbers(Y) -> byte @A {
+        ; Compares two strings for sorting.
+        ; Returns -1 (255), 0 or 1 depeding on wether string1 sorts before, equal or after string2.
+        ; Note that you can also directly compare strings and string values with eachother using
+        ; comparison operators ==, < etcetera (it will use strcmp for you under water automatically).
+        %asm {{
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strcmp_mem
+        }}
+    }
+
+    asmsub lower(uword st @AY) {
+        ; Lowercases the petscii string in-place.
+        ; (for efficiency, non-letter characters > 128 will also not be left intact,
+        ;  but regular text doesn't usually contain those characters anyway.)
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            beq  _done
+            and  #$7f
+            cmp  #97
+            bcc  +
+            cmp  #123
+            bcs  +
+            and  #%11011111
++           sta  (P8ZP_SCRATCH_W1),y
+            iny
+            bne  -
+_done       rts
+        }}
+    }
+
+    asmsub upper(uword st @AY) {
+        ; Uppercases the petscii string in-place.
+        %asm {{
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+-           lda  (P8ZP_SCRATCH_W1),y
+            beq  _done
+            cmp  #65
+            bcc  +
+            cmp  #91
+            bcs  +
+            ora  #%00100000
++           sta  (P8ZP_SCRATCH_W1),y
+            iny
+            bne  -
+_done       rts
+        }}
+    }
+}

compiler/res/prog8lib/test_stack.p8

@@ -1,48 +1,50 @@
+; utility debug code to print the X (evalstack) and SP (cpu stack) registers.
+
 %import textio
 
 test_stack {
 
     asmsub test() {
         %asm {{
-            stx  _saveX
-            lda  #13
-            jsr  txt.chrout
-            lda  #'-'
-            ldy  #12
--           jsr  txt.chrout
-            dey
-            bne  -
-            lda  #13
-            jsr  txt.chrout
-            lda  #'x'
-            jsr  txt.chrout
-            lda  #'='
-            jsr  txt.chrout
-            lda  _saveX
-            jsr  txt.print_ub
-            lda  #' '
-            jsr  txt.chrout
-            lda  #'s'
-            jsr  txt.chrout
-            lda  #'p'
-            jsr  txt.chrout
-            lda  #'='
-            jsr  txt.chrout
-            tsx
-            txa
-            jsr  txt.print_ub
-            lda  #13
-            jsr  txt.chrout
-            lda  #'-'
-            ldy  #12
--           jsr  txt.chrout
-            dey
-            bne  -
-            lda  #13
-            jsr  txt.chrout
-            ldx  _saveX
-            rts
-_saveX   .byte 0
+	stx  _saveX
+	lda  #13
+	jsr  txt.chrout
+	lda  #'-'
+	ldy  #12
+-	jsr  txt.chrout
+	dey
+	bne  -
+	lda  #13
+	jsr  txt.chrout
+	lda  #'x'
+	jsr  txt.chrout
+	lda  #'='
+	jsr  txt.chrout
+	lda  _saveX
+	jsr  txt.print_ub
+	lda  #' '
+	jsr  txt.chrout
+	lda  #'s'
+	jsr  txt.chrout
+	lda  #'p'
+	jsr  txt.chrout
+	lda  #'='
+	jsr  txt.chrout
+	tsx
+	txa
+	jsr  txt.print_ub
+	lda  #13
+	jsr  txt.chrout
+	lda  #'-'
+	ldy  #12
+-	jsr  txt.chrout
+	dey
+	bne  -
+	lda  #13
+	jsr  txt.chrout
+	ldx  _saveX
+	rts
+_saveX	.byte 0
         }}
     }
 }

compiler/res/version.txt

@@ -1 +1 @@
-5.1
+6.4

compiler/src/prog8/CompilerMain.kt

@@ -1,12 +1,14 @@
 package prog8
 
-import kotlinx.cli.*
+import kotlinx.cli.ArgParser
+import kotlinx.cli.ArgType
+import kotlinx.cli.default
+import kotlinx.cli.multiple
 import prog8.ast.base.AstException
 import prog8.compiler.CompilationResult
 import prog8.compiler.compileProgram
 import prog8.compiler.target.C64Target
 import prog8.compiler.target.Cx16Target
-import prog8.compiler.target.CompilationTarget
 import prog8.parser.ParsingFailedError
 import java.nio.file.FileSystems
 import java.nio.file.Path
@@ -32,20 +34,20 @@ fun pathFrom(stringPath: String, vararg rest: String): Path  = FileSystems.getDe
 
 
 private fun compileMain(args: Array<String>) {
-    val cli = CommandLineInterface("prog8compiler")
-    val startEmulator by cli.flagArgument("-emu", "auto-start emulator after successful compilation")
-    val outputDir by cli.flagValueArgument("-out", "directory", "directory for output files instead of current directory", ".")
-    val dontWriteAssembly by cli.flagArgument("-noasm", "don't create assembly code")
-    val dontOptimize by cli.flagArgument("-noopt", "don't perform any optimizations")
-    val watchMode by cli.flagArgument("-watch", "continuous compilation mode (watches for file changes), greatly increases compilation speed")
-    val slowCodegenWarnings by cli.flagArgument("-slowwarn", "show debug warnings about slow/problematic assembly code generation")
-    val compilationTarget by cli.flagValueArgument("-target", "compilertarget",
-            "target output of the compiler, currently '${C64Target.name}' and '${Cx16Target.name}' available", C64Target.name)
-    val moduleFiles by cli.positionalArgumentsList("modules", "main module file(s) to compile", minArgs = 1)
+    val cli = ArgParser("prog8compiler", prefixStyle = ArgParser.OptionPrefixStyle.JVM)
+    val startEmulator by cli.option(ArgType.Boolean, fullName = "emu", description = "auto-start emulator after successful compilation")
+    val outputDir by cli.option(ArgType.String, fullName = "out", description = "directory for output files instead of current directory").default(".")
+    val dontWriteAssembly by cli.option(ArgType.Boolean, fullName = "noasm", description="don't create assembly code")
+    val dontOptimize by cli.option(ArgType.Boolean, fullName = "noopt", description = "don't perform any optimizations")
+    val watchMode by cli.option(ArgType.Boolean, fullName = "watch", description = "continuous compilation mode (watches for file changes), greatly increases compilation speed")
+    val slowCodegenWarnings by cli.option(ArgType.Boolean, fullName = "slowwarn", description="show debug warnings about slow/problematic assembly code generation")
+    val compilationTarget by cli.option(ArgType.String, fullName = "target", description = "target output of the compiler, currently '${C64Target.name}' and '${Cx16Target.name}' available").default(C64Target.name)
+    val moduleFiles by cli.argument(ArgType.String, fullName = "modules", description = "main module file(s) to compile").multiple(999)
 
     try {
         cli.parse(args)
-    } catch (e: Exception) {
+    } catch (e: IllegalStateException) {
+        System.err.println(e.message)
         exitProcess(1)
     }
 
@@ -55,19 +57,22 @@ private fun compileMain(args: Array<String>) {
         exitProcess(1)
     }
 
-    if(watchMode) {
+    if(watchMode==true) {
         val watchservice = FileSystems.getDefault().newWatchService()
+        val allImportedFiles = mutableSetOf<Path>()
 
         while(true) {
             println("Continuous watch mode active. Modules: $moduleFiles")
             val results = mutableListOf<CompilationResult>()
             for(filepathRaw in moduleFiles) {
                 val filepath = pathFrom(filepathRaw).normalize()
-                val compilationResult = compileProgram(filepath, !dontOptimize, !dontWriteAssembly, slowCodegenWarnings, compilationTarget, outputPath)
+                val compilationResult = compileProgram(filepath, dontOptimize!=true, dontWriteAssembly!=true, slowCodegenWarnings==true, compilationTarget, outputPath)
                 results.add(compilationResult)
             }
 
-            val allImportedFiles = results.flatMap { it.importedFiles }
+            val allNewlyImportedFiles = results.flatMap { it.importedFiles }
+            allImportedFiles.addAll(allNewlyImportedFiles)
+
             println("Imported files (now watching:)")
             for (importedFile in allImportedFiles) {
                 print("  ")
@@ -98,7 +103,7 @@ private fun compileMain(args: Array<String>) {
             val filepath = pathFrom(filepathRaw).normalize()
             val compilationResult: CompilationResult
             try {
-                compilationResult = compileProgram(filepath, !dontOptimize, !dontWriteAssembly, slowCodegenWarnings, compilationTarget, outputPath)
+                compilationResult = compileProgram(filepath, dontOptimize!=true, dontWriteAssembly!=true, slowCodegenWarnings==true, compilationTarget, outputPath)
                 if(!compilationResult.success)
                     exitProcess(1)
             } catch (x: ParsingFailedError) {
@@ -107,11 +112,11 @@ private fun compileMain(args: Array<String>) {
                 exitProcess(1)
             }
 
-            if (startEmulator) {
+            if (startEmulator==true) {
                 if (compilationResult.programName.isEmpty())
                     println("\nCan't start emulator because no program was assembled.")
-                else if(startEmulator) {
-                    CompilationTarget.instance.machine.launchEmulator(compilationResult.programName)
+                else {
+                    compilationResult.compTarget.machine.launchEmulator(compilationResult.programName)
                 }
             }
         }

compiler/src/prog8/ast/processing/ImportedModuleDirectiveRemover.kt

@@ -1,22 +0,0 @@
-package prog8.ast.processing
-
-import prog8.ast.INameScope
-import prog8.ast.Node
-import prog8.ast.statements.Directive
-
-
-internal class ImportedModuleDirectiveRemover: AstWalker() {
-    /**
-     * Most global directives don't apply for imported modules, so remove them
-     */
-
-    private val moduleLevelDirectives = listOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address")
-    private val noModifications = emptyList<IAstModification>()
-
-    override fun before(directive: Directive, parent: Node): Iterable<IAstModification> {
-        if(directive.directive in moduleLevelDirectives) {
-            return listOf(IAstModification.Remove(directive, parent as INameScope))
-        }
-        return noModifications
-    }
-}

compiler/src/prog8/ast/processing/StatementReorderer.kt

@@ -1,370 +0,0 @@
-package prog8.ast.processing
-
-import prog8.ast.*
-import prog8.ast.base.*
-import prog8.ast.expressions.*
-import prog8.ast.statements.*
-
-
-internal class StatementReorderer(val program: Program, val errors: ErrorReporter) : AstWalker() {
-    // Reorders the statements in a way the compiler needs.
-    // - 'main' block must be the very first statement UNLESS it has an address set.
-    // - library blocks are put last.
-    // - blocks are ordered by address, where blocks without address are placed last.
-    // - in every block and module, most directives and vardecls are moved to the top. (not in subroutines!)
-    // - the 'start' subroutine is moved to the top.
-    // - (syntax desugaring) a vardecl with a non-const initializer value is split into a regular vardecl and an assignment statement.
-    // - (syntax desugaring) struct value assignment is expanded into several struct member assignments.
-    // - in-place assignments are reordered a bit so that they are mostly of the form A = A <operator> <rest>
-    // - sorts the choices in when statement.
-    // - insert AddressOf (&) expression where required (string params to a UWORD function param etc).
-
-    private val noModifications = emptyList<IAstModification>()
-    private val directivesToMove = setOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address", "%option")
-
-    override fun after(module: Module, parent: Node): Iterable<IAstModification> {
-        val (blocks, other) = module.statements.partition { it is Block }
-        module.statements = other.asSequence().plus(blocks.sortedBy { (it as Block).address ?: Int.MAX_VALUE }).toMutableList()
-
-        val mainBlock = module.statements.filterIsInstance<Block>().firstOrNull { it.name=="main" }
-        if(mainBlock!=null && mainBlock.address==null) {
-            module.statements.remove(mainBlock)
-            module.statements.add(0, mainBlock)
-        }
-
-        reorderVardeclsAndDirectives(module.statements)
-        return noModifications
-    }
-
-    private fun reorderVardeclsAndDirectives(statements: MutableList<Statement>) {
-        val varDecls = statements.filterIsInstance<VarDecl>()
-        statements.removeAll(varDecls)
-        statements.addAll(0, varDecls)
-
-        val directives = statements.filterIsInstance<Directive>().filter {it.directive in directivesToMove}
-        statements.removeAll(directives)
-        statements.addAll(0, directives)
-    }
-
-    override fun before(block: Block, parent: Node): Iterable<IAstModification> {
-        parent as Module
-        if(block.isInLibrary) {
-            return listOf(
-                    IAstModification.Remove(block, parent),
-                    IAstModification.InsertLast(block, parent)
-            )
-        }
-
-        reorderVardeclsAndDirectives(block.statements)
-        return noModifications
-    }
-
-    override fun before(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
-        if(subroutine.name=="start" && parent is Block) {
-            if(parent.statements.filterIsInstance<Subroutine>().first().name!="start") {
-                return listOf(
-                        IAstModification.Remove(subroutine, parent),
-                        IAstModification.InsertFirst(subroutine, parent)
-                )
-            }
-        }
-        return noModifications
-    }
-
-    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
-        when (val expr2 = arrayIndexedExpression.indexer.origExpression) {
-            is NumericLiteralValue -> {
-                arrayIndexedExpression.indexer.indexNum = expr2
-                arrayIndexedExpression.indexer.origExpression = null
-                return noModifications
-            }
-            is IdentifierReference -> {
-                arrayIndexedExpression.indexer.indexVar = expr2
-                arrayIndexedExpression.indexer.origExpression = null
-                return noModifications
-            }
-            is Expression -> {
-                // replace complex indexing with a temp variable
-                return getAutoIndexerVarFor(arrayIndexedExpression)
-            }
-            else -> return noModifications
-        }
-    }
-
-    private fun getAutoIndexerVarFor(expr: ArrayIndexedExpression): MutableList<IAstModification> {
-        val modifications = mutableListOf<IAstModification>()
-        val subroutine = expr.definingSubroutine()!!
-        val statement = expr.containingStatement()
-        val indexerVarPrefix = "prog8_autovar_index_"
-        val indexerVarName: String
-        val repo = subroutine.asmGenInfo.usedAutoArrayIndexerForStatements
-
-        val freeVar = repo.filter { statement !in it.value }.map { it.key }.firstOrNull()
-        if(freeVar==null) {
-            // add another loop index var to be used for this expression
-            val statementId = expr.hashCode()
-            indexerVarName = "$indexerVarPrefix$statementId"
-            // create the indexer var at block level scope
-            val vardecl = VarDecl(VarDeclType.VAR, DataType.UBYTE, ZeropageWish.PREFER_ZEROPAGE,
-                    null, indexerVarName, null, null, isArray = false, autogeneratedDontRemove = true, position = expr.position)
-            modifications.add(IAstModification.InsertFirst(vardecl, subroutine))
-            var statements = repo[indexerVarName]
-            if(statements==null) {
-                statements = mutableSetOf()
-                repo[indexerVarName] = statements
-            }
-            statements.add(statement)
-        } else {
-            // reuse an already created indexer autovar
-            repo.getValue(freeVar).add(statement)
-            indexerVarName = freeVar
-        }
-
-        // assign the indexing expression to the helper variable, and replace the indexer with just the variable
-        val indexerExpression = expr.indexer.origExpression!!
-        val target = AssignTarget(IdentifierReference(listOf(indexerVarName), indexerExpression.position), null, null, indexerExpression.position)
-        val assign = Assignment(target, indexerExpression, indexerExpression.position)
-        val beforeWhat = expr.containingStatement()
-        modifications.add(IAstModification.InsertBefore(beforeWhat, assign, beforeWhat.definingScope()))
-        modifications.add(IAstModification.SetExpression( {
-            expr.indexer.indexVar = it as IdentifierReference
-            expr.indexer.indexNum = null
-            expr.indexer.origExpression = null
-        }, target.identifier!!.copy(), expr.indexer))
-
-        return modifications
-    }
-
-    override fun after(whenStatement: WhenStatement, parent: Node): Iterable<IAstModification> {
-        val choices = whenStatement.choiceValues(program).sortedBy {
-            it.first?.first() ?: Int.MAX_VALUE
-        }
-        whenStatement.choices.clear()
-        choices.mapTo(whenStatement.choices) { it.second }
-        return noModifications
-    }
-
-    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
-        val declValue = decl.value
-        if(declValue!=null && decl.type== VarDeclType.VAR && decl.datatype in NumericDatatypes) {
-            val declConstValue = declValue.constValue(program)
-            if(declConstValue==null) {
-                // move the vardecl (without value) to the scope and replace this with a regular assignment
-                // Unless we're dealing with a floating point variable because that will actually make things less efficient at the moment (because floats are mostly calcualated via the stack)
-                if(decl.datatype!=DataType.FLOAT) {
-                    decl.value = null
-                    decl.allowInitializeWithZero = false
-                    val target = AssignTarget(IdentifierReference(listOf(decl.name), decl.position), null, null, decl.position)
-                    val assign = Assignment(target, declValue, decl.position)
-                    return listOf(
-                            IAstModification.ReplaceNode(decl, assign, parent),
-                            IAstModification.InsertFirst(decl, decl.definingScope())
-                    )
-                }
-            }
-        }
-        return noModifications
-    }
-
-    override fun before(assignment: Assignment, parent: Node): Iterable<IAstModification> {
-        val valueType = assignment.value.inferType(program)
-        val targetType = assignment.target.inferType(program)
-        var assignments = emptyList<Assignment>()
-
-        if(targetType.istype(DataType.STRUCT) && (valueType.istype(DataType.STRUCT) || valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes )) {
-            assignments = if (assignment.value is ArrayLiteralValue) {
-                flattenStructAssignmentFromStructLiteral(assignment)    //  'structvar = [ ..... ] '
-            } else {
-                flattenStructAssignmentFromIdentifier(assignment)    //   'structvar1 = structvar2'
-            }
-        }
-
-        if(targetType.typeOrElse(DataType.STRUCT) in ArrayDatatypes && valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes ) {
-            assignments = if (assignment.value is ArrayLiteralValue) {
-                flattenArrayAssignmentFromArrayLiteral(assignment)    //  'arrayvar = [ ..... ] '
-            } else {
-                flattenArrayAssignmentFromIdentifier(assignment)    //   'arrayvar1 = arrayvar2'
-            }
-        }
-
-        if(assignments.isNotEmpty()) {
-            val modifications = mutableListOf<IAstModification>()
-            val scope = assignment.definingScope()
-            assignments.reversed().mapTo(modifications) { IAstModification.InsertAfter(assignment, it, scope) }
-            modifications.add(IAstModification.Remove(assignment, scope))
-            return modifications
-        }
-
-        return noModifications
-    }
-
-    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
-        // rewrite in-place assignment expressions a bit so that the assignment target usually is the leftmost operand
-        val binExpr = assignment.value as? BinaryExpression
-        if(binExpr!=null) {
-            if(binExpr.left isSameAs assignment.target) {
-                // A = A <operator> 5, unchanged
-                return noModifications
-            }
-
-            if(binExpr.operator in associativeOperators) {
-                if (binExpr.right isSameAs assignment.target) {
-                    // A = v <associative-operator> A  ==>  A = A <associative-operator> v
-                    return listOf(IAstModification.SwapOperands(binExpr))
-                }
-
-                val leftBinExpr = binExpr.left as? BinaryExpression
-                if(leftBinExpr?.operator == binExpr.operator) {
-                    return if(leftBinExpr.left isSameAs assignment.target) {
-                        // A = (A <associative-operator> x) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
-                        val newRight = BinaryExpression(leftBinExpr.right, binExpr.operator, binExpr.right, binExpr.position)
-                        val newValue = BinaryExpression(leftBinExpr.left, binExpr.operator, newRight, binExpr.position)
-                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
-                    } else {
-                        // A = (x <associative-operator> A) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
-                        val newRight = BinaryExpression(leftBinExpr.left, binExpr.operator, binExpr.right, binExpr.position)
-                        val newValue = BinaryExpression(leftBinExpr.right, binExpr.operator, newRight, binExpr.position)
-                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
-                    }
-                }
-                val rightBinExpr = binExpr.right as? BinaryExpression
-                if(rightBinExpr?.operator == binExpr.operator) {
-                    return if(rightBinExpr.left isSameAs assignment.target) {
-                        // A = x <associative-operator> (A <same-operator> y) ==> A = A <associative-operator> (x <same-operator> y)
-                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.right, binExpr.position)
-                        val newValue = BinaryExpression(rightBinExpr.left, binExpr.operator, newRight, binExpr.position)
-                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
-                    } else {
-                        // A = x <associative-operator> (y <same-operator> A) ==> A = A <associative-operator> (x <same-operator> y)
-                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.left, binExpr.position)
-                        val newValue = BinaryExpression(rightBinExpr.right, binExpr.operator, newRight, binExpr.position)
-                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
-                    }
-                }
-            }
-        }
-
-        return noModifications
-    }
-
-    private fun flattenArrayAssignmentFromArrayLiteral(assign: Assignment): List<Assignment> {
-        val identifier = assign.target.identifier!!
-        val targetVar = identifier.targetVarDecl(program.namespace)!!
-        val alv = assign.value as? ArrayLiteralValue
-        return flattenArrayAssign(targetVar, alv, identifier, assign.position)
-    }
-
-    private fun flattenArrayAssignmentFromIdentifier(assign: Assignment): List<Assignment> {
-        val identifier = assign.target.identifier!!
-        val targetVar = identifier.targetVarDecl(program.namespace)!!
-        val sourceIdent = assign.value as IdentifierReference
-        val sourceVar = sourceIdent.targetVarDecl(program.namespace)!!
-        if(!sourceVar.isArray) {
-            errors.err("value must be an array",  sourceIdent.position)
-            return emptyList()
-        }
-        val alv = sourceVar.value as? ArrayLiteralValue
-        return flattenArrayAssign(targetVar, alv, identifier, assign.position)
-    }
-
-    private fun flattenArrayAssign(targetVar: VarDecl, alv: ArrayLiteralValue?, identifier: IdentifierReference, position: Position): List<Assignment> {
-        if(targetVar.arraysize==null) {
-            errors.err("array has no defined size", identifier.position)
-            return emptyList()
-        }
-
-        if(alv==null || alv.value.size != targetVar.arraysize!!.constIndex()) {
-            errors.err("element count mismatch", position)
-            return emptyList()
-        }
-
-        // TODO use a pointer loop instead of individual assignments
-        return alv.value.mapIndexed { index, value ->
-            val idx = ArrayIndexedExpression(identifier, ArrayIndex(NumericLiteralValue(DataType.UBYTE, index, position), position), position)
-            Assignment(AssignTarget(null, idx, null, position), value, value.position)
-        }
-    }
-
-    private fun flattenStructAssignmentFromStructLiteral(structAssignment: Assignment): List<Assignment> {
-        val identifier = structAssignment.target.identifier!!
-        val identifierName = identifier.nameInSource.single()
-        val targetVar = identifier.targetVarDecl(program.namespace)!!
-        val struct = targetVar.struct!!
-
-        val slv = structAssignment.value as? ArrayLiteralValue
-        if(slv==null || slv.value.size != struct.numberOfElements) {
-            errors.err("element count mismatch", structAssignment.position)
-            return emptyList()
-        }
-
-        return struct.statements.zip(slv.value).map { (targetDecl, sourceValue) ->
-            targetDecl as VarDecl
-            val mangled = mangledStructMemberName(identifierName, targetDecl.name)
-            val idref = IdentifierReference(listOf(mangled), structAssignment.position)
-            val assign = Assignment(AssignTarget(idref, null, null, structAssignment.position),
-                    sourceValue, sourceValue.position)
-            assign.linkParents(structAssignment)
-            assign
-        }
-    }
-
-    private fun flattenStructAssignmentFromIdentifier(structAssignment: Assignment): List<Assignment> {
-        // TODO use memcopy beyond a certain number of elements
-        val identifier = structAssignment.target.identifier!!
-        val identifierName = identifier.nameInSource.single()
-        val targetVar = identifier.targetVarDecl(program.namespace)!!
-        val struct = targetVar.struct!!
-        when (structAssignment.value) {
-            is IdentifierReference -> {
-                val sourceVar = (structAssignment.value as IdentifierReference).targetVarDecl(program.namespace)!!
-                when {
-                    sourceVar.struct!=null -> {
-                        // struct memberwise copy
-                        val sourceStruct = sourceVar.struct!!
-                        if(sourceStruct!==targetVar.struct) {
-                            // structs are not the same in assignment
-                            return listOf()     // error will be printed elsewhere
-                        }
-                        if(struct.statements.size!=sourceStruct.statements.size)
-                            return listOf()     // error will be printed elsewhere
-                        return struct.statements.zip(sourceStruct.statements).map { member ->
-                            val targetDecl = member.first as VarDecl
-                            val sourceDecl = member.second as VarDecl
-                            if(targetDecl.name != sourceDecl.name)
-                                throw FatalAstException("struct member mismatch")
-                            val mangled = mangledStructMemberName(identifierName, targetDecl.name)
-                            val idref = IdentifierReference(listOf(mangled), structAssignment.position)
-                            val sourcemangled = mangledStructMemberName(sourceVar.name, sourceDecl.name)
-                            val sourceIdref = IdentifierReference(listOf(sourcemangled), structAssignment.position)
-                            val assign = Assignment(AssignTarget(idref, null, null, structAssignment.position), sourceIdref, member.second.position)
-                            assign.linkParents(structAssignment)
-                            assign
-                        }
-                    }
-                    sourceVar.isArray -> {
-                        val array = (sourceVar.value as ArrayLiteralValue).value
-                        if(struct.statements.size!=array.size)
-                            return listOf()     // error will be printed elsewhere
-                        return struct.statements.zip(array).map {
-                            val decl = it.first as VarDecl
-                            val mangled = mangledStructMemberName(identifierName, decl.name)
-                            val targetName = IdentifierReference(listOf(mangled), structAssignment.position)
-                            val target = AssignTarget(targetName, null, null, structAssignment.position)
-                            val assign = Assignment(target, it.second, structAssignment.position)
-                            assign.linkParents(structAssignment)
-                            assign
-                        }
-                    }
-                    else -> {
-                        throw FatalAstException("can only assign arrays or structs to structs")
-                    }
-                }
-            }
-            is ArrayLiteralValue -> {
-                throw IllegalArgumentException("not going to flatten a structLv assignment here")
-            }
-            else -> throw FatalAstException("strange struct value")
-        }
-    }
-
-}

compiler/src/prog8/ast/processing/VariousCleanups.kt

@@ -1,44 +0,0 @@
-package prog8.ast.processing
-
-import prog8.ast.INameScope
-import prog8.ast.Node
-import prog8.ast.expressions.NumericLiteralValue
-import prog8.ast.expressions.TypecastExpression
-import prog8.ast.statements.AnonymousScope
-import prog8.ast.statements.NopStatement
-
-
-internal class VariousCleanups: AstWalker() {
-    private val noModifications = emptyList<IAstModification>()
-
-    override fun before(nopStatement: NopStatement, parent: Node): Iterable<IAstModification> {
-        return listOf(IAstModification.Remove(nopStatement, parent as INameScope))
-    }
-
-    override fun before(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
-        return if(parent is INameScope)
-            listOf(ScopeFlatten(scope, parent as INameScope))
-        else
-            noModifications
-    }
-
-    class ScopeFlatten(val scope: AnonymousScope, val into: INameScope) : IAstModification {
-        override fun perform() {
-            val idx = into.statements.indexOf(scope)
-            if(idx>=0) {
-                into.statements.addAll(idx+1, scope.statements)
-                into.statements.remove(scope)
-            }
-        }
-    }
-
-    override fun before(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
-        if(typecast.expression is NumericLiteralValue) {
-            val value = (typecast.expression as NumericLiteralValue).cast(typecast.type)
-            if(value.isValid)
-                return listOf(IAstModification.ReplaceNode(typecast, value.valueOrZero(), parent))
-        }
-
-        return noModifications
-    }
-}

compiler/src/prog8/compiler/BeforeAsmGenerationAstChanger.kt

@@ -5,12 +5,14 @@ import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
 import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.target.ICompilationTarget
 
 
-internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: ErrorReporter) : AstWalker() {
+internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: IErrorReporter, private val compTarget: ICompilationTarget) : AstWalker() {
 
     private val noModifications = emptyList<IAstModification>()
 
@@ -37,7 +39,7 @@ internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: E
         // But it can only be done if the target variable IS NOT OCCURRING AS AN OPERAND ITSELF.
         if(!assignment.isAugmentable
                 && assignment.target.identifier != null
-                && assignment.target.isInRegularRAM(program.namespace)) {
+                && compTarget.isInRegularRAM(assignment.target, program)) {
             val binExpr = assignment.value as? BinaryExpression
             if (binExpr != null && binExpr.operator !in comparisonOperators) {
                 if (binExpr.left !is BinaryExpression) {
@@ -106,11 +108,12 @@ internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: E
         }
 
 
-        // add the implicit return statement at the end (if it's not there yet), but only if it's not a kernel routine.
+        // add the implicit return statement at the end (if it's not there yet), but only if it's not a kernal routine.
         // and if an assembly block doesn't contain a rts/rti, and some other situations.
         val mods = mutableListOf<IAstModification>()
         val returnStmt = Return(null, subroutine.position)
         if (subroutine.asmAddress == null
+                && !subroutine.inline
                 && subroutine.statements.isNotEmpty()
                 && subroutine.amountOfRtsInAsm() == 0
                 && subroutine.statements.lastOrNull { it !is VarDecl } !is Return
@@ -152,16 +155,6 @@ internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: E
         // The only place for now where we can do this is for:
         //    asmsub register pair parameter.
 
-        if(typecast.type in WordDatatypes) {
-            val fcall = typecast.parent as? IFunctionCall
-            if (fcall != null) {
-                val sub = fcall.target.targetStatement(program.namespace) as? Subroutine
-                if (sub != null && sub.isAsmSubroutine) {
-                    return listOf(IAstModification.ReplaceNode(typecast, typecast.expression, parent))
-                }
-            }
-        }
-
         if(sourceDt in PassByReferenceDatatypes) {
             if(typecast.type==DataType.UWORD) {
                 if(typecast.expression is IdentifierReference) {
@@ -170,6 +163,12 @@ internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: E
                             AddressOf(typecast.expression as IdentifierReference, typecast.position),
                             parent
                     ))
+                } else if(typecast.expression is IFunctionCall) {
+                    return listOf(IAstModification.ReplaceNode(
+                            typecast,
+                            typecast.expression,
+                            parent
+                    ))
                 }
             } else {
                 errors.err("cannot cast pass-by-reference value to type ${typecast.type} (only to UWORD)", typecast.position)
@@ -208,4 +207,93 @@ internal class BeforeAsmGenerationAstChanger(val program: Program, val errors: E
         }
         return noModifications
     }
+
+    override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        if(functionCallStatement.target.nameInSource==listOf("cmp")) {
+            // if the datatype of the arguments of cmp() are different, cast the byte one to word.
+            val arg1 = functionCallStatement.args[0]
+            val arg2 = functionCallStatement.args[1]
+            val dt1 = arg1.inferType(program).typeOrElse(DataType.STRUCT)
+            val dt2 = arg2.inferType(program).typeOrElse(DataType.STRUCT)
+            if(dt1 in ByteDatatypes) {
+                if(dt2 in ByteDatatypes)
+                    return noModifications
+                val cast1 = TypecastExpression(arg1, if(dt1==DataType.UBYTE) DataType.UWORD else DataType.WORD, true, functionCallStatement.position)
+                return listOf(IAstModification.ReplaceNode(arg1, cast1, functionCallStatement))
+            } else {
+                if(dt2 in WordDatatypes)
+                    return noModifications
+                val cast2 = TypecastExpression(arg2, if(dt2==DataType.UBYTE) DataType.UWORD else DataType.WORD, true, functionCallStatement.position)
+                return listOf(IAstModification.ReplaceNode(arg2, cast2, functionCallStatement))
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
+
+        val containingStatement = getContainingStatement(arrayIndexedExpression)
+        if(getComplexArrayIndexedExpressions(containingStatement).size > 1) {
+            errors.err("it's not possible to use more than one complex array indexing expression in a single statement; break it up via a temporary variable for instance", containingStatement.position)
+            return noModifications
+        }
+
+
+        val index = arrayIndexedExpression.indexer.indexExpr
+        if(index !is NumericLiteralValue && index !is IdentifierReference) {
+            // replace complex indexing expression with a temp variable to hold the computed index first
+            return getAutoIndexerVarFor(arrayIndexedExpression)
+        }
+
+        return noModifications
+    }
+
+    private fun getComplexArrayIndexedExpressions(stmt: Statement): List<ArrayIndexedExpression> {
+
+        class Searcher : IAstVisitor {
+            val complexArrayIndexedExpressions = mutableListOf<ArrayIndexedExpression>()
+            override fun visit(arrayIndexedExpression: ArrayIndexedExpression) {
+                val ix = arrayIndexedExpression.indexer.indexExpr
+                if(ix !is NumericLiteralValue && ix !is IdentifierReference)
+                    complexArrayIndexedExpressions.add(arrayIndexedExpression)
+            }
+
+            override fun visit(branchStatement: BranchStatement) {}
+
+            override fun visit(forLoop: ForLoop) {}
+
+            override fun visit(ifStatement: IfStatement) {
+                ifStatement.condition.accept(this)
+            }
+
+            override fun visit(untilLoop: UntilLoop) {
+                untilLoop.condition.accept(this)
+            }
+        }
+
+        val searcher = Searcher()
+        stmt.accept(searcher)
+        return searcher.complexArrayIndexedExpressions
+    }
+
+    private fun getContainingStatement(expression: Expression): Statement {
+        var node: Node = expression
+        while(node !is Statement)
+            node = node.parent
+
+        return node
+    }
+
+    private fun getAutoIndexerVarFor(expr: ArrayIndexedExpression): MutableList<IAstModification> {
+        val modifications = mutableListOf<IAstModification>()
+        val statement = expr.containingStatement()
+        // replace the indexer with just the variable (simply use a cx16 virtual register r9, that we HOPE is not used for other things in the expression...)
+        // assign the indexing expression to the helper variable, but only if that hasn't been done already
+        val target = AssignTarget(IdentifierReference(listOf("cx16", "r9"), expr.indexer.position), null, null, expr.indexer.position)
+        val assign = Assignment(target, expr.indexer.indexExpr, expr.indexer.position)
+        modifications.add(IAstModification.InsertBefore(statement, assign, statement.definingScope()))
+        modifications.add(IAstModification.ReplaceNode(expr.indexer.indexExpr, target.identifier!!.copy(), expr.indexer))
+        return modifications
+    }
+
 }

compiler/src/prog8/compiler/Compiler.kt

@@ -1,9 +1,30 @@
 package prog8.compiler
 
+import prog8.ast.AstToSourceCode
+import prog8.ast.IBuiltinFunctions
+import prog8.ast.IMemSizer
+import prog8.ast.Program
+import prog8.ast.base.AstException
+import prog8.ast.base.Position
+import prog8.ast.expressions.Expression
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.Directive
+import prog8.compiler.astprocessing.*
+import prog8.compiler.functions.*
+import prog8.compiler.target.C64Target
+import prog8.compiler.target.Cx16Target
+import prog8.compiler.target.ICompilationTarget
+import prog8.compiler.target.asmGeneratorFor
+import prog8.optimizer.*
+import prog8.parser.ModuleImporter
+import prog8.parser.ParsingFailedError
+import prog8.parser.moduleName
 import java.io.File
 import java.io.InputStream
 import java.nio.file.Path
-import kotlin.math.abs
+import kotlin.system.exitProcess
+import kotlin.system.measureTimeMillis
+
 
 enum class OutputType {
     RAW,
@@ -28,33 +49,283 @@ data class CompilationOptions(val output: OutputType,
                               val zeropage: ZeropageType,
                               val zpReserved: List<IntRange>,
                               val floats: Boolean,
-                              val noSysInit: Boolean) {
+                              val noSysInit: Boolean,
+                              val compTarget: ICompilationTarget) {
     var slowCodegenWarnings = false
+    var optimize = false
 }
 
 
 class CompilerException(message: String?) : Exception(message)
 
-fun Number.toHex(): String {
-    //  0..15 -> "0".."15"
-    //  16..255 -> "$10".."$ff"
-    //  256..65536 -> "$0100".."$ffff"
-    // negative values are prefixed with '-'.
-    val integer = this.toInt()
-    if(integer<0)
-        return '-' + abs(integer).toHex()
-    return when (integer) {
-        in 0 until 16 -> integer.toString()
-        in 0 until 0x100 -> "$"+integer.toString(16).padStart(2,'0')
-        in 0 until 0x10000 -> "$"+integer.toString(16).padStart(4,'0')
-        else -> throw CompilerException("number too large for 16 bits $this")
+class CompilationResult(val success: Boolean,
+                        val programAst: Program,
+                        val programName: String,
+                        val compTarget: ICompilationTarget,
+                        val importedFiles: List<Path>)
+
+
+fun compileProgram(filepath: Path,
+                   optimize: Boolean,
+                   writeAssembly: Boolean,
+                   slowCodegenWarnings: Boolean,
+                   compilationTarget: String,
+                   outputDir: Path): CompilationResult {
+    var programName = ""
+    lateinit var programAst: Program
+    lateinit var importedFiles: List<Path>
+    val errors = ErrorReporter()
+
+    val compTarget =
+        when(compilationTarget) {
+            C64Target.name -> C64Target
+            Cx16Target.name -> Cx16Target
+            else -> {
+                System.err.println("invalid compilation target")
+                exitProcess(1)
+            }
+        }
+
+    try {
+        val totalTime = measureTimeMillis {
+            // import main module and everything it needs
+            val (ast, compilationOptions, imported) = parseImports(filepath, errors, compTarget)
+            compilationOptions.slowCodegenWarnings = slowCodegenWarnings
+            compilationOptions.optimize = optimize
+            programAst = ast
+            importedFiles = imported
+            processAst(programAst, errors, compilationOptions)
+            if (compilationOptions.optimize)
+                optimizeAst(programAst, errors, BuiltinFunctionsFacade(BuiltinFunctions), compTarget)
+            postprocessAst(programAst, errors, compilationOptions)
+
+            // printAst(programAst)
+
+            if(writeAssembly)
+                programName = writeAssembly(programAst, errors, outputDir, compilationOptions)
+        }
+        System.out.flush()
+        System.err.flush()
+        println("\nTotal compilation+assemble time: ${totalTime / 1000.0} sec.")
+        return CompilationResult(true, programAst, programName, compTarget, importedFiles)
+
+    } catch (px: ParsingFailedError) {
+        System.err.print("\u001b[91m")  // bright red
+        System.err.println(px.message)
+        System.err.print("\u001b[0m")  // reset
+    } catch (ax: AstException) {
+        System.err.print("\u001b[91m")  // bright red
+        System.err.println(ax.toString())
+        System.err.print("\u001b[0m")  // reset
+    } catch (x: Exception) {
+        print("\u001b[91m")  // bright red
+        println("\n* internal error *")
+        print("\u001b[0m")  // reset
+        System.out.flush()
+        throw x
+    } catch (x: NotImplementedError) {
+        print("\u001b[91m")  // bright red
+        println("\n* internal error: missing feature/code *")
+        print("\u001b[0m")  // reset
+        System.out.flush()
+        throw x
     }
+
+    val failedProgram = Program("failed", mutableListOf(), BuiltinFunctionsFacade(BuiltinFunctions), compTarget)
+    return CompilationResult(false, failedProgram, programName, compTarget, emptyList())
+}
+
+private class BuiltinFunctionsFacade(functions: Map<String, FSignature>): IBuiltinFunctions {
+    lateinit var program: Program
+
+    override val names = functions.keys
+    override val purefunctionNames = functions.filter { it.value.pure }.map { it.key }.toSet()
+
+    override fun constValue(name: String, args: List<Expression>, position: Position, memsizer: IMemSizer): NumericLiteralValue? {
+        val func = BuiltinFunctions[name]
+        if(func!=null) {
+            val exprfunc = func.constExpressionFunc
+            if(exprfunc!=null) {
+                return try {
+                    exprfunc(args, position, program, memsizer)
+                } catch(x: NotConstArgumentException) {
+                    // const-evaluating the builtin function call failed.
+                    null
+                } catch(x: CannotEvaluateException) {
+                    // const-evaluating the builtin function call failed.
+                    null
+                }
+            }
+            else if(func.known_returntype==null)
+                return null  // builtin function $name can't be used here because it doesn't return a value
+        }
+        return null
+    }
+    override fun returnType(name: String, args: MutableList<Expression>) =
+        builtinFunctionReturnType(name, args, program)
+}
+
+private fun parseImports(filepath: Path, errors: IErrorReporter, compTarget: ICompilationTarget): Triple<Program, CompilationOptions, List<Path>> {
+    val compilationTargetName = compTarget.name
+    println("Compiler target: $compilationTargetName. Parsing...")
+    val importer = ModuleImporter()
+    val bf = BuiltinFunctionsFacade(BuiltinFunctions)
+    val programAst = Program(moduleName(filepath.fileName), mutableListOf(), bf, compTarget)
+    bf.program = programAst
+    importer.importModule(programAst, filepath, compTarget, compilationTargetName)
+    errors.report()
+
+    val importedFiles = programAst.modules.filter { !it.source.startsWith("@embedded@") }.map { it.source }
+    val compilerOptions = determineCompilationOptions(programAst, compTarget)
+    if (compilerOptions.launcher == LauncherType.BASIC && compilerOptions.output != OutputType.PRG)
+        throw ParsingFailedError("${programAst.modules.first().position} BASIC launcher requires output type PRG.")
+
+    // depending on the machine and compiler options we may have to include some libraries
+    for(lib in compTarget.machine.importLibs(compilerOptions, compilationTargetName))
+        importer.importLibraryModule(programAst, lib, compTarget, compilationTargetName)
+
+    // always import prog8_lib and math
+    importer.importLibraryModule(programAst, "math", compTarget, compilationTargetName)
+    importer.importLibraryModule(programAst, "prog8_lib", compTarget, compilationTargetName)
+    errors.report()
+    return Triple(programAst, compilerOptions, importedFiles)
+}
+
+private fun determineCompilationOptions(program: Program, compTarget: ICompilationTarget): CompilationOptions {
+    val mainModule = program.mainModule
+    val outputType = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%output" }
+            as? Directive)?.args?.single()?.name?.toUpperCase()
+    val launcherType = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%launcher" }
+            as? Directive)?.args?.single()?.name?.toUpperCase()
+    val zpoption: String? = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%zeropage" }
+            as? Directive)?.args?.single()?.name?.toUpperCase()
+    val allOptions = program.modules.flatMap { it.statements }.filter { it is Directive && it.directive == "%option" }.flatMap { (it as Directive).args }.toSet()
+    val floatsEnabled = allOptions.any { it.name == "enable_floats" }
+    val noSysInit = allOptions.any { it.name == "no_sysinit" }
+    var zpType: ZeropageType =
+            if (zpoption == null)
+                if(floatsEnabled) ZeropageType.FLOATSAFE else ZeropageType.KERNALSAFE
+            else
+                try {
+                    ZeropageType.valueOf(zpoption)
+                } catch (x: IllegalArgumentException) {
+                    ZeropageType.KERNALSAFE
+                    // error will be printed by the astchecker
+                }
+
+    if (zpType==ZeropageType.FLOATSAFE && compTarget.name == Cx16Target.name) {
+        System.err.println("Warning: Cx16 target must use zp option basicsafe instead of floatsafe")
+        zpType = ZeropageType.BASICSAFE
+    }
+
+    val zpReserved = mainModule.statements
+            .asSequence()
+            .filter { it is Directive && it.directive == "%zpreserved" }
+            .map { (it as Directive).args }
+            .map { it[0].int!!..it[1].int!! }
+            .toList()
+
+    if(outputType!=null && !OutputType.values().any {it.name==outputType}) {
+        System.err.println("invalid output type $outputType")
+        exitProcess(1)
+    }
+    if(launcherType!=null && !LauncherType.values().any {it.name==launcherType}) {
+        System.err.println("invalid launcher type $launcherType")
+        exitProcess(1)
+    }
+
+    return CompilationOptions(
+            if (outputType == null) OutputType.PRG else OutputType.valueOf(outputType),
+            if (launcherType == null) LauncherType.BASIC else LauncherType.valueOf(launcherType),
+            zpType, zpReserved, floatsEnabled, noSysInit,
+            compTarget
+    )
+}
+
+private fun processAst(programAst: Program, errors: IErrorReporter, compilerOptions: CompilationOptions) {
+    // perform initial syntax checks and processings
+    println("Processing for target ${compilerOptions.compTarget.name}...")
+    programAst.checkIdentifiers(errors, compilerOptions.compTarget)
+    errors.report()
+    programAst.constantFold(errors, compilerOptions.compTarget)
+    errors.report()
+    programAst.reorderStatements(errors)
+    errors.report()
+    programAst.addTypecasts(errors)
+    errors.report()
+    programAst.variousCleanups()
+    programAst.checkValid(compilerOptions, errors, compilerOptions.compTarget)
+    errors.report()
+    programAst.checkIdentifiers(errors, compilerOptions.compTarget)
+    errors.report()
+}
+
+private fun optimizeAst(programAst: Program, errors: IErrorReporter, functions: IBuiltinFunctions, compTarget: ICompilationTarget) {
+    // optimize the parse tree
+    println("Optimizing...")
+    while (true) {
+        // keep optimizing expressions and statements until no more steps remain
+        val optsDone1 = programAst.simplifyExpressions()
+        val optsDone2 = programAst.splitBinaryExpressions(compTarget)
+        val optsDone3 = programAst.optimizeStatements(errors, functions, compTarget, ::loadAsmIncludeFile)
+        programAst.constantFold(errors, compTarget) // because simplified statements and expressions can result in more constants that can be folded away
+        errors.report()
+        if (optsDone1 + optsDone2 + optsDone3 == 0)
+            break
+    }
+
+    val remover = UnusedCodeRemover(programAst, errors, compTarget, ::loadAsmIncludeFile)
+    remover.visit(programAst)
+    remover.applyModifications()
+    errors.report()
+}
+
+private fun postprocessAst(programAst: Program, errors: IErrorReporter, compilerOptions: CompilationOptions) {
+    programAst.addTypecasts(errors)
+    errors.report()
+    programAst.variousCleanups()
+    programAst.checkValid(compilerOptions, errors, compilerOptions.compTarget)          // check if final tree is still valid
+    errors.report()
+    val callGraph = CallGraph(programAst, ::loadAsmIncludeFile)
+    callGraph.checkRecursiveCalls(errors)
+    errors.report()
+    programAst.verifyFunctionArgTypes()
+    programAst.moveMainAndStartToFirst()
+}
+
+private fun writeAssembly(programAst: Program,
+                          errors: IErrorReporter,
+                          outputDir: Path,
+                          compilerOptions: CompilationOptions): String {
+    // asm generation directly from the Ast
+    programAst.processAstBeforeAsmGeneration(errors, compilerOptions.compTarget)
+    errors.report()
+
+    // printAst(programAst)
+
+    compilerOptions.compTarget.machine.initializeZeropage(compilerOptions)
+    val assembly = asmGeneratorFor(compilerOptions.compTarget,
+            programAst,
+            errors,
+            compilerOptions.compTarget.machine.zeropage,
+            compilerOptions,
+            outputDir).compileToAssembly()
+    assembly.assemble(compilerOptions)
+    errors.report()
+    return assembly.name
+}
+
+fun printAst(programAst: Program) {
+    println()
+    val printer = AstToSourceCode(::print, programAst)
+    printer.visit(programAst)
+    println()
 }
 
 fun loadAsmIncludeFile(filename: String, source: Path): String {
     return if (filename.startsWith("library:")) {
         val resource = tryGetEmbeddedResource(filename.substring(8))
-                ?: throw IllegalArgumentException("library file '$filename' not found")
+            ?: throw IllegalArgumentException("library file '$filename' not found")
         resource.bufferedReader().use { it.readText() }
     } else {
         // first try in the isSameAs folder as where the containing file was imported from

compiler/src/prog8/compiler/ErrorReporting.kt

@@ -1,9 +1,18 @@
-package prog8.ast.base
+package prog8.compiler
 
+import prog8.ast.base.Position
 import prog8.parser.ParsingFailedError
 
 
-class ErrorReporter {
+interface IErrorReporter {
+    fun err(msg: String, position: Position)
+    fun warn(msg: String, position: Position)
+    fun isEmpty(): Boolean
+    fun report()
+}
+
+
+internal class ErrorReporter: IErrorReporter {
     private enum class MessageSeverity {
         WARNING,
         ERROR
@@ -13,10 +22,14 @@ class ErrorReporter {
     private val messages = mutableListOf<CompilerMessage>()
     private val alreadyReportedMessages = mutableSetOf<String>()
 
-    fun err(msg: String, position: Position) = messages.add(CompilerMessage(MessageSeverity.ERROR, msg, position))
-    fun warn(msg: String, position: Position) = messages.add(CompilerMessage(MessageSeverity.WARNING, msg, position))
+    override fun err(msg: String, position: Position) {
+        messages.add(CompilerMessage(MessageSeverity.ERROR, msg, position))
+    }
+    override fun warn(msg: String, position: Position) {
+        messages.add(CompilerMessage(MessageSeverity.WARNING, msg, position))
+    }
 
-    fun handle() {
+    override fun report() {
         var numErrors = 0
         var numWarnings = 0
         messages.forEach {
@@ -40,5 +53,5 @@ class ErrorReporter {
             throw ParsingFailedError("There are $numErrors errors and $numWarnings warnings.")
     }
 
-    fun isEmpty() = messages.isEmpty()
+    override fun isEmpty() = messages.isEmpty()
 }

compiler/src/prog8/compiler/Main.kt

@@ -1,246 +0,0 @@
-package prog8.compiler
-
-import prog8.ast.AstToSourceCode
-import prog8.ast.Program
-import prog8.ast.base.*
-import prog8.ast.statements.Directive
-import prog8.compiler.target.C64Target
-import prog8.compiler.target.CompilationTarget
-import prog8.compiler.target.Cx16Target
-import prog8.optimizer.*
-import prog8.optimizer.UnusedCodeRemover
-import prog8.optimizer.constantFold
-import prog8.optimizer.optimizeStatements
-import prog8.optimizer.simplifyExpressions
-import prog8.parser.ModuleImporter
-import prog8.parser.ParsingFailedError
-import prog8.parser.moduleName
-import java.nio.file.Path
-import kotlin.system.exitProcess
-import kotlin.system.measureTimeMillis
-
-
-class CompilationResult(val success: Boolean,
-                        val programAst: Program,
-                        val programName: String,
-                        val importedFiles: List<Path>)
-
-
-fun compileProgram(filepath: Path,
-                   optimize: Boolean,
-                   writeAssembly: Boolean,
-                   slowCodegenWarnings: Boolean,
-                   compilationTarget: String,
-                   outputDir: Path): CompilationResult {
-    var programName = ""
-    lateinit var programAst: Program
-    lateinit var importedFiles: List<Path>
-    val errors = ErrorReporter()
-
-    when(compilationTarget) {
-        C64Target.name -> CompilationTarget.instance = C64Target
-        Cx16Target.name -> CompilationTarget.instance = Cx16Target
-        else -> {
-            System.err.println("invalid compilation target")
-            exitProcess(1)
-        }
-    }
-
-    try {
-        val totalTime = measureTimeMillis {
-            // import main module and everything it needs
-            val (ast, compilationOptions, imported) = parseImports(filepath, errors)
-            compilationOptions.slowCodegenWarnings = slowCodegenWarnings
-            programAst = ast
-            importedFiles = imported
-            processAst(programAst, errors, compilationOptions)
-            if (optimize)
-                optimizeAst(programAst, errors)
-            postprocessAst(programAst, errors, compilationOptions)
-
-            // printAst(programAst)
-
-            if(writeAssembly)
-                programName = writeAssembly(programAst, errors, outputDir, optimize, compilationOptions)
-        }
-        System.out.flush()
-        System.err.flush()
-        println("\nTotal compilation+assemble time: ${totalTime / 1000.0} sec.")
-        return CompilationResult(true, programAst, programName, importedFiles)
-
-    } catch (px: ParsingFailedError) {
-        System.err.print("\u001b[91m")  // bright red
-        System.err.println(px.message)
-        System.err.print("\u001b[0m")  // reset
-    } catch (ax: AstException) {
-        System.err.print("\u001b[91m")  // bright red
-        System.err.println(ax.toString())
-        System.err.print("\u001b[0m")  // reset
-    } catch (x: Exception) {
-        print("\u001b[91m")  // bright red
-        println("\n* internal error *")
-        print("\u001b[0m")  // reset
-        System.out.flush()
-        throw x
-    } catch (x: NotImplementedError) {
-        print("\u001b[91m")  // bright red
-        println("\n* internal error: missing feature/code *")
-        print("\u001b[0m")  // reset
-        System.out.flush()
-        throw x
-    }
-
-    return CompilationResult(false, Program("failed", mutableListOf()), programName, emptyList())
-}
-
-private fun parseImports(filepath: Path, errors: ErrorReporter): Triple<Program, CompilationOptions, List<Path>> {
-    println("Compiler target: ${CompilationTarget.instance.name}. Parsing...")
-    val importer = ModuleImporter()
-    val programAst = Program(moduleName(filepath.fileName), mutableListOf())
-    importer.importModule(programAst, filepath)
-    errors.handle()
-
-    val importedFiles = programAst.modules.filter { !it.source.startsWith("@embedded@") }.map { it.source }
-
-    val compilerOptions = determineCompilationOptions(programAst)
-    if (compilerOptions.launcher == LauncherType.BASIC && compilerOptions.output != OutputType.PRG)
-        throw ParsingFailedError("${programAst.modules.first().position} BASIC launcher requires output type PRG.")
-
-    // depending on the machine and compiler options we may have to include some libraries
-    CompilationTarget.instance.machine.importLibs(compilerOptions, importer, programAst)
-
-    // always import prog8_lib and math
-    importer.importLibraryModule(programAst, "math")
-    importer.importLibraryModule(programAst, "prog8_lib")
-    errors.handle()
-    return Triple(programAst, compilerOptions, importedFiles)
-}
-
-private fun determineCompilationOptions(program: Program): CompilationOptions {
-    val mainModule = program.modules.first()
-    val outputType = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%output" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    val launcherType = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%launcher" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    val zpoption: String? = (mainModule.statements.singleOrNull { it is Directive && it.directive == "%zeropage" }
-            as? Directive)?.args?.single()?.name?.toUpperCase()
-    val allOptions = program.modules.flatMap { it.statements }.filter { it is Directive && it.directive == "%option" }.flatMap { (it as Directive).args }.toSet()
-    val floatsEnabled = allOptions.any { it.name == "enable_floats" }
-    val noSysInit = allOptions.any { it.name == "no_sysinit" }
-    var zpType: ZeropageType =
-            if (zpoption == null)
-                if(floatsEnabled) ZeropageType.FLOATSAFE else ZeropageType.KERNALSAFE
-            else
-                try {
-                    ZeropageType.valueOf(zpoption)
-                } catch (x: IllegalArgumentException) {
-                    ZeropageType.KERNALSAFE
-                    // error will be printed by the astchecker
-                }
-
-    if (zpType==ZeropageType.FLOATSAFE && CompilationTarget.instance.name == Cx16Target.name) {
-        System.err.println("Warning: Cx16 target must use zp option basicsafe instead of floatsafe")
-        zpType = ZeropageType.BASICSAFE
-    }
-
-    val zpReserved = mainModule.statements
-            .asSequence()
-            .filter { it is Directive && it.directive == "%zpreserved" }
-            .map { (it as Directive).args }
-            .map { it[0].int!!..it[1].int!! }
-            .toList()
-
-    if(outputType!=null && !OutputType.values().any {it.name==outputType}) {
-        System.err.println("invalid output type $outputType")
-        exitProcess(1)
-    }
-    if(launcherType!=null && !LauncherType.values().any {it.name==launcherType}) {
-        System.err.println("invalid launcher type $launcherType")
-        exitProcess(1)
-    }
-
-    return CompilationOptions(
-            if (outputType == null) OutputType.PRG else OutputType.valueOf(outputType),
-            if (launcherType == null) LauncherType.BASIC else LauncherType.valueOf(launcherType),
-            zpType, zpReserved, floatsEnabled, noSysInit
-    )
-}
-
-private fun processAst(programAst: Program, errors: ErrorReporter, compilerOptions: CompilationOptions) {
-    // perform initial syntax checks and processings
-    println("Processing for target ${CompilationTarget.instance.name}...")
-    programAst.checkIdentifiers(errors)
-    errors.handle()
-    programAst.constantFold(errors)
-    errors.handle()
-    programAst.reorderStatements(errors)
-    errors.handle()
-    programAst.addTypecasts(errors)
-    errors.handle()
-    programAst.variousCleanups()
-    programAst.checkValid(compilerOptions, errors)
-    errors.handle()
-    programAst.checkIdentifiers(errors)
-    errors.handle()
-}
-
-private fun optimizeAst(programAst: Program, errors: ErrorReporter) {
-    // optimize the parse tree
-    println("Optimizing...")
-    while (true) {
-        // keep optimizing expressions and statements until no more steps remain
-        val optsDone1 = programAst.simplifyExpressions()
-        val optsDone2 = programAst.splitBinaryExpressions()
-        val optsDone3 = programAst.optimizeStatements(errors)
-        programAst.constantFold(errors) // because simplified statements and expressions can result in more constants that can be folded away
-        errors.handle()
-        if (optsDone1 + optsDone2 + optsDone3 == 0)
-            break
-    }
-
-    val remover = UnusedCodeRemover(programAst, errors)
-    remover.visit(programAst)
-    remover.applyModifications()
-    errors.handle()
-}
-
-private fun postprocessAst(programAst: Program, errors: ErrorReporter, compilerOptions: CompilationOptions) {
-    programAst.addTypecasts(errors)
-    errors.handle()
-    programAst.variousCleanups()
-    programAst.checkValid(compilerOptions, errors)          // check if final tree is still valid
-    errors.handle()
-    val callGraph = CallGraph(programAst)
-    callGraph.checkRecursiveCalls(errors)
-    errors.handle()
-    programAst.verifyFunctionArgTypes()
-    programAst.moveMainAndStartToFirst()
-}
-
-private fun writeAssembly(programAst: Program, errors: ErrorReporter, outputDir: Path,
-                          optimize: Boolean, compilerOptions: CompilationOptions): String {
-    // asm generation directly from the Ast,
-    programAst.processAstBeforeAsmGeneration(errors)
-    errors.handle()
-
-    // printAst(programAst)
-
-    CompilationTarget.instance.machine.initializeZeropage(compilerOptions)
-    val assembly = CompilationTarget.instance.asmGenerator(
-            programAst,
-            errors,
-            CompilationTarget.instance.machine.zeropage,
-            compilerOptions,
-            outputDir).compileToAssembly(optimize)
-    assembly.assemble(compilerOptions)
-    errors.handle()
-    return assembly.name
-}
-
-fun printAst(programAst: Program) {
-    println()
-    val printer = AstToSourceCode(::print, programAst)
-    printer.visit(programAst)
-    println()
-}
-

compiler/src/prog8/compiler/Zeropage.kt

@@ -21,7 +21,7 @@ abstract class Zeropage(protected val options: CompilationOptions) {
 
     fun available() = if(options.zeropage==ZeropageType.DONTUSE) 0 else free.size
 
-    fun allocate(scopedname: String, datatype: DataType, position: Position?, errors: ErrorReporter): Int {
+    fun allocate(scopedname: String, datatype: DataType, position: Position?, errors: IErrorReporter): Int {
         assert(scopedname.isEmpty() || !allocations.values.any { it.first==scopedname } ) {"scopedname can't be allocated twice"}
 
         if(options.zeropage==ZeropageType.DONTUSE)

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

@@ -1,4 +1,4 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.INameScope
 import prog8.ast.Module
@@ -6,16 +6,21 @@ import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
 import prog8.ast.statements.*
+import prog8.ast.walk.IAstVisitor
 import prog8.compiler.CompilationOptions
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+import prog8.compiler.functions.builtinFunctionReturnType
 import prog8.compiler.target.C64Target
-import prog8.compiler.target.CompilationTarget
 import prog8.compiler.target.Cx16Target
-import prog8.functions.BuiltinFunctions
+import prog8.compiler.target.ICompilationTarget
 import java.io.File
 
 internal class AstChecker(private val program: Program,
                           private val compilerOptions: CompilationOptions,
-                          private val errors: ErrorReporter) : IAstVisitor {
+                          private val errors: IErrorReporter,
+                          private val compTarget: ICompilationTarget
+) : IAstVisitor {
 
     override fun visit(program: Program) {
         assert(program === this.program)
@@ -34,37 +39,6 @@ internal class AstChecker(private val program: Program,
                 if (startSub.parameters.isNotEmpty() || startSub.returntypes.isNotEmpty())
                     errors.err("program entrypoint subroutine can't have parameters and/or return values", startSub.position)
             }
-
-            // the main module cannot contain 'regular' statements (they will never be executed!)
-            for (statement in mainBlock.statements) {
-                val ok = when (statement) {
-                    is Block -> true
-                    is Directive -> true
-                    is Label -> true
-                    is VarDecl -> true
-                    is InlineAssembly -> true
-                    is INameScope -> true
-                    is NopStatement -> true
-                    else -> false
-                }
-                if (!ok) {
-                    errors.err("main block contains regular statements, this is not allowed (they'll never get executed). Use subroutines.", statement.position)
-                    break
-                }
-            }
-        }
-
-        // there can be an optional single 'irq' block with a 'irq' subroutine in it,
-        // which will be used as the 60hz irq routine in the vm if it's present
-        val irqBlocks = program.modules.flatMap { it.statements }.filter { it is Block && it.name=="irq" }.map { it as Block }
-        if(irqBlocks.size>1)
-            errors.err("more than one 'irq' block", irqBlocks[0].position)
-        for(irqBlock in irqBlocks) {
-            val irqSub = irqBlock.subScope("irq") as? Subroutine
-            if (irqSub != null) {
-                if (irqSub.parameters.isNotEmpty() || irqSub.returntypes.isNotEmpty())
-                    errors.err("irq entrypoint subroutine can't have parameters and/or return values", irqSub.position)
-            }
         }
 
         super.visit(program)
@@ -116,7 +90,7 @@ internal class AstChecker(private val program: Program,
         if(iterableDt !in IterableDatatypes && forLoop.iterable !is RangeExpr) {
             errors.err("can only loop over an iterable type", forLoop.position)
         } else {
-            val loopvar = forLoop.loopVar.targetVarDecl(program.namespace)
+            val loopvar = forLoop.loopVar.targetVarDecl(program)
             if(loopvar==null || loopvar.type== VarDeclType.CONST) {
                 errors.err("for loop requires a variable to loop with", forLoop.position)
             } else {
@@ -167,24 +141,44 @@ internal class AstChecker(private val program: Program,
     }
 
     override fun visit(jump: Jump) {
-        if(jump.identifier!=null) {
-            val targetStatement = checkFunctionOrLabelExists(jump.identifier, jump)
+        val ident = jump.identifier
+        if(ident!=null) {
+            val targetStatement = checkFunctionOrLabelExists(ident, jump)
             if(targetStatement!=null) {
                 if(targetStatement is BuiltinFunctionStatementPlaceholder)
                     errors.err("can't jump to a builtin function", jump.position)
             }
         }
 
-        if(jump.address!=null && (jump.address < 0 || jump.address > 65535))
+        val addr = jump.address
+        if(addr!=null && (addr < 0 || addr > 65535))
             errors.err("jump address must be valid integer 0..\$ffff", jump.position)
         super.visit(jump)
     }
 
     override fun visit(block: Block) {
-        if(block.address!=null && (block.address<0 || block.address>65535)) {
+        val addr = block.address
+        if(addr!=null && (addr<0 || addr>65535)) {
             errors.err("block memory address must be valid integer 0..\$ffff", block.position)
         }
 
+        for (statement in block.statements) {
+            val ok = when (statement) {
+                is Block,
+                is Directive,
+                is Label,
+                is VarDecl,
+                is InlineAssembly,
+                is INameScope,
+                is NopStatement -> true
+                else -> false
+            }
+            if (!ok) {
+                errors.err("non-declarative statement occurs in block scope, where it will never be executed. Move it to a subroutine instead.", statement.position)
+                break
+            }
+        }
+
         super.visit(block)
     }
 
@@ -202,10 +196,20 @@ internal class AstChecker(private val program: Program,
         if(subroutine.name in BuiltinFunctions)
             err("cannot redefine a built-in function")
 
+        if(subroutine.parameters.size>16)
+            err("subroutines are limited to 16 parameters")
+
         val uniqueNames = subroutine.parameters.asSequence().map { it.name }.toSet()
         if(uniqueNames.size!=subroutine.parameters.size)
             err("parameter names must be unique")
 
+        if(subroutine.inline) {
+            if (subroutine.containsDefinedVariables())
+                err("can't inline a subroutine that defines variables")
+            if (!subroutine.isAsmSubroutine && subroutine.parameters.isNotEmpty())
+                err("can't inline a non-asm subroutine that has parameters")
+        }
+
         super.visit(subroutine)
 
         // user-defined subroutines can only have zero or one return type
@@ -214,13 +218,13 @@ internal class AstChecker(private val program: Program,
             err("subroutines can only have one return value")
 
         // subroutine must contain at least one 'return' or 'goto'
-        // (or if it has an asm block, that must contain a 'rts' or 'jmp')
+        // (or if it has an asm block, that must contain a 'rts' or 'jmp' or 'bra')
         if(subroutine.statements.count { it is Return || it is Jump } == 0) {
             if (subroutine.amountOfRtsInAsm() == 0) {
                 if (subroutine.returntypes.isNotEmpty()) {
                     // for asm subroutines with an address, no statement check is possible.
-                    if (subroutine.asmAddress == null)
-                        err("subroutine has result value(s) and thus must have at least one 'return' or 'goto' in it (or 'rts' / 'jmp' in case of %asm)")
+                    if (subroutine.asmAddress == null && !subroutine.inline)
+                        err("non-inline subroutine has result value(s) and thus must have at least one 'return' or 'goto' in it (or rts/jmp/bra in case of %asm)")
                 }
             }
         }
@@ -288,9 +292,28 @@ internal class AstChecker(private val program: Program,
                             regCounts[CpuRegister.X]=regCounts.getValue(CpuRegister.X)+1
                             regCounts[CpuRegister.Y]=regCounts.getValue(CpuRegister.Y)+1
                         }
-                        null ->
-                            if(p.statusflag!=null)
-                                statusflagCounts[p.statusflag] = statusflagCounts.getValue(p.statusflag) + 1
+                        RegisterOrPair.FAC1, RegisterOrPair.FAC2 -> { /* no sensible way to count this */ }
+                        RegisterOrPair.R0,
+                        RegisterOrPair.R1,
+                        RegisterOrPair.R2,
+                        RegisterOrPair.R3,
+                        RegisterOrPair.R4,
+                        RegisterOrPair.R5,
+                        RegisterOrPair.R6,
+                        RegisterOrPair.R7,
+                        RegisterOrPair.R8,
+                        RegisterOrPair.R9,
+                        RegisterOrPair.R10,
+                        RegisterOrPair.R11,
+                        RegisterOrPair.R12,
+                        RegisterOrPair.R13,
+                        RegisterOrPair.R14,
+                        RegisterOrPair.R15 -> { /* no sensible way to count this */ }
+                        null -> {
+                            val statusf = p.statusflag
+                            if (statusf != null)
+                                statusflagCounts[statusf] = statusflagCounts.getValue(statusf) + 1
+                        }
                     }
                 }
             }
@@ -315,10 +338,6 @@ internal class AstChecker(private val program: Program,
             if(statusFlagsNoCarry.isNotEmpty())
                 err("can only use Carry as status flag parameter")
 
-            val carryParameter = subroutine.asmParameterRegisters.singleOrNull { it.statusflag==Statusflag.Pc }
-            if(carryParameter!=null && carryParameter !== subroutine.asmParameterRegisters.last())
-                err("carry parameter has to come last")
-
         } else {
             // Pass-by-reference datatypes can not occur as parameters to a subroutine directly
             // Instead, their reference (address) should be passed (as an UWORD).
@@ -340,15 +359,22 @@ internal class AstChecker(private val program: Program,
         super.visit(whileLoop)
     }
 
+    override fun visit(repeatLoop: RepeatLoop) {
+        val iterations = repeatLoop.iterations?.constValue(program)
+        if(iterations != null && iterations.number.toInt() > 65535)
+            errors.err("repeat cannot go over 65535 iterations", iterations.position)
+        super.visit(repeatLoop)
+    }
+
     override fun visit(assignment: Assignment) {
         if(assignment.value is FunctionCall) {
-            val stmt = (assignment.value as FunctionCall).target.targetStatement(program.namespace)
+            val stmt = (assignment.value as FunctionCall).target.targetStatement(program)
             if (stmt is Subroutine) {
                 val idt = assignment.target.inferType(program)
                 if(!idt.isKnown) {
                      errors.err("return type mismatch", assignment.value.position)
                 }
-                if(stmt.returntypes.size <= 1 && stmt.returntypes.single() isNotAssignableTo idt.typeOrElse(DataType.BYTE)) {
+                if(stmt.returntypes.isEmpty() || (stmt.returntypes.size == 1 && stmt.returntypes.single() isNotAssignableTo idt.typeOrElse(DataType.BYTE))) {
                     errors.err("return type mismatch", assignment.value.position)
                 }
             }
@@ -356,7 +382,7 @@ internal class AstChecker(private val program: Program,
 
         val targetIdent = assignment.target.identifier
         if(targetIdent!=null) {
-            val targetVar = targetIdent.targetVarDecl(program.namespace)
+            val targetVar = targetIdent.targetVarDecl(program)
             if(targetVar?.struct != null) {
                 val sourceStructLv = assignment.value as? ArrayLiteralValue
                 if (sourceStructLv != null) {
@@ -365,7 +391,7 @@ internal class AstChecker(private val program: Program,
                 } else {
                     val sourceIdent = assignment.value as? IdentifierReference
                     if (sourceIdent != null) {
-                        val sourceVar = sourceIdent.targetVarDecl(program.namespace)
+                        val sourceVar = sourceIdent.targetVarDecl(program)
                         if (sourceVar?.struct != null) {
                             if (sourceVar.struct !== targetVar.struct)
                                 errors.err("assignment of different struct types", assignment.position)
@@ -383,7 +409,7 @@ internal class AstChecker(private val program: Program,
         if(valueDt.isKnown && !(valueDt isAssignableTo targetDt)) {
             if(targetDt.typeOrElse(DataType.STRUCT) in IterableDatatypes)
                 errors.err("cannot assign value to string or array", assignment.value.position)
-            else
+            else if(!(valueDt.istype(DataType.STR) && targetDt.istype(DataType.UWORD)))
                 errors.err("value's type doesn't match target", assignment.value.position)
         }
 
@@ -453,16 +479,13 @@ internal class AstChecker(private val program: Program,
     }
 
     override fun visit(addressOf: AddressOf) {
-        val variable=addressOf.identifier.targetVarDecl(program.namespace)
-        if(variable==null)
-            errors.err("pointer-of operand must be the name of a heap variable", addressOf.position)
-        else {
-            if(variable.datatype !in ArrayDatatypes
-                    && variable.type!=VarDeclType.MEMORY
-                    && variable.struct == null
-                    && variable.datatype != DataType.STR && variable.datatype!=DataType.STRUCT)
+        val variable=addressOf.identifier.targetVarDecl(program)
+        if(variable!=null
+            && variable.datatype !in ArrayDatatypes
+            && variable.type!=VarDeclType.MEMORY
+            && variable.struct == null
+            && variable.datatype != DataType.STR && variable.datatype!=DataType.STRUCT)
                 errors.err("invalid pointer-of operand type", addressOf.position)
-        }
         super.visit(addressOf)
     }
 
@@ -470,7 +493,7 @@ internal class AstChecker(private val program: Program,
         fun err(msg: String, position: Position?=null) = errors.err(msg, position ?: decl.position)
 
         // the initializer value can't refer to the variable itself (recursive definition)
-        if(decl.value?.referencesIdentifier(decl.name) == true || decl.arraysize?.indexVar?.referencesIdentifier(decl.name) == true)
+        if(decl.value?.referencesIdentifier(decl.name) == true || decl.arraysize?.indexExpr?.referencesIdentifier(decl.name) == true)
             err("recursive var declaration")
 
         // CONST can only occur on simple types (byte, word, float)
@@ -584,7 +607,7 @@ internal class AstChecker(private val program: Program,
                         err("memory address must be valid integer 0..\$ffff", decl.value?.position)
                     }
                 } else {
-                    err("value of memory mapped variable can only be a number, perhaps you meant to use an address pointer type instead?", decl.value?.position)
+                    err("value of memory mapped variable can only be a fixed number, perhaps you meant to use an address pointer type instead?", decl.value?.position)
                 }
             }
         }
@@ -603,23 +626,28 @@ internal class AstChecker(private val program: Program,
             }
         }
 
-        // array length limits
+        // array length limits and constant lenghts
         if(decl.isArray) {
-            val length = decl.arraysize!!.constIndex() ?: 1
-            when (decl.datatype) {
-                DataType.STR, DataType.ARRAY_UB, DataType.ARRAY_B -> {
-                    if(length==0 || length>256)
-                        err("string and byte array length must be 1-256")
+            val length = decl.arraysize!!.constIndex()
+            if(length==null)
+                err("array length must be a constant")
+            else {
+                when (decl.datatype) {
+                    DataType.STR, DataType.ARRAY_UB, DataType.ARRAY_B -> {
+                        if (length == 0 || length > 256)
+                            err("string and byte array length must be 1-256")
+                    }
+                    DataType.ARRAY_UW, DataType.ARRAY_W -> {
+                        if (length == 0 || length > 128)
+                            err("word array length must be 1-128")
+                    }
+                    DataType.ARRAY_F -> {
+                        if (length == 0 || length > 51)
+                            err("float array length must be 1-51")
+                    }
+                    else -> {
+                    }
                 }
-                DataType.ARRAY_UW, DataType.ARRAY_W -> {
-                    if(length==0 || length>128)
-                        err("word array length must be 1-128")
-                }
-                DataType.ARRAY_F -> {
-                    if(length==0 || length>51)
-                        err("float array length must be 1-51")
-                }
-                else -> {}
             }
         }
 
@@ -711,7 +739,7 @@ internal class AstChecker(private val program: Program,
                     err("this directive may only occur in a block or at module level")
                 if(directive.args.isEmpty())
                     err("missing option directive argument(s)")
-                else if(directive.args.map{it.name in setOf("enable_floats", "force_output", "no_sysinit")}.any { !it })
+                else if(directive.args.map{it.name in setOf("enable_floats", "force_output", "no_sysinit", "align_word", "align_page")}.any { !it })
                     err("invalid option directive argument(s)")
             }
             "%target" -> {
@@ -746,7 +774,7 @@ internal class AstChecker(private val program: Program,
 
         fun isPassByReferenceElement(e: Expression): Boolean {
             if(e is IdentifierReference) {
-                val decl = e.targetVarDecl(program.namespace)!!
+                val decl = e.targetVarDecl(program)!!
                 return decl.datatype in PassByReferenceDatatypes
             }
             return e is StringLiteralValue
@@ -754,7 +782,7 @@ internal class AstChecker(private val program: Program,
 
         if(array.parent is VarDecl) {
             if (!array.value.all { it is NumericLiteralValue || it is AddressOf || isPassByReferenceElement(it) })
-                errors.err("array literal for variable initialization contains invalid types", array.position)
+                errors.err("array literal for variable initialization contains non-constant elements", array.position)
         } else if(array.parent is ForLoop) {
             if (!array.value.all { it.constValue(program) != null })
                 errors.err("array literal for iteration must contain constants. Try using a separate array variable instead?", array.position)
@@ -843,6 +871,10 @@ internal class AstChecker(private val program: Program,
     override fun visit(typecast: TypecastExpression) {
         if(typecast.type in IterableDatatypes)
             errors.err("cannot type cast to string or array type", typecast.position)
+
+        if(!typecast.expression.inferType(program).isKnown)
+            errors.err("this expression doesn't return a value", typecast.expression.position)
+
         super.visit(typecast)
     }
 
@@ -895,12 +927,12 @@ internal class AstChecker(private val program: Program,
                 errors.warn("sgn() of unsigned type is always 0 or 1, this is perhaps not what was intended", functionCall.args.first().position)
         }
 
-        val error = VerifyFunctionArgTypes.checkTypes(functionCall, functionCall.definingScope(), program)
+        val error = VerifyFunctionArgTypes.checkTypes(functionCall, program)
         if(error!=null)
             errors.err(error, functionCall.position)
 
         // check the functions that return multiple returnvalues.
-        val stmt = functionCall.target.targetStatement(program.namespace)
+        val stmt = functionCall.target.targetStatement(program)
         if (stmt is Subroutine) {
             if (stmt.returntypes.size > 1) {
                 // Currently, it's only possible to handle ONE (or zero) return values from a subroutine.
@@ -921,6 +953,20 @@ internal class AstChecker(private val program: Program,
             }
         }
 
+        // functions that don't return a value, can't be used in an expression or assignment
+        if(targetStatement is Subroutine) {
+            if(targetStatement.returntypes.isEmpty()) {
+                if(functionCall.parent is Expression || functionCall.parent is Assignment)
+                    errors.err("subroutine doesn't return a value", functionCall.position)
+            }
+        }
+        else if(targetStatement is BuiltinFunctionStatementPlaceholder) {
+            if(builtinFunctionReturnType(targetStatement.name, functionCall.args, program).isUnknown) {
+                if(functionCall.parent is Expression || functionCall.parent is Assignment)
+                    errors.err("function doesn't return a value", functionCall.position)
+            }
+        }
+
         super.visit(functionCall)
     }
 
@@ -928,11 +974,19 @@ internal class AstChecker(private val program: Program,
         val targetStatement = checkFunctionOrLabelExists(functionCallStatement.target, functionCallStatement)
         if(targetStatement!=null)
             checkFunctionCall(targetStatement, functionCallStatement.args, functionCallStatement.position)
-        if(!functionCallStatement.void && targetStatement is Subroutine && targetStatement.returntypes.isNotEmpty()) {
-            if(targetStatement.returntypes.size==1)
-                errors.warn("result value of subroutine call is discarded (use void?)", functionCallStatement.position)
-            else
-                errors.warn("result values of subroutine call are discarded (use void?)", functionCallStatement.position)
+        if (!functionCallStatement.void) {
+            // check for unused return values
+            if (targetStatement is Subroutine && targetStatement.returntypes.isNotEmpty()) {
+                if(targetStatement.returntypes.size==1)
+                    errors.warn("result value of subroutine call is discarded (use void?)", functionCallStatement.position)
+                else
+                    errors.warn("result values of subroutine call are discarded (use void?)", functionCallStatement.position)
+            }
+            else if(targetStatement is BuiltinFunctionStatementPlaceholder) {
+                val rt = builtinFunctionReturnType(targetStatement.name, functionCallStatement.args, program)
+                if(rt.isKnown)
+                    errors.warn("result value of a function call is discarded (use void?)", functionCallStatement.position)
+            }
         }
 
         if(functionCallStatement.target.nameInSource.last() == "sort") {
@@ -950,7 +1004,8 @@ internal class AstChecker(private val program: Program,
             }
         }
 
-        val error = VerifyFunctionArgTypes.checkTypes(functionCallStatement, functionCallStatement.definingScope(), program)
+        val error =
+            VerifyFunctionArgTypes.checkTypes(functionCallStatement, program)
         if(error!=null) {
             errors.err(error, functionCallStatement.args.firstOrNull()?.position ?: functionCallStatement.position)
         }
@@ -977,7 +1032,7 @@ internal class AstChecker(private val program: Program,
                     errors.err("swap requires args of numerical type", position)
             }
             else if(target.name=="all" || target.name=="any") {
-                if((args[0] as? AddressOf)?.identifier?.targetVarDecl(program.namespace)?.datatype == DataType.STR) {
+                if((args[0] as? AddressOf)?.identifier?.targetVarDecl(program)?.datatype == DataType.STR) {
                     errors.err("any/all on a string is useless (is always true unless the string is empty)", position)
                 }
                 if(args[0].inferType(program).typeOrElse(DataType.STR) == DataType.STR) {
@@ -985,14 +1040,34 @@ internal class AstChecker(private val program: Program,
                 }
             }
         } else if(target is Subroutine) {
-            if(target.regXasResult())
-                errors.warn("subroutine call return value in X register is discarded and replaced by 0", position)
             if(target.isAsmSubroutine) {
                 for (arg in args.zip(target.parameters)) {
                     val argIDt = arg.first.inferType(program)
                     if (!argIDt.isKnown)
                         return
                 }
+
+                // check that cx16 virtual registers aren't used as arguments in a conflicting way
+                val params = target.asmParameterRegisters.withIndex().toList()
+                for(arg in args.withIndex()) {
+                    var ident: IdentifierReference? = null
+                    if(arg.value is IdentifierReference)
+                        ident = arg.value as IdentifierReference
+                    else if(arg.value is FunctionCall) {
+                        val fcall = arg.value as FunctionCall
+                        if(fcall.target.nameInSource == listOf("lsb") || fcall.target.nameInSource == listOf("msb"))
+                            ident = fcall.args[0] as? IdentifierReference
+                    }
+                    if(ident!=null && ident.nameInSource[0] == "cx16" && ident.nameInSource[1].startsWith("r")) {
+                        val reg = RegisterOrPair.valueOf(ident.nameInSource[1].toUpperCase())
+                        val same = params.filter { it.value.registerOrPair==reg }
+                        for(s in same) {
+                            if(s.index!=arg.index) {
+                                errors.err("conflicting register $reg used as argument but is also a target register for another parameter", ident.position)
+                            }
+                        }
+                    }
+                }
             }
         }
     }
@@ -1012,7 +1087,7 @@ internal class AstChecker(private val program: Program,
                 }
             }
         } else if(postIncrDecr.target.arrayindexed != null) {
-            val target = postIncrDecr.target.arrayindexed?.arrayvar?.targetStatement(program.namespace)
+            val target = postIncrDecr.target.arrayindexed?.arrayvar?.targetStatement(program)
             if(target==null) {
                 errors.err("undefined symbol", postIncrDecr.position)
             }
@@ -1027,7 +1102,7 @@ internal class AstChecker(private val program: Program,
     }
 
     override fun visit(arrayIndexedExpression: ArrayIndexedExpression) {
-        val target = arrayIndexedExpression.arrayvar.targetStatement(program.namespace)
+        val target = arrayIndexedExpression.arrayvar.targetStatement(program)
         if(target is VarDecl) {
             if(target.datatype !in IterableDatatypes)
                 errors.err("indexing requires an iterable variable", arrayIndexedExpression.position)
@@ -1050,15 +1125,9 @@ internal class AstChecker(private val program: Program,
             errors.err("indexing requires a variable to act upon", arrayIndexedExpression.position)
 
         // check index value 0..255
-        val dtxNum = arrayIndexedExpression.indexer.indexNum?.inferType(program)?.typeOrElse(DataType.STRUCT)
-        if(dtxNum!=null && dtxNum != DataType.UBYTE && dtxNum != DataType.BYTE)
+        val dtxNum = arrayIndexedExpression.indexer.indexExpr.inferType(program)
+        if(!dtxNum.istype(DataType.UBYTE) && !dtxNum.istype(DataType.BYTE))
             errors.err("array indexing is limited to byte size 0..255", arrayIndexedExpression.position)
-        val dtxVar = arrayIndexedExpression.indexer.indexVar?.inferType(program)?.typeOrElse(DataType.STRUCT)
-        if(dtxVar!=null && dtxVar != DataType.UBYTE && dtxVar != DataType.BYTE)
-            errors.err("array indexing is limited to byte size 0..255", arrayIndexedExpression.position)
-
-        if(arrayIndexedExpression.indexer.origExpression!=null)
-            throw FatalAstException("array indexer should have been replaced with a temp var @ ${arrayIndexedExpression.indexer.position}")
 
         super.visit(arrayIndexedExpression)
     }
@@ -1067,12 +1136,18 @@ internal class AstChecker(private val program: Program,
         val conditionType = whenStatement.condition.inferType(program).typeOrElse(DataType.STRUCT)
         if(conditionType !in IntegerDatatypes)
             errors.err("when condition must be an integer value", whenStatement.position)
-        val choiceValues = whenStatement.choiceValues(program)
-        val occurringValues = choiceValues.map {it.first}
-        val tally = choiceValues.associate { it.second to occurringValues.count { ov->it.first==ov} }
-        tally.filter { it.value>1 }.forEach {
-            errors.err("choice value occurs multiple times", it.key.position)
+        val tally = mutableSetOf<Int>()
+        for((choices, choiceNode) in whenStatement.choiceValues(program)) {
+            if(choices!=null) {
+                for (c in choices) {
+                    if(c in tally)
+                        errors.err("choice value already occurs earlier", choiceNode.position)
+                    else
+                        tally.add(c)
+                }
+            }
         }
+
         if(whenStatement.choices.isEmpty())
             errors.err("empty when statement", whenStatement.position)
 
@@ -1119,7 +1194,7 @@ internal class AstChecker(private val program: Program,
     }
 
     private fun checkFunctionOrLabelExists(target: IdentifierReference, statement: Statement): Statement? {
-        val targetStatement = target.targetStatement(program.namespace)
+        val targetStatement = target.targetStatement(program)
         if(targetStatement is Label || targetStatement is Subroutine || targetStatement is BuiltinFunctionStatementPlaceholder)
             return targetStatement
         else if(targetStatement==null)
@@ -1209,7 +1284,7 @@ internal class AstChecker(private val program: Program,
 
                     // check if the floating point values are all within range
                     val doubles = value.value.map {it.constValue(program)?.number!!.toDouble()}.toDoubleArray()
-                    if(doubles.any { it < CompilationTarget.instance.machine.FLOAT_MAX_NEGATIVE || it > CompilationTarget.instance.machine.FLOAT_MAX_POSITIVE })
+                    if(doubles.any { it < compTarget.machine.FLOAT_MAX_NEGATIVE || it > compTarget.machine.FLOAT_MAX_POSITIVE })
                         return err("floating point value overflow")
                     return true
                 }
@@ -1283,7 +1358,7 @@ internal class AstChecker(private val program: Program,
         val array = value.value.map {
             when (it) {
                 is NumericLiteralValue -> it.number.toInt()
-                is AddressOf -> it.identifier.heapId(program.namespace)
+                is AddressOf -> it.identifier.hashCode() and 0xffff
                 is TypecastExpression -> {
                     val constVal = it.expression.constValue(program)
                     val cast = constVal?.cast(it.type)
@@ -1337,12 +1412,15 @@ internal class AstChecker(private val program: Program,
                 if(sourceDatatype==DataType.STRUCT) {
                     val structLv = sourceValue as ArrayLiteralValue
                     val numValues = structLv.value.size
-                    val targetstruct = target.identifier!!.targetVarDecl(program.namespace)!!.struct!!
+                    val targetstruct = target.identifier!!.targetVarDecl(program)!!.struct!!
                     return targetstruct.numberOfElements == numValues
                 }
                 false
             }
-            else -> errors.err("cannot assign new value to variable of type $targetDatatype", position)
+            else -> {
+                errors.err("cannot assign new value to variable of type $targetDatatype", position)
+                false
+            }
         }
 
         if(result)

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

@@ -1,31 +1,32 @@
-package prog8.ast.base
+package prog8.compiler.astprocessing
 
-import prog8.ast.Module
 import prog8.ast.Program
-import prog8.ast.processing.*
+import prog8.ast.base.FatalAstException
 import prog8.ast.statements.Directive
-import prog8.compiler.CompilationOptions
 import prog8.compiler.BeforeAsmGenerationAstChanger
+import prog8.compiler.CompilationOptions
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
 
 
-internal fun Program.checkValid(compilerOptions: CompilationOptions, errors: ErrorReporter) {
-    val checker = AstChecker(this, compilerOptions, errors)
+internal fun Program.checkValid(compilerOptions: CompilationOptions, errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val checker = AstChecker(this, compilerOptions, errors, compTarget)
     checker.visit(this)
 }
 
-internal fun Program.processAstBeforeAsmGeneration(errors: ErrorReporter) {
-    val fixer = BeforeAsmGenerationAstChanger(this, errors)
+internal fun Program.processAstBeforeAsmGeneration(errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val fixer = BeforeAsmGenerationAstChanger(this, errors, compTarget)
     fixer.visit(this)
     fixer.applyModifications()
 }
 
-internal fun Program.reorderStatements(errors: ErrorReporter) {
+internal fun Program.reorderStatements(errors: IErrorReporter) {
     val reorder = StatementReorderer(this, errors)
     reorder.visit(this)
     reorder.applyModifications()
 }
 
-internal fun Program.addTypecasts(errors: ErrorReporter) {
+internal fun Program.addTypecasts(errors: IErrorReporter) {
     val caster = TypecastsAdder(this, errors)
     caster.visit(this)
     caster.applyModifications()
@@ -36,15 +37,9 @@ internal fun Program.verifyFunctionArgTypes() {
     fixer.visit(this)
 }
 
-internal fun Module.checkImportedValid() {
-    val imr = ImportedModuleDirectiveRemover()
-    imr.visit(this, this.parent)
-    imr.applyModifications()
-}
+internal fun Program.checkIdentifiers(errors: IErrorReporter, compTarget: ICompilationTarget) {
 
-internal fun Program.checkIdentifiers(errors: ErrorReporter) {
-
-    val checker2 = AstIdentifiersChecker(this, errors)
+    val checker2 = AstIdentifiersChecker(this, errors, compTarget)
     checker2.visit(this)
 
     if(errors.isEmpty()) {

compiler/src/prog8/compiler/astprocessing/AstIdentifiersChecker.kt

@@ -1,14 +1,20 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.Module
 import prog8.ast.Program
-import prog8.ast.base.*
-import prog8.ast.expressions.*
+import prog8.ast.base.DataType
+import prog8.ast.base.NumericDatatypes
+import prog8.ast.base.Position
+import prog8.ast.expressions.ArrayLiteralValue
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.expressions.StringLiteralValue
 import prog8.ast.statements.*
-import prog8.compiler.target.CompilationTarget
-import prog8.functions.BuiltinFunctions
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+import prog8.compiler.target.ICompilationTarget
 
-internal class AstIdentifiersChecker(private val program: Program, private val errors: ErrorReporter) : IAstVisitor {
+internal class AstIdentifiersChecker(private val program: Program, private val errors: IErrorReporter, private val compTarget: ICompilationTarget) : IAstVisitor {
     private var blocks = mutableMapOf<String, Block>()
 
     private fun nameError(name: String, position: Position, existing: Statement) {
@@ -22,7 +28,7 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
     }
 
     override fun visit(block: Block) {
-        if(block.name in CompilationTarget.instance.machine.opcodeNames)
+        if(block.name in compTarget.machine.opcodeNames)
             errors.err("can't use a cpu opcode name as a symbol: '${block.name}'", block.position)
 
         val existing = blocks[block.name]
@@ -31,14 +37,21 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
         else
             blocks[block.name] = block
 
+        if(!block.isInLibrary) {
+            val libraries = program.modules.filter { it.isLibraryModule }
+            val libraryBlockNames = libraries.flatMap { it.statements.filterIsInstance<Block>().map { b -> b.name } }
+            if(block.name in libraryBlockNames)
+                errors.err("block is already defined in an included library module", block.position)
+        }
+
         super.visit(block)
     }
 
     override fun visit(directive: Directive) {
         if(directive.directive=="%target") {
             val compatibleTarget = directive.args.single().name
-            if (compatibleTarget != CompilationTarget.instance.name)
-                errors.err("module's compilation target ($compatibleTarget) differs from active target (${CompilationTarget.instance.name})", directive.position)
+            if (compatibleTarget != compTarget.name)
+                errors.err("module's compilation target ($compatibleTarget) differs from active target (${compTarget.name})", directive.position)
         }
 
         super.visit(directive)
@@ -50,7 +63,7 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
         if(decl.name in BuiltinFunctions)
             errors.err("builtin function cannot be redefined", decl.position)
 
-        if(decl.name in CompilationTarget.instance.machine.opcodeNames)
+        if(decl.name in compTarget.machine.opcodeNames)
             errors.err("can't use a cpu opcode name as a symbol: '${decl.name}'", decl.position)
 
         if(decl.datatype==DataType.STRUCT) {
@@ -89,7 +102,7 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
     }
 
     override fun visit(subroutine: Subroutine) {
-        if(subroutine.name in CompilationTarget.instance.machine.opcodeNames) {
+        if(subroutine.name in compTarget.machine.opcodeNames) {
             errors.err("can't use a cpu opcode name as a symbol: '${subroutine.name}'", subroutine.position)
         } else if(subroutine.name in BuiltinFunctions) {
             // the builtin functions can't be redefined
@@ -112,9 +125,12 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
                 val labelOrVar = subroutine.getLabelOrVariable(name)
                 if(labelOrVar!=null && labelOrVar.position != subroutine.position)
                     nameError(name, labelOrVar.position, subroutine)
-                val sub = subroutine.statements.singleOrNull { it is Subroutine && it.name==name}
+                val sub = subroutine.statements.firstOrNull { it is Subroutine && it.name==name}
                 if(sub!=null)
-                    nameError(name, sub.position, subroutine)
+                    nameError(name, subroutine.position, sub)
+                val block = program.allBlocks().firstOrNull { it.name==name }
+                if(block!=null)
+                    nameError(name, subroutine.position, block)
             }
 
             if(subroutine.isAsmSubroutine && subroutine.statements.any{it !is InlineAssembly}) {
@@ -126,7 +142,7 @@ internal class AstIdentifiersChecker(private val program: Program, private val e
     }
 
     override fun visit(label: Label) {
-        if(label.name in CompilationTarget.instance.machine.opcodeNames)
+        if(label.name in compTarget.machine.opcodeNames)
             errors.err("can't use a cpu opcode name as a symbol: '${label.name}'", label.position)
 
         if(label.name in BuiltinFunctions) {

compiler/src/prog8/compiler/astprocessing/AstVariousTransforms.kt

@@ -1,10 +1,16 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.Node
 import prog8.ast.Program
-import prog8.ast.base.*
-import prog8.ast.expressions.*
-import prog8.ast.statements.*
+import prog8.ast.base.DataType
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.StringLiteralValue
+import prog8.ast.statements.AnonymousScope
+import prog8.ast.statements.ParameterVarDecl
+import prog8.ast.statements.Subroutine
+import prog8.ast.statements.VarDecl
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
 
 
 internal class AstVariousTransforms(private val program: Program) : AstWalker() {
@@ -26,7 +32,7 @@ internal class AstVariousTransforms(private val program: Program) : AstWalker()
     }
 
     override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
-        // For non-kernel subroutines and non-asm parameters:
+        // For non-kernal subroutines and non-asm parameters:
         // inject subroutine params as local variables (if they're not there yet).
         val symbolsInSub = subroutine.allDefinedSymbols()
         val namesInSub = symbolsInSub.map{ it.first }.toSet()

compiler/src/prog8/compiler/astprocessing/LiteralsToAutoVars.kt

@@ -1,10 +1,15 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.Node
 import prog8.ast.Program
-import prog8.ast.base.*
-import prog8.ast.expressions.*
-import prog8.ast.statements.*
+import prog8.ast.base.DataType
+import prog8.ast.expressions.ArrayLiteralValue
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.StringLiteralValue
+import prog8.ast.statements.VarDecl
+import prog8.ast.statements.WhenChoice
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
 
 
 internal class LiteralsToAutoVars(private val program: Program) : AstWalker() {
@@ -12,13 +17,10 @@ internal class LiteralsToAutoVars(private val program: Program) : AstWalker() {
 
     override fun after(string: StringLiteralValue, parent: Node): Iterable<IAstModification> {
         if(string.parent !is VarDecl && string.parent !is WhenChoice) {
-            // replace the literal string by a identifier reference to a new local vardecl
-            val vardecl = VarDecl.createAuto(string)
-            val identifier = IdentifierReference(listOf(vardecl.name), vardecl.position)
-            return listOf(
-                    IAstModification.ReplaceNode(string, identifier, parent),
-                    IAstModification.InsertFirst(vardecl, string.definingScope())
-            )
+            // replace the literal string by a identifier reference to the interned string
+            val scopedName = program.internString(string)
+            val identifier = IdentifierReference(scopedName, string.position)
+            return listOf(IAstModification.ReplaceNode(string, identifier, parent))
         }
         return noModifications
     }

compiler/src/prog8/compiler/astprocessing/ReflectionAstWalker.kt

@@ -1,4 +1,4 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 
 /*

compiler/src/prog8/compiler/astprocessing/StatementReorderer.kt

@@ -0,0 +1,399 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Module
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
+
+
+internal class StatementReorderer(val program: Program, val errors: IErrorReporter) : AstWalker() {
+    // Reorders the statements in a way the compiler needs.
+    // - 'main' block must be the very first statement UNLESS it has an address set.
+    // - library blocks are put last.
+    // - blocks are ordered by address, where blocks without address are placed last.
+    // - in every block and module, most directives and vardecls are moved to the top. (not in subroutines!)
+    // - the 'start' subroutine is moved to the top.
+    // - (syntax desugaring) a vardecl with a non-const initializer value is split into a regular vardecl and an assignment statement.
+    // - (syntax desugaring) struct value assignment is expanded into several struct member assignments.
+    // - in-place assignments are reordered a bit so that they are mostly of the form A = A <operator> <rest>
+    // - sorts the choices in when statement.
+    // - insert AddressOf (&) expression where required (string params to a UWORD function param etc).
+
+    private val noModifications = emptyList<IAstModification>()
+    private val directivesToMove = setOf("%output", "%launcher", "%zeropage", "%zpreserved", "%address", "%option")
+
+    override fun after(module: Module, parent: Node): Iterable<IAstModification> {
+        val (blocks, other) = module.statements.partition { it is Block }
+        module.statements = other.asSequence().plus(blocks.sortedBy { (it as Block).address ?: Int.MAX_VALUE }).toMutableList()
+
+        val mainBlock = module.statements.filterIsInstance<Block>().firstOrNull { it.name=="main" }
+        if(mainBlock!=null && mainBlock.address==null) {
+            module.statements.remove(mainBlock)
+            module.statements.add(0, mainBlock)
+        }
+
+        reorderVardeclsAndDirectives(module.statements)
+        return noModifications
+    }
+
+    private fun reorderVardeclsAndDirectives(statements: MutableList<Statement>) {
+        val varDecls = statements.filterIsInstance<VarDecl>()
+        statements.removeAll(varDecls)
+        statements.addAll(0, varDecls)
+
+        val directives = statements.filterIsInstance<Directive>().filter {it.directive in directivesToMove}
+        statements.removeAll(directives)
+        statements.addAll(0, directives)
+    }
+
+    override fun before(block: Block, parent: Node): Iterable<IAstModification> {
+        parent as Module
+        if(block.isInLibrary) {
+            return listOf(
+                    IAstModification.Remove(block, parent),
+                    IAstModification.InsertLast(block, parent)
+            )
+        }
+
+        reorderVardeclsAndDirectives(block.statements)
+        return noModifications
+    }
+
+    override fun before(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
+        if(subroutine.name=="start" && parent is Block) {
+            if(parent.statements.filterIsInstance<Subroutine>().first().name!="start") {
+                return listOf(
+                        IAstModification.Remove(subroutine, parent),
+                        IAstModification.InsertFirst(subroutine, parent)
+                )
+            }
+        }
+
+        val subs = subroutine.statements.filterIsInstance<Subroutine>()
+        if(subs.isNotEmpty()) {
+            // all subroutines defined within this subroutine are moved to the end
+            return subs.map { IAstModification.Remove(it, subroutine) } +
+                    subs.map { IAstModification.InsertLast(it, subroutine) }
+        }
+
+        return noModifications
+    }
+
+    override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
+
+        val arrayVar = arrayIndexedExpression.arrayvar.targetVarDecl(program)
+        if(arrayVar!=null && arrayVar.datatype ==  DataType.UWORD) {
+            // rewrite   pointervar[index]  into  @(pointervar+index)
+            val indexer = arrayIndexedExpression.indexer
+            val add = BinaryExpression(arrayIndexedExpression.arrayvar, "+", indexer.indexExpr, arrayIndexedExpression.position)
+            return if(parent is AssignTarget) {
+                // we're part of the target of an assignment, we have to actually change the assign target itself
+                val memwrite = DirectMemoryWrite(add, arrayIndexedExpression.position)
+                val newtarget = AssignTarget(null, null, memwrite, arrayIndexedExpression.position)
+                listOf(IAstModification.ReplaceNode(parent, newtarget, parent.parent))
+            } else {
+                val memread = DirectMemoryRead(add, arrayIndexedExpression.position)
+                listOf(IAstModification.ReplaceNode(arrayIndexedExpression, memread, parent))
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        // when using a simple bit shift and assigning it to a variable of a different type,
+        // try to make the bit shifting 'wide enough' to fall into the variable's type.
+        // with this, for instance, uword x = 1 << 10  will result in 1024 rather than 0 (the ubyte result).
+        if(expr.operator=="<<" || expr.operator==">>") {
+            val leftDt = expr.left.inferType(program)
+            when (parent) {
+                is Assignment -> {
+                    val targetDt = parent.target.inferType(program)
+                    if(leftDt != targetDt) {
+                        val cast = TypecastExpression(expr.left, targetDt.typeOrElse(DataType.STRUCT), true, parent.position)
+                        return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                    }
+                }
+                is VarDecl -> {
+                    if(!leftDt.istype(parent.datatype)) {
+                        val cast = TypecastExpression(expr.left, parent.datatype, true, parent.position)
+                        return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                    }
+                }
+                is IFunctionCall -> {
+                    val argnum = parent.args.indexOf(expr)
+                    when (val callee = parent.target.targetStatement(program)) {
+                        is Subroutine -> {
+                            val paramType = callee.parameters[argnum].type
+                            if(leftDt isAssignableTo paramType) {
+                                val cast = TypecastExpression(expr.left, paramType, true, parent.position)
+                                return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                            }
+                        }
+                        is BuiltinFunctionStatementPlaceholder -> {
+                            val func = BuiltinFunctions.getValue(callee.name)
+                            val paramTypes = func.parameters[argnum].possibleDatatypes
+                            for(type in paramTypes) {
+                                if(leftDt isAssignableTo type) {
+                                    val cast = TypecastExpression(expr.left, type, true, parent.position)
+                                    return listOf(IAstModification.ReplaceNode(expr.left, cast, expr))
+                                }
+                            }
+                        }
+                        else -> throw FatalAstException("weird callee")
+                    }
+                }
+                else -> return noModifications
+            }
+        }
+        else if(expr.operator in logicalOperators) {
+            // make sure that logical expressions like "var and other-logical-expression
+            // is rewritten as "var!=0 and other-logical-expression", to avoid bitwise boolean and
+            // generating the wrong results later
+
+            fun wrapped(expr: Expression): Expression =
+                BinaryExpression(expr, "!=", NumericLiteralValue(DataType.UBYTE, 0, expr.position), expr.position)
+
+            fun isLogicalExpr(expr: Expression?): Boolean {
+                if(expr is BinaryExpression && expr.operator in (logicalOperators + comparisonOperators))
+                    return true
+                if(expr is PrefixExpression && expr.operator in logicalOperators)
+                    return true
+                return false
+            }
+
+            return if(isLogicalExpr(expr.left)) {
+                if(isLogicalExpr(expr.right))
+                    noModifications
+                else
+                    listOf(IAstModification.ReplaceNode(expr.right, wrapped(expr.right), expr))
+            } else {
+                if(isLogicalExpr(expr.right))
+                    listOf(IAstModification.ReplaceNode(expr.left, wrapped(expr.left), expr))
+                else {
+                    listOf(
+                        IAstModification.ReplaceNode(expr.left, wrapped(expr.left), expr),
+                        IAstModification.ReplaceNode(expr.right, wrapped(expr.right), expr)
+                    )
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun after(whenStatement: WhenStatement, parent: Node): Iterable<IAstModification> {
+        val choices = whenStatement.choiceValues(program).sortedBy {
+            it.first?.first() ?: Int.MAX_VALUE
+        }
+        whenStatement.choices.clear()
+        choices.mapTo(whenStatement.choices) { it.second }
+        return noModifications
+    }
+
+    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
+        val declValue = decl.value
+        if(declValue!=null && decl.type== VarDeclType.VAR && decl.datatype in NumericDatatypes) {
+            val declConstValue = declValue.constValue(program)
+            if(declConstValue==null) {
+                // move the vardecl (without value) to the scope and replace this with a regular assignment
+                // Unless we're dealing with a floating point variable because that will actually make things less efficient at the moment (because floats are mostly calcualated via the stack)
+                if(decl.datatype!=DataType.FLOAT) {
+                    decl.value = null
+                    decl.allowInitializeWithZero = false
+                    val target = AssignTarget(IdentifierReference(listOf(decl.name), decl.position), null, null, decl.position)
+                    val assign = Assignment(target, declValue, decl.position)
+                    return listOf(
+                            IAstModification.ReplaceNode(decl, assign, parent),
+                            IAstModification.InsertFirst(decl, decl.definingScope())
+                    )
+                }
+            }
+        }
+        return noModifications
+    }
+
+    override fun before(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        val valueType = assignment.value.inferType(program)
+        val targetType = assignment.target.inferType(program)
+
+        if(targetType.istype(DataType.STRUCT) && (valueType.istype(DataType.STRUCT) || valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes )) {
+            if (assignment.value is ArrayLiteralValue) {
+                errors.err("cannot assign non-const array value, use separate assignment per field", assignment.position)
+            } else {
+                return copyStructValue(assignment)
+            }
+        }
+
+        if(targetType.typeOrElse(DataType.STRUCT) in ArrayDatatypes && valueType.typeOrElse(DataType.STRUCT) in ArrayDatatypes ) {
+            if (assignment.value is ArrayLiteralValue) {
+                errors.err("cannot assign non-const array value, use separate assignment per element", assignment.position)
+            } else {
+                return copyArrayValue(assignment)
+            }
+        }
+
+        return noModifications
+    }
+
+    override fun after(assignment: Assignment, parent: Node): Iterable<IAstModification> {
+        // rewrite in-place assignment expressions a bit so that the assignment target usually is the leftmost operand
+        val binExpr = assignment.value as? BinaryExpression
+        if(binExpr!=null) {
+            if(binExpr.left isSameAs assignment.target) {
+                // A = A <operator> 5, unchanged
+                return noModifications
+            }
+
+            if(binExpr.operator in associativeOperators) {
+                if (binExpr.right isSameAs assignment.target) {
+                    // A = v <associative-operator> A  ==>  A = A <associative-operator> v
+                    return listOf(IAstModification.SwapOperands(binExpr))
+                }
+
+                val leftBinExpr = binExpr.left as? BinaryExpression
+                if(leftBinExpr?.operator == binExpr.operator) {
+                    return if(leftBinExpr.left isSameAs assignment.target) {
+                        // A = (A <associative-operator> x) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(leftBinExpr.right, binExpr.operator, binExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(leftBinExpr.left, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    } else {
+                        // A = (x <associative-operator> A) <same-operator> y ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(leftBinExpr.left, binExpr.operator, binExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(leftBinExpr.right, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    }
+                }
+                val rightBinExpr = binExpr.right as? BinaryExpression
+                if(rightBinExpr?.operator == binExpr.operator) {
+                    return if(rightBinExpr.left isSameAs assignment.target) {
+                        // A = x <associative-operator> (A <same-operator> y) ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.right, binExpr.position)
+                        val newValue = BinaryExpression(rightBinExpr.left, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    } else {
+                        // A = x <associative-operator> (y <same-operator> A) ==> A = A <associative-operator> (x <same-operator> y)
+                        val newRight = BinaryExpression(binExpr.left, binExpr.operator, rightBinExpr.left, binExpr.position)
+                        val newValue = BinaryExpression(rightBinExpr.right, binExpr.operator, newRight, binExpr.position)
+                        listOf(IAstModification.ReplaceNode(binExpr, newValue, assignment))
+                    }
+                }
+            }
+        }
+
+        return noModifications
+    }
+
+    private fun copyArrayValue(assign: Assignment): List<IAstModification> {
+        val identifier = assign.target.identifier!!
+        val targetVar = identifier.targetVarDecl(program)!!
+
+        if(targetVar.arraysize==null)
+            errors.err("array has no defined size", assign.position)
+
+        if(assign.value !is IdentifierReference) {
+            errors.err("invalid array value to assign to other array", assign.value.position)
+            return noModifications
+        }
+        val sourceIdent = assign.value as IdentifierReference
+        val sourceVar = sourceIdent.targetVarDecl(program)!!
+        if(!sourceVar.isArray) {
+            errors.err("value must be an array", sourceIdent.position)
+        } else {
+            if (sourceVar.arraysize!!.constIndex() != targetVar.arraysize!!.constIndex())
+                errors.err("element count mismatch", assign.position)
+            if (sourceVar.datatype != targetVar.datatype)
+                errors.err("element type mismatch", assign.position)
+        }
+
+        if(!errors.isEmpty())
+            return noModifications
+
+        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), assign.position),
+            mutableListOf(
+                AddressOf(sourceIdent, assign.position),
+                AddressOf(identifier, assign.position),
+                NumericLiteralValue.optimalInteger(targetVar.arraysize!!.constIndex()!!, assign.position)
+            ),
+            true,
+            assign.position
+        )
+        return listOf(IAstModification.ReplaceNode(assign, memcopy, assign.parent))
+    }
+
+    private fun copyStructValue(structAssignment: Assignment): List<IAstModification> {
+        val identifier = structAssignment.target.identifier!!
+        val targetVar = identifier.targetVarDecl(program)!!
+        val struct = targetVar.struct!!
+        when (structAssignment.value) {
+            is IdentifierReference -> {
+                val sourceVar = (structAssignment.value as IdentifierReference).targetVarDecl(program)!!
+                val memsize = struct.memsize(program.memsizer)
+                when {
+                    sourceVar.struct!=null -> {
+                        // struct memberwise copy
+                        val sourceStruct = sourceVar.struct!!
+                        if(sourceStruct!==targetVar.struct) {
+                            errors.err("struct type mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(struct.statements.size!=sourceStruct.statements.size) {
+                            errors.err("struct element count mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(memsize!=sourceStruct.memsize(program.memsizer)) {
+                            errors.err("memory size mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), structAssignment.position),
+                            mutableListOf(
+                                AddressOf(structAssignment.value as IdentifierReference, structAssignment.position),
+                                AddressOf(identifier, structAssignment.position),
+                                NumericLiteralValue.optimalInteger(memsize, structAssignment.position)
+                            ),
+                            true,
+                            structAssignment.position
+                        )
+                        return listOf(IAstModification.ReplaceNode(structAssignment, memcopy, structAssignment.parent))
+                    }
+                    sourceVar.isArray -> {
+                        val array = sourceVar.value as ArrayLiteralValue
+                        if(struct.statements.size!=array.value.size) {
+                            errors.err("struct element count mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        if(memsize!=array.memsize(program.memsizer)) {
+                            errors.err("memory size mismatch", structAssignment.position)
+                            return listOf()
+                        }
+                        val memcopy = FunctionCallStatement(IdentifierReference(listOf("sys", "memcopy"), structAssignment.position),
+                            mutableListOf(
+                                AddressOf(structAssignment.value as IdentifierReference, structAssignment.position),
+                                AddressOf(identifier, structAssignment.position),
+                                NumericLiteralValue.optimalInteger(memsize, structAssignment.position)
+                            ),
+                            true,
+                            structAssignment.position
+                        )
+                        return listOf(IAstModification.ReplaceNode(structAssignment, memcopy, structAssignment.parent))
+                    }
+                    else -> {
+                        throw FatalAstException("can only assign arrays or structs to structs")
+                    }
+                }
+            }
+            is ArrayLiteralValue -> {
+                throw IllegalArgumentException("not going to do a structLv assignment here")
+            }
+            else -> throw FatalAstException("strange struct value")
+        }
+    }
+
+}

compiler/src/prog8/compiler/astprocessing/TypecastsAdder.kt

@@ -1,16 +1,18 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.IFunctionCall
-import prog8.ast.INameScope
 import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
 import prog8.ast.statements.*
-import prog8.functions.BuiltinFunctions
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.functions.BuiltinFunctions
 
 
-class TypecastsAdder(val program: Program, val errors: ErrorReporter) : AstWalker() {
+class TypecastsAdder(val program: Program, val errors: IErrorReporter) : AstWalker() {
     /*
      * Make sure any value assignments get the proper type casts if needed to cast them into the target variable's type.
      * (this includes function call arguments)
@@ -106,18 +108,18 @@ class TypecastsAdder(val program: Program, val errors: ErrorReporter) : AstWalke
     }
 
     override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
-        return afterFunctionCallArgs(functionCallStatement, functionCallStatement.definingScope())
+        return afterFunctionCallArgs(functionCallStatement)
     }
 
     override fun after(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
-        return afterFunctionCallArgs(functionCall, functionCall.definingScope())
+        return afterFunctionCallArgs(functionCall)
     }
 
-    private fun afterFunctionCallArgs(call: IFunctionCall, scope: INameScope): Iterable<IAstModification> {
+    private fun afterFunctionCallArgs(call: IFunctionCall): Iterable<IAstModification> {
         // see if a typecast is needed to convert the arguments into the required parameter's type
         val modifications = mutableListOf<IAstModification>()
 
-        when(val sub = call.target.targetStatement(scope)) {
+        when(val sub = call.target.targetStatement(program)) {
             is Subroutine -> {
                 sub.parameters.zip(call.args).forEachIndexed { index, pair ->
                     val argItype = pair.second.inferType(program)

compiler/src/prog8/compiler/astprocessing/VariousCleanups.kt

@@ -0,0 +1,73 @@
+package prog8.compiler.astprocessing
+
+import prog8.ast.IFunctionCall
+import prog8.ast.INameScope
+import prog8.ast.Node
+import prog8.ast.base.Position
+import prog8.ast.expressions.DirectMemoryRead
+import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.expressions.TypecastExpression
+import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+
+
+internal class VariousCleanups: AstWalker() {
+    private val noModifications = emptyList<IAstModification>()
+
+    override fun before(nopStatement: NopStatement, parent: Node): Iterable<IAstModification> {
+        return listOf(IAstModification.Remove(nopStatement, parent as INameScope))
+    }
+
+    override fun before(scope: AnonymousScope, parent: Node): Iterable<IAstModification> {
+        return if(parent is INameScope)
+            listOf(ScopeFlatten(scope, parent as INameScope))
+        else
+            noModifications
+    }
+
+    class ScopeFlatten(val scope: AnonymousScope, val into: INameScope) : IAstModification {
+        override fun perform() {
+            val idx = into.statements.indexOf(scope)
+            if(idx>=0) {
+                into.statements.addAll(idx+1, scope.statements)
+                into.statements.remove(scope)
+            }
+        }
+    }
+
+    override fun before(typecast: TypecastExpression, parent: Node): Iterable<IAstModification> {
+        if(typecast.expression is NumericLiteralValue) {
+            val value = (typecast.expression as NumericLiteralValue).cast(typecast.type)
+            if(value.isValid)
+                return listOf(IAstModification.ReplaceNode(typecast, value.valueOrZero(), parent))
+        }
+
+        return noModifications
+    }
+
+    override fun before(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
+        return before(functionCallStatement as IFunctionCall, parent, functionCallStatement.position)
+    }
+
+    override fun before(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
+        return before(functionCall as IFunctionCall, parent, functionCall.position)
+    }
+
+    private fun before(functionCall: IFunctionCall, parent: Node, position: Position): Iterable<IAstModification> {
+        if(functionCall.target.nameInSource==listOf("peek")) {
+            // peek(a) is synonymous with @(a)
+            val memread = DirectMemoryRead(functionCall.args.single(), position)
+            return listOf(IAstModification.ReplaceNode(functionCall as Node, memread, parent))
+        }
+        if(functionCall.target.nameInSource==listOf("poke")) {
+            // poke(a, v) is synonymous with @(a) = v
+            val tgt = AssignTarget(null, null, DirectMemoryWrite(functionCall.args[0], position), position)
+            val assign = Assignment(tgt, functionCall.args[1], position)
+            return listOf(IAstModification.ReplaceNode(functionCall as Node, assign, parent))
+        }
+        return noModifications
+    }
+
+}

compiler/src/prog8/compiler/astprocessing/VerifyFunctionArgTypes.kt

@@ -1,26 +1,26 @@
-package prog8.ast.processing
+package prog8.compiler.astprocessing
 
 import prog8.ast.IFunctionCall
-import prog8.ast.INameScope
 import prog8.ast.Program
 import prog8.ast.base.DataType
-import prog8.ast.base.FatalAstException
 import prog8.ast.expressions.Expression
 import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.TypecastExpression
 import prog8.ast.statements.*
+import prog8.ast.walk.IAstVisitor
 import prog8.compiler.CompilerException
-import prog8.functions.BuiltinFunctions
+import prog8.compiler.functions.BuiltinFunctions
 
 class VerifyFunctionArgTypes(val program: Program) : IAstVisitor {
 
     override fun visit(functionCall: FunctionCall) {
-        val error = checkTypes(functionCall as IFunctionCall, functionCall.definingScope(), program)
+        val error = checkTypes(functionCall as IFunctionCall, program)
         if(error!=null)
             throw CompilerException(error)
     }
 
     override fun visit(functionCallStatement: FunctionCallStatement) {
-        val error = checkTypes(functionCallStatement as IFunctionCall, functionCallStatement.definingScope(), program)
+        val error = checkTypes(functionCallStatement as IFunctionCall, program)
         if (error!=null)
             throw CompilerException(error)
     }
@@ -39,12 +39,13 @@ class VerifyFunctionArgTypes(val program: Program) : IAstVisitor {
             return false
         }
 
-        fun checkTypes(call: IFunctionCall, scope: INameScope, program: Program): String? {
+        fun checkTypes(call: IFunctionCall, program: Program): String? {
             val argITypes = call.args.map { it.inferType(program) }
-            if(argITypes.any { !it.isKnown })
-                throw FatalAstException("unknown dt")
+            val firstUnknownDt = argITypes.indexOfFirst { it.isUnknown }
+            if(firstUnknownDt>=0)
+                return "argument ${firstUnknownDt+1} invalid argument type"
             val argtypes = argITypes.map { it.typeOrElse(DataType.STRUCT) }
-            val target = call.target.targetStatement(scope)
+            val target = call.target.targetStatement(program)
             if (target is Subroutine) {
                 if(call.args.size != target.parameters.size)
                     return "invalid number of arguments"
@@ -60,8 +61,15 @@ class VerifyFunctionArgTypes(val program: Program) : IAstVisitor {
                         // multiple return values will NOT work inside an expression.
                         // they MIGHT work in a regular assignment or just a function call statement.
                         val parent = if(call is Statement) call.parent else if(call is Expression) call.parent else null
-                        if(call !is FunctionCallStatement && parent !is Assignment && parent !is VarDecl) {
-                            return "can't use subroutine call that returns multiple return values here (try moving it into a separate assignment)"
+                        if (call !is FunctionCallStatement) {
+                            val checkParent =
+                                if(parent is TypecastExpression)
+                                    parent.parent
+                                else
+                                    parent
+                            if (checkParent !is Assignment && checkParent !is VarDecl) {
+                                return "can't use subroutine call that returns multiple return values here"
+                            }
                         }
                     }
                 }

compiler/src/prog8/compiler/functions/BuiltinFunctions.kt

@@ -1,5 +1,6 @@
-package prog8.functions
+package prog8.compiler.functions
 
+import prog8.ast.IMemSizer
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
@@ -12,7 +13,7 @@ import kotlin.math.*
 class FParam(val name: String, val possibleDatatypes: Set<DataType>)
 
 
-typealias ConstExpressionCaller = (args: List<Expression>, position: Position, program: Program) -> NumericLiteralValue
+typealias ConstExpressionCaller = (args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer) -> NumericLiteralValue
 
 
 class ReturnConvention(val dt: DataType, val reg: RegisterOrPair?, val floatFac1: Boolean)
@@ -87,6 +88,7 @@ class FSignature(val name: String,
     }
 }
 
+@Suppress("UNUSED_ANONYMOUS_PARAMETER")
 private val functionSignatures: List<FSignature> = listOf(
         // this set of function have no return value and operate in-place:
     FSignature("rol"         , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
@@ -95,80 +97,54 @@ private val functionSignatures: List<FSignature> = listOf(
     FSignature("ror2"        , false, listOf(FParam("item", setOf(DataType.UBYTE, DataType.UWORD))), null),
     FSignature("sort"        , false, listOf(FParam("array", ArrayDatatypes)), null),
     FSignature("reverse"     , false, listOf(FParam("array", ArrayDatatypes)), null),
+    FSignature("cmp"         , false, listOf(FParam("value1", IntegerDatatypes), FParam("value2", NumericDatatypes)), null),
         // these few have a return value depending on the argument(s):
-    FSignature("max"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg -> collectionArg(a, p, prg, ::builtinMax) },    // type depends on args
-    FSignature("min"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg -> collectionArg(a, p, prg, ::builtinMin) },    // type depends on args
-    FSignature("sum"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg -> collectionArg(a, p, prg, ::builtinSum) },      // type depends on args
+    FSignature("max"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinMax) },    // type depends on args
+    FSignature("min"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinMin) },    // type depends on args
+    FSignature("sum"         , true, listOf(FParam("values", ArrayDatatypes)), null) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinSum) },      // type depends on args
     FSignature("abs"         , true, listOf(FParam("value", NumericDatatypes)), null, ::builtinAbs),      // type depends on argument
     FSignature("len"         , true, listOf(FParam("values", IterableDatatypes)), null, ::builtinLen),    // type is UBYTE or UWORD depending on actual length
     FSignature("sizeof"      , true, listOf(FParam("object", DataType.values().toSet())), DataType.UBYTE, ::builtinSizeof),
+    FSignature("offsetof"    , true, listOf(FParam("object", DataType.values().toSet())), DataType.UBYTE, ::builtinOffsetof),
         // normal functions follow:
     FSignature("sgn"         , true, listOf(FParam("value", NumericDatatypes)), DataType.BYTE, ::builtinSgn ),
-    FSignature("sin"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::sin) },
+    FSignature("sin"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::sin) },
     FSignature("sin8"        , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.BYTE, ::builtinSin8 ),
     FSignature("sin8u"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UBYTE, ::builtinSin8u ),
     FSignature("sin16"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.WORD, ::builtinSin16 ),
     FSignature("sin16u"      , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinSin16u ),
-    FSignature("cos"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::cos) },
+    FSignature("cos"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::cos) },
     FSignature("cos8"        , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.BYTE, ::builtinCos8 ),
     FSignature("cos8u"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UBYTE, ::builtinCos8u ),
     FSignature("cos16"       , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.WORD, ::builtinCos16 ),
     FSignature("cos16u"      , true, listOf(FParam("angle8", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinCos16u ),
-    FSignature("tan"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::tan) },
-    FSignature("atan"        , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::atan) },
-    FSignature("ln"          , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::log) },
-    FSignature("log2"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, ::log2) },
-    FSignature("sqrt16"      , true, listOf(FParam("value", setOf(DataType.UWORD))), DataType.UBYTE) { a, p, prg -> oneIntArgOutputInt(a, p, prg) { sqrt(it.toDouble()).toInt() } },
-    FSignature("sqrt"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::sqrt) },
-    FSignature("rad"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::toRadians) },
-    FSignature("deg"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArg(a, p, prg, Math::toDegrees) },
-    FSignature("round"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArgOutputWord(a, p, prg, Math::round) },
-    FSignature("floor"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArgOutputWord(a, p, prg, Math::floor) },
-    FSignature("ceil"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg -> oneDoubleArgOutputWord(a, p, prg, Math::ceil) },
-    FSignature("any"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg -> collectionArg(a, p, prg, ::builtinAny) },
-    FSignature("all"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg -> collectionArg(a, p, prg, ::builtinAll) },
-    FSignature("lsb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg -> oneIntArgOutputInt(a, p, prg) { x: Int -> x and 255 }},
-    FSignature("msb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg -> oneIntArgOutputInt(a, p, prg) { x: Int -> x ushr 8 and 255}},
+    FSignature("tan"         , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::tan) },
+    FSignature("atan"        , true, listOf(FParam("rads", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::atan) },
+    FSignature("ln"          , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::log) },
+    FSignature("log2"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, ::log2) },
+    FSignature("sqrt16"      , true, listOf(FParam("value", setOf(DataType.UWORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { sqrt(it.toDouble()).toInt() } },
+    FSignature("sqrt"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::sqrt) },
+    FSignature("rad"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::toRadians) },
+    FSignature("deg"         , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArg(a, p, prg, Math::toDegrees) },
+    FSignature("round"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::round) },
+    FSignature("floor"       , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::floor) },
+    FSignature("ceil"        , true, listOf(FParam("value", setOf(DataType.FLOAT))), DataType.FLOAT) { a, p, prg, ct -> oneDoubleArgOutputWord(a, p, prg, Math::ceil) },
+    FSignature("any"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinAny) },
+    FSignature("all"         , true, listOf(FParam("values", ArrayDatatypes)), DataType.UBYTE) { a, p, prg, ct -> collectionArg(a, p, prg, ::builtinAll) },
+    FSignature("lsb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { x: Int -> x and 255 } },
+    FSignature("msb"         , true, listOf(FParam("value", setOf(DataType.UWORD, DataType.WORD))), DataType.UBYTE) { a, p, prg, ct -> oneIntArgOutputInt(a, p, prg) { x: Int -> x ushr 8 and 255} },
     FSignature("mkword"      , true, listOf(FParam("msb", setOf(DataType.UBYTE)), FParam("lsb", setOf(DataType.UBYTE))), DataType.UWORD, ::builtinMkword),
+    FSignature("peek"        , true, listOf(FParam("address", setOf(DataType.UWORD))), DataType.UBYTE),
+    FSignature("peekw"       , true, listOf(FParam("address", setOf(DataType.UWORD))), DataType.UWORD),
+    FSignature("poke"        , false, listOf(FParam("address", setOf(DataType.UWORD)), FParam("value", setOf(DataType.UBYTE))), null),
+    FSignature("pokew"       , false, listOf(FParam("address", setOf(DataType.UWORD)), FParam("value", setOf(DataType.UWORD))), null),
+    FSignature("fastrnd8"    , false, emptyList(), DataType.UBYTE),
     FSignature("rnd"         , false, emptyList(), DataType.UBYTE),
     FSignature("rndw"        , false, emptyList(), DataType.UWORD),
     FSignature("rndf"        , false, emptyList(), DataType.FLOAT),
-    FSignature("exit"        , false, listOf(FParam("returnvalue", setOf(DataType.UBYTE))), null),
-    FSignature("rsave"       , false, emptyList(), null),
-    FSignature("rrestore"    , false, emptyList(), null),
-    FSignature("set_carry"   , false, emptyList(), null),
-    FSignature("clear_carry" , false, emptyList(), null),
-    FSignature("set_irqd"    , false, emptyList(), null),
-    FSignature("clear_irqd"  , false, emptyList(), null),
-    FSignature("read_flags"  , false, emptyList(), DataType.UBYTE),
+    FSignature("memory"      , true, listOf(FParam("name", setOf(DataType.STR)), FParam("size", setOf(DataType.UWORD))), DataType.UWORD),
     FSignature("swap"        , false, listOf(FParam("first", NumericDatatypes), FParam("second", NumericDatatypes)), null),
-    FSignature("memcopy"     , false, listOf(
-                            FParam("from", IterableDatatypes + DataType.UWORD),
-                            FParam("to", IterableDatatypes + DataType.UWORD),
-                            FParam("numbytes", setOf(DataType.UBYTE, DataType.UWORD))), null),
-    FSignature("memset"      , false, listOf(
-                            FParam("address", IterableDatatypes + DataType.UWORD),
-                            FParam("numbytes", setOf(DataType.UWORD)),
-                            FParam("bytevalue", ByteDatatypes)), null),
-    FSignature("memsetw"     , false, listOf(
-                            FParam("address", IterableDatatypes + DataType.UWORD),
-                            FParam("numwords", setOf(DataType.UWORD)),
-                            FParam("wordvalue", setOf(DataType.UWORD, DataType.WORD))), null),
-    FSignature("strlen"      , true, listOf(FParam("string", setOf(DataType.STR))), DataType.UBYTE, ::builtinStrlen),
-    FSignature("substr"      , false, listOf(
-            FParam("source", IterableDatatypes + DataType.UWORD),
-            FParam("target", IterableDatatypes + DataType.UWORD),
-            FParam("start", setOf(DataType.UBYTE)),
-            FParam("length", setOf(DataType.UBYTE))), null),
-    FSignature("leftstr"     , false, listOf(
-            FParam("source", IterableDatatypes + DataType.UWORD),
-            FParam("target", IterableDatatypes + DataType.UWORD),
-            FParam("length", setOf(DataType.UBYTE))), null),
-    FSignature("rightstr"    , false, listOf(
-            FParam("source", IterableDatatypes + DataType.UWORD),
-            FParam("target", IterableDatatypes + DataType.UWORD),
-            FParam("length", setOf(DataType.UBYTE))), null),
-    FSignature("strcmp"      , false, listOf(FParam("s1", IterableDatatypes + DataType.UWORD), FParam("s2", IterableDatatypes + DataType.UWORD)), DataType.BYTE, null)
+
 )
 
 val BuiltinFunctions = functionSignatures.associateBy { it.name }
@@ -298,7 +274,8 @@ private fun collectionArg(args: List<Expression>, position: Position, program: P
     return NumericLiteralValue.optimalNumeric(function(constElements.mapNotNull { it }), args[0].position)
 }
 
-private fun builtinAbs(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinAbs(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     // 1 arg, type = float or int, result type= isSameAs as argument type
     if(args.size!=1)
         throw SyntaxError("abs requires one numeric argument", position)
@@ -311,7 +288,29 @@ private fun builtinAbs(args: List<Expression>, position: Position, program: Prog
     }
 }
 
-private fun builtinSizeof(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+private fun builtinOffsetof(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
+    // 1 arg, type = anything, result type = ubyte
+    if(args.size!=1)
+        throw SyntaxError("offsetof requires one argument", position)
+    val idref = args[0] as? IdentifierReference
+        ?: throw SyntaxError("offsetof argument should be an identifier", position)
+
+    val vardecl = idref.targetVarDecl(program)!!
+    val struct = vardecl.struct
+    if (struct == null || vardecl.datatype == DataType.STRUCT)
+        throw SyntaxError("offsetof can only be used on struct members", position)
+
+    val membername = idref.nameInSource.last()
+    var offset = 0
+    for(member in struct.statements) {
+        if((member as VarDecl).name == membername)
+            return NumericLiteralValue(DataType.UBYTE, offset, position)
+        offset += memsizer.memorySize(member.datatype)
+    }
+    throw SyntaxError("undefined struct member", position)
+}
+
+private fun builtinSizeof(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     // 1 arg, type = anything, result type = ubyte
     if(args.size!=1)
         throw SyntaxError("sizeof requires one argument", position)
@@ -320,17 +319,16 @@ private fun builtinSizeof(args: List<Expression>, position: Position, program: P
 
     val dt = args[0].inferType(program)
     if(dt.isKnown) {
-        val target = (args[0] as IdentifierReference).targetStatement(program.namespace)
+        val target = (args[0] as IdentifierReference).targetStatement(program)
                 ?: throw CannotEvaluateException("sizeof", "no target")
 
-        fun structSize(target: StructDecl) =
-                NumericLiteralValue(DataType.UBYTE, target.statements.map { (it as VarDecl).datatype.memorySize() }.sum(), position)
+        fun structSize(target: StructDecl) = NumericLiteralValue(DataType.UBYTE, target.memsize(memsizer), position)
 
         return when {
             dt.typeOrElse(DataType.STRUCT) in ArrayDatatypes -> {
                 val length = (target as VarDecl).arraysize!!.constIndex() ?: throw CannotEvaluateException("sizeof", "unknown array size")
                 val elementDt = ArrayElementTypes.getValue(dt.typeOrElse(DataType.STRUCT))
-                numericLiteral(elementDt.memorySize() * length, position)
+                numericLiteral(memsizer.memorySize(elementDt) * length, position)
             }
             dt.istype(DataType.STRUCT) -> {
                 when (target) {
@@ -340,36 +338,20 @@ private fun builtinSizeof(args: List<Expression>, position: Position, program: P
                 }
             }
             dt.istype(DataType.STR) -> throw SyntaxError("sizeof str is undefined, did you mean len?", position)
-            else -> NumericLiteralValue(DataType.UBYTE, dt.typeOrElse(DataType.STRUCT).memorySize(), position)
+            else -> NumericLiteralValue(DataType.UBYTE, memsizer.memorySize(dt.typeOrElse(DataType.STRUCT)), position)
         }
     } else {
         throw SyntaxError("sizeof invalid argument type", position)
     }
 }
 
-private fun builtinStrlen(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
-    if (args.size != 1)
-        throw SyntaxError("strlen requires one argument", position)
-    val argument=args[0]
-    if(argument is StringLiteralValue)
-        return NumericLiteralValue.optimalInteger(argument.value.length, argument.position)
-    val vardecl = (argument as IdentifierReference).targetVarDecl(program.namespace)
-    if(vardecl!=null) {
-        if(vardecl.datatype!=DataType.STR && vardecl.datatype!=DataType.UWORD)
-            throw SyntaxError("strlen must have string argument", position)
-        if(vardecl.autogeneratedDontRemove && vardecl.value!=null) {
-            return NumericLiteralValue.optimalInteger((vardecl.value as StringLiteralValue).value.length, argument.position)
-        }
-    }
-    throw NotConstArgumentException()
-}
-
-private fun builtinLen(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinLen(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     // note: in some cases the length is > 255 and then we have to return a UWORD type instead of a UBYTE.
     if(args.size!=1)
         throw SyntaxError("len requires one argument", position)
 
-    val directMemVar = ((args[0] as? DirectMemoryRead)?.addressExpression as? IdentifierReference)?.targetVarDecl(program.namespace)
+    val directMemVar = ((args[0] as? DirectMemoryRead)?.addressExpression as? IdentifierReference)?.targetVarDecl(program)
     var arraySize = directMemVar?.arraysize?.constIndex()
     if(arraySize != null)
         return NumericLiteralValue.optimalInteger(arraySize, position)
@@ -377,7 +359,7 @@ private fun builtinLen(args: List<Expression>, position: Position, program: Prog
         return NumericLiteralValue.optimalInteger((args[0] as ArrayLiteralValue).value.size, position)
     if(args[0] !is IdentifierReference)
         throw SyntaxError("len argument should be an identifier", position)
-    val target = (args[0] as IdentifierReference).targetVarDecl(program.namespace)
+    val target = (args[0] as IdentifierReference).targetVarDecl(program)
             ?: throw CannotEvaluateException("len", "no target vardecl")
 
     return when(target.datatype) {
@@ -388,7 +370,7 @@ private fun builtinLen(args: List<Expression>, position: Position, program: Prog
             NumericLiteralValue.optimalInteger(arraySize, args[0].position)
         }
         DataType.STR -> {
-            val refLv = target.value as StringLiteralValue
+            val refLv = target.value as? StringLiteralValue ?: throw CannotEvaluateException("len", "stringsize unknown")
             NumericLiteralValue.optimalInteger(refLv.value.length, args[0].position)
         }
         DataType.STRUCT -> throw SyntaxError("cannot use len on struct, did you mean sizeof?", args[0].position)
@@ -398,7 +380,8 @@ private fun builtinLen(args: List<Expression>, position: Position, program: Prog
 }
 
 
-private fun builtinMkword(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinMkword(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 2)
         throw SyntaxError("mkword requires msb and lsb arguments", position)
     val constMsb = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -407,7 +390,8 @@ private fun builtinMkword(args: List<Expression>, position: Position, program: P
     return NumericLiteralValue(DataType.UWORD, result, position)
 }
 
-private fun builtinSin8(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin8(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("sin8 requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -415,7 +399,8 @@ private fun builtinSin8(args: List<Expression>, position: Position, program: Pro
     return NumericLiteralValue(DataType.BYTE, (127.0 * sin(rad)).toInt().toShort(), position)
 }
 
-private fun builtinSin8u(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin8u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("sin8u requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -423,7 +408,8 @@ private fun builtinSin8u(args: List<Expression>, position: Position, program: Pr
     return NumericLiteralValue(DataType.UBYTE, (128.0 + 127.5 * sin(rad)).toInt().toShort(), position)
 }
 
-private fun builtinCos8(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos8(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("cos8 requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -431,7 +417,8 @@ private fun builtinCos8(args: List<Expression>, position: Position, program: Pro
     return NumericLiteralValue(DataType.BYTE, (127.0 * cos(rad)).toInt().toShort(), position)
 }
 
-private fun builtinCos8u(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos8u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("cos8u requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -439,7 +426,8 @@ private fun builtinCos8u(args: List<Expression>, position: Position, program: Pr
     return NumericLiteralValue(DataType.UBYTE, (128.0 + 127.5 * cos(rad)).toInt().toShort(), position)
 }
 
-private fun builtinSin16(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin16(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("sin16 requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -447,7 +435,8 @@ private fun builtinSin16(args: List<Expression>, position: Position, program: Pr
     return NumericLiteralValue(DataType.WORD, (32767.0 * sin(rad)).toInt(), position)
 }
 
-private fun builtinSin16u(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSin16u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("sin16u requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -455,7 +444,8 @@ private fun builtinSin16u(args: List<Expression>, position: Position, program: P
     return NumericLiteralValue(DataType.UWORD, (32768.0 + 32767.5 * sin(rad)).toInt(), position)
 }
 
-private fun builtinCos16(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos16(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("cos16 requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -463,7 +453,8 @@ private fun builtinCos16(args: List<Expression>, position: Position, program: Pr
     return NumericLiteralValue(DataType.WORD, (32767.0 * cos(rad)).toInt(), position)
 }
 
-private fun builtinCos16u(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinCos16u(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("cos16u requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()
@@ -471,7 +462,8 @@ private fun builtinCos16u(args: List<Expression>, position: Position, program: P
     return NumericLiteralValue(DataType.UWORD, (32768.0 + 32767.5 * cos(rad)).toInt(), position)
 }
 
-private fun builtinSgn(args: List<Expression>, position: Position, program: Program): NumericLiteralValue {
+@Suppress("UNUSED_PARAMETER")
+private fun builtinSgn(args: List<Expression>, position: Position, program: Program, memsizer: IMemSizer): NumericLiteralValue {
     if (args.size != 1)
         throw SyntaxError("sgn requires one argument", position)
     val constval = args[0].constValue(program) ?: throw NotConstArgumentException()

compiler/src/prog8/compiler/target/CompilationTarget.kt

@@ -1,47 +0,0 @@
-package prog8.compiler.target
-
-import prog8.ast.Program
-import prog8.ast.base.ErrorReporter
-import prog8.compiler.CompilationOptions
-import prog8.compiler.Zeropage
-import prog8.compiler.target.c64.C64MachineDefinition
-import prog8.compiler.target.c64.Petscii
-import prog8.compiler.target.c64.codegen.AsmGen
-import prog8.compiler.target.cx16.CX16MachineDefinition
-import java.nio.file.Path
-
-
-internal interface CompilationTarget {
-    val name: String
-    val machine: IMachineDefinition
-    fun encodeString(str: String, altEncoding: Boolean): List<Short>
-    fun decodeString(bytes: List<Short>, altEncoding: Boolean): String
-    fun asmGenerator(program: Program, errors: ErrorReporter, zp: Zeropage, options: CompilationOptions, path: Path): IAssemblyGenerator
-
-    companion object {
-        lateinit var instance: CompilationTarget
-    }
-}
-
-
-internal object C64Target: CompilationTarget {
-    override val name = "c64"
-    override val machine = C64MachineDefinition
-    override fun encodeString(str: String, altEncoding: Boolean) =
-            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
-    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
-            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
-    override fun asmGenerator(program: Program, errors: ErrorReporter, zp: Zeropage, options: CompilationOptions, path: Path) =
-            AsmGen(program, errors, zp, options, path)
-}
-
-internal object Cx16Target: CompilationTarget {
-    override val name = "cx16"
-    override val machine = CX16MachineDefinition
-    override fun encodeString(str: String, altEncoding: Boolean) =
-            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
-    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
-            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
-    override fun asmGenerator(program: Program, errors: ErrorReporter, zp: Zeropage, options: CompilationOptions, path: Path) =
-            AsmGen(program, errors, zp, options, path)
-}

compiler/src/prog8/compiler/target/IAssemblyGenerator.kt

@@ -3,10 +3,12 @@ package prog8.compiler.target
 import prog8.compiler.CompilationOptions
 
 internal interface IAssemblyGenerator {
-    fun compileToAssembly(optimize: Boolean): IAssemblyProgram
+    fun compileToAssembly(): IAssemblyProgram
 }
 
 internal const val generatedLabelPrefix = "_prog8_label_"
+internal const val subroutineFloatEvalResultVar1 = "_prog8_float_eval_result1"
+internal const val subroutineFloatEvalResultVar2 = "_prog8_float_eval_result2"
 
 internal interface IAssemblyProgram {
     val name: String

compiler/src/prog8/compiler/target/ICompilationTarget.kt

@@ -0,0 +1,119 @@
+package prog8.compiler.target
+
+import prog8.ast.IMemSizer
+import prog8.ast.IStringEncoding
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.IdentifierReference
+import prog8.ast.expressions.NumericLiteralValue
+import prog8.ast.statements.AssignTarget
+import prog8.compiler.CompilationOptions
+import prog8.compiler.IErrorReporter
+import prog8.compiler.Zeropage
+import prog8.compiler.target.c64.C64MachineDefinition
+import prog8.compiler.target.c64.Petscii
+import prog8.compiler.target.cpu6502.codegen.AsmGen
+import prog8.compiler.target.cx16.CX16MachineDefinition
+import java.nio.file.Path
+
+
+interface ICompilationTarget: IStringEncoding, IMemSizer {
+    val name: String
+    val machine: IMachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean): List<Short>
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean): String
+
+    fun isInRegularRAM(target: AssignTarget, program: Program): Boolean {
+        val memAddr = target.memoryAddress
+        val arrayIdx = target.arrayindexed
+        val ident = target.identifier
+        when {
+            memAddr != null -> {
+                return when (memAddr.addressExpression) {
+                    is NumericLiteralValue -> {
+                        machine.isRegularRAMaddress((memAddr.addressExpression as NumericLiteralValue).number.toInt())
+                    }
+                    is IdentifierReference -> {
+                        val decl = (memAddr.addressExpression as IdentifierReference).targetVarDecl(program)
+                        if ((decl?.type == VarDeclType.VAR || decl?.type == VarDeclType.CONST) && decl.value is NumericLiteralValue)
+                            machine.isRegularRAMaddress((decl.value as NumericLiteralValue).number.toInt())
+                        else
+                            false
+                    }
+                    else -> false
+                }
+            }
+            arrayIdx != null -> {
+                val targetStmt = arrayIdx.arrayvar.targetVarDecl(program)
+                return if (targetStmt?.type == VarDeclType.MEMORY) {
+                    val addr = targetStmt.value as? NumericLiteralValue
+                    if (addr != null)
+                        machine.isRegularRAMaddress(addr.number.toInt())
+                    else
+                        false
+                } else true
+            }
+            ident != null -> {
+                val decl = ident.targetVarDecl(program)!!
+                return if (decl.type == VarDeclType.MEMORY && decl.value is NumericLiteralValue)
+                    machine.isRegularRAMaddress((decl.value as NumericLiteralValue).number.toInt())
+                else
+                    true
+            }
+            else -> return true
+        }
+    }
+}
+
+
+internal object C64Target: ICompilationTarget {
+    override val name = "c64"
+    override val machine = C64MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean) =
+            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
+            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            in PassByReferenceDatatypes -> machine.POINTER_MEM_SIZE
+            else -> -9999999
+        }
+    }
+}
+
+internal object Cx16Target: ICompilationTarget {
+    override val name = "cx16"
+    override val machine = CX16MachineDefinition
+    override fun encodeString(str: String, altEncoding: Boolean) =
+            if(altEncoding) Petscii.encodeScreencode(str, true) else Petscii.encodePetscii(str, true)
+    override fun decodeString(bytes: List<Short>, altEncoding: Boolean) =
+            if(altEncoding) Petscii.decodeScreencode(bytes, true) else Petscii.decodePetscii(bytes, true)
+
+    override fun memorySize(dt: DataType): Int {
+        return when(dt) {
+            in ByteDatatypes -> 1
+            in WordDatatypes -> 2
+            DataType.FLOAT -> machine.FLOAT_MEM_SIZE
+            in PassByReferenceDatatypes -> machine.POINTER_MEM_SIZE
+            else -> -9999999
+        }
+    }
+}
+
+
+internal fun asmGeneratorFor(
+    compTarget: ICompilationTarget,
+    program: Program,
+    errors: IErrorReporter,
+    zp: Zeropage,
+    options: CompilationOptions,
+    outputDir: Path
+): IAssemblyGenerator
+{
+    // at the moment we only have one code generation backend (for 6502 and 65c02)
+    return AsmGen(program, errors, zp, options, compTarget, outputDir)
+}

compiler/src/prog8/compiler/target/IMachineDefinition.kt

@@ -1,22 +1,20 @@
 package prog8.compiler.target
 
-import prog8.ast.Program
 import prog8.compiler.CompilationOptions
 import prog8.compiler.Zeropage
-import prog8.parser.ModuleImporter
 
 
-internal interface IMachineFloat {
+interface IMachineFloat {
     fun toDouble(): Double
     fun makeFloatFillAsm(): String
 }
 
-internal enum class CpuType {
+enum class CpuType {
     CPU6502,
     CPU65c02
 }
 
-internal interface IMachineDefinition {
+interface IMachineDefinition {
     val FLOAT_MAX_NEGATIVE: Double
     val FLOAT_MAX_POSITIVE: Double
     val FLOAT_MEM_SIZE: Int
@@ -32,8 +30,8 @@ internal interface IMachineDefinition {
 
     fun initializeZeropage(compilerOptions: CompilationOptions)
     fun getFloat(num: Number): IMachineFloat
-    fun getFloatRomConst(number: Double): String?
-    fun importLibs(compilerOptions: CompilationOptions, importer: ModuleImporter, program: Program)
+
+    fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String>
     fun launchEmulator(programName: String)
     fun isRegularRAMaddress(address: Int): Boolean
 }

compiler/src/prog8/compiler/target/c64/AssemblyProgram.kt

@@ -2,13 +2,12 @@ package prog8.compiler.target.c64
 
 import prog8.compiler.CompilationOptions
 import prog8.compiler.OutputType
-import prog8.compiler.target.CompilationTarget
 import prog8.compiler.target.IAssemblyProgram
 import prog8.compiler.target.generatedLabelPrefix
 import java.nio.file.Path
 import kotlin.system.exitProcess
 
-class AssemblyProgram(override val name: String, outputDir: Path) : IAssemblyProgram {
+class AssemblyProgram(override val name: String, outputDir: Path, private val compTarget: String) : IAssemblyProgram {
     private val assemblyFile = outputDir.resolve("$name.asm")
     private val prgFile = outputDir.resolve("$name.prg")
     private val binFile = outputDir.resolve("$name.bin")
@@ -23,12 +22,12 @@ class AssemblyProgram(override val name: String, outputDir: Path) : IAssemblyPro
         val outFile = when (options.output) {
             OutputType.PRG -> {
                 command.add("--cbm-prg")
-                println("\nCreating prg for target ${CompilationTarget.instance.name}.")
+                println("\nCreating prg for target $compTarget.")
                 prgFile
             }
             OutputType.RAW -> {
                 command.add("--nostart")
-                println("\nCreating raw binary for target ${CompilationTarget.instance.name}.")
+                println("\nCreating raw binary for target $compTarget.")
                 binFile
             }
         }

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

@@ -1,13 +1,10 @@
 package prog8.compiler.target.c64
 
-import prog8.ast.Program
 import prog8.compiler.*
 import prog8.compiler.target.CpuType
 import prog8.compiler.target.IMachineDefinition
 import prog8.compiler.target.IMachineFloat
-import prog8.parser.ModuleImporter
 import java.io.IOException
-import java.math.RoundingMode
 import kotlin.math.absoluteValue
 import kotlin.math.pow
 
@@ -24,7 +21,6 @@ internal object C64MachineDefinition: IMachineDefinition {
     override val RAW_LOAD_ADDRESS = 0xc000
 
     // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
-    // and some heavily used string constants derived from the two values above
     override val ESTACK_LO = 0xce00        //  $ce00-$ceff inclusive
     override val ESTACK_HI = 0xcf00        //  $ce00-$ceff inclusive
 
@@ -32,44 +28,11 @@ internal object C64MachineDefinition: IMachineDefinition {
 
     override fun getFloat(num: Number) = Mflpt5.fromNumber(num)
 
-    override fun getFloatRomConst(number: Double): String? {
-        // try to match the ROM float constants to save memory
-        val mflpt5 = Mflpt5.fromNumber(number)
-        val floatbytes = shortArrayOf(mflpt5.b0, mflpt5.b1, mflpt5.b2, mflpt5.b3, mflpt5.b4)
-        when {
-            floatbytes.contentEquals(shortArrayOf(0x00, 0x00, 0x00, 0x00, 0x00)) -> return "floats.FL_ZERO_const"        // not a ROM const
-            floatbytes.contentEquals(shortArrayOf(0x81, 0x00, 0x00, 0x00, 0x00)) -> return "floats.FL_ONE_const"        // not a ROM const
-            floatbytes.contentEquals(shortArrayOf(0x82, 0x49, 0x0f, 0xda, 0xa1)) -> return "floats.FL_PIVAL"
-            floatbytes.contentEquals(shortArrayOf(0x90, 0x80, 0x00, 0x00, 0x00)) -> return "floats.FL_N32768"
-            floatbytes.contentEquals(shortArrayOf(0x81, 0x00, 0x00, 0x00, 0x00)) -> return "floats.FL_FONE"
-            floatbytes.contentEquals(shortArrayOf(0x80, 0x35, 0x04, 0xf3, 0x34)) -> return "floats.FL_SQRHLF"
-            floatbytes.contentEquals(shortArrayOf(0x81, 0x35, 0x04, 0xf3, 0x34)) -> return "floats.FL_SQRTWO"
-            floatbytes.contentEquals(shortArrayOf(0x80, 0x80, 0x00, 0x00, 0x00)) -> return "floats.FL_NEGHLF"
-            floatbytes.contentEquals(shortArrayOf(0x80, 0x31, 0x72, 0x17, 0xf8)) -> return "floats.FL_LOG2"
-            floatbytes.contentEquals(shortArrayOf(0x84, 0x20, 0x00, 0x00, 0x00)) -> return "floats.FL_TENC"
-            floatbytes.contentEquals(shortArrayOf(0x9e, 0x6e, 0x6b, 0x28, 0x00)) -> return "floats.FL_NZMIL"
-            floatbytes.contentEquals(shortArrayOf(0x80, 0x00, 0x00, 0x00, 0x00)) -> return "floats.FL_FHALF"
-            floatbytes.contentEquals(shortArrayOf(0x81, 0x38, 0xaa, 0x3b, 0x29)) -> return "floats.FL_LOGEB2"
-            floatbytes.contentEquals(shortArrayOf(0x81, 0x49, 0x0f, 0xda, 0xa2)) -> return "floats.FL_PIHALF"
-            floatbytes.contentEquals(shortArrayOf(0x83, 0x49, 0x0f, 0xda, 0xa2)) -> return "floats.FL_TWOPI"
-            floatbytes.contentEquals(shortArrayOf(0x7f, 0x00, 0x00, 0x00, 0x00)) -> return "floats.FL_FR4"
-            else -> {
-                // attempt to correct for a few rounding issues
-                when (number.toBigDecimal().setScale(10, RoundingMode.HALF_DOWN).toDouble()) {
-                    3.1415926536 -> return "floats.FL_PIVAL"
-                    1.4142135624 -> return "floats.FL_SQRTWO"
-                    0.7071067812 -> return "floats.FL_SQRHLF"
-                    0.6931471806 -> return "floats.FL_LOG2"
-                    else -> {}
-                }
-            }
-        }
-        return null
-    }
-
-    override fun importLibs(compilerOptions: CompilationOptions, importer: ModuleImporter, program: Program) {
-        if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            importer.importLibraryModule(program, "syslib")
+    override fun importLibs(compilerOptions: CompilationOptions,compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
     }
 
     override fun launchEmulator(programName: String) {
@@ -139,13 +102,14 @@ internal object C64MachineDefinition: IMachineDefinition {
                             0x73, 0x74, 0x75, 0x76, 0x77, 0x78, 0x79, 0x7a, 0x7b, 0x7c,
                             0x7d, 0x7e, 0x7f, 0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8a,
                             0x8b, 0x8c, 0x8d, 0x8e, 0x8f, 0xff
-                            // 0x90-0xfa is 'kernel work storage area'
+                            // 0x90-0xfa is 'kernal work storage area'
                     ))
                 }
 
                 if (options.zeropage == ZeropageType.FLOATSAFE) {
                     // remove the zero page locations used for floating point operations from the free list
                     free.removeAll(listOf(
+                            0x22, 0x23, 0x24, 0x25,
                             0x10, 0x11, 0x12, 0x26, 0x27, 0x28, 0x29, 0x2a,
                             0x57, 0x58, 0x59, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f, 0x60,
                             0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, 0x68,

compiler/src/prog8/compiler/target/c64/Petscii.kt

@@ -172,7 +172,7 @@ object Petscii {
             '\u258c',    //  ▌    0xA1 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0xA2 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0xA3 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0xA4 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0xA4 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0xA5 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0xA6 -> MEDIUM SHADE
             '\u2595',    //  ▕    0xA7 -> RIGHT ONE EIGHTH BLOCK
@@ -236,7 +236,7 @@ object Petscii {
             '\u258c',    //  ▌    0xE1 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0xE2 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0xE3 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0xE4 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0xE4 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0xE5 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0xE6 -> MEDIUM SHADE
             '\u2595',    //  ▕    0xE7 -> RIGHT ONE EIGHTH BLOCK
@@ -431,7 +431,7 @@ object Petscii {
             '\u258c',    //  ▌    0xA1 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0xA2 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0xA3 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0xA4 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0xA4 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0xA5 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0xA6 -> MEDIUM SHADE
             '\u2595',    //  ▕    0xA7 -> RIGHT ONE EIGHTH BLOCK
@@ -495,7 +495,7 @@ object Petscii {
             '\u258c',    //  ▌    0xE1 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0xE2 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0xE3 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0xE4 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0xE4 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0xE5 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0xE6 -> MEDIUM SHADE
             '\u2595',    //  ▕    0xE7 -> RIGHT ONE EIGHTH BLOCK
@@ -626,7 +626,7 @@ object Petscii {
             '\u258c',    //  ▌    0x61 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0x62 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0x63 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0x64 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0x64 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0x65 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0x66 -> MEDIUM SHADE
             '\u2595',    //  ▕    0x67 -> RIGHT ONE EIGHTH BLOCK
@@ -885,7 +885,7 @@ object Petscii {
             '\u258c',    //  ▌    0x61 -> LEFT HALF BLOCK
             '\u2584',    //  ▄    0x62 -> LOWER HALF BLOCK
             '\u2594',    //  ▔    0x63 -> UPPER ONE EIGHTH BLOCK
-            '\u2581',    //  ▁    0x64 -> LOWER ONE EIGHTH BLOCK
+            '_',         //  ▁    0x64 -> LOWER ONE EIGHTH BLOCK
             '\u258f',    //  ▏    0x65 -> LEFT ONE EIGHTH BLOCK
             '\u2592',    //  ▒    0x66 -> MEDIUM SHADE
             '\u2595',    //  ▕    0x67 -> RIGHT ONE EIGHTH BLOCK

compiler/src/prog8/compiler/target/c64/codegen/FunctionCallAsmGen.kt

@@ -1,270 +0,0 @@
-package prog8.compiler.target.c64.codegen
-
-import prog8.ast.IFunctionCall
-import prog8.ast.Node
-import prog8.ast.Program
-import prog8.ast.base.*
-import prog8.ast.expressions.*
-import prog8.ast.statements.RegisterOrStatusflag
-import prog8.ast.statements.Subroutine
-import prog8.ast.statements.SubroutineParameter
-import prog8.compiler.AssemblyError
-import prog8.compiler.target.c64.codegen.assignment.*
-
-
-internal class FunctionCallAsmGen(private val program: Program, private val asmgen: AsmGen) {
-
-    internal fun translateFunctionCallStatement(stmt: IFunctionCall) {
-        val sub = stmt.target.targetSubroutine(program.namespace)!!
-        val preserveStatusRegisterAfterCall = sub.asmReturnvaluesRegisters.any {it.statusflag!=null}
-        translateFunctionCall(stmt, preserveStatusRegisterAfterCall)
-        // functioncalls no longer return results on the stack, so simply ignore the results in the registers
-        if(preserveStatusRegisterAfterCall)
-            asmgen.out("  plp\t; restore status flags from call")
-    }
-
-
-    internal fun translateFunctionCall(stmt: IFunctionCall, preserveStatusRegisterAfterCall: Boolean) {
-        // output the code to setup the parameters and perform the actual call
-        // does NOT output the code to deal with the result values!
-        val sub = stmt.target.targetSubroutine(program.namespace) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
-        val saveX = CpuRegister.X in sub.asmClobbers || sub.regXasResult() || sub.regXasParam()
-        if(saveX)
-            asmgen.saveRegister(CpuRegister.X, preserveStatusRegisterAfterCall, (stmt as Node).definingSubroutine()!!)
-
-        val subName = asmgen.asmSymbolName(stmt.target)
-        if(stmt.args.isNotEmpty()) {
-            if(sub.asmParameterRegisters.isEmpty()) {
-                // via variables
-                for(arg in sub.parameters.withIndex().zip(stmt.args)) {
-                    argumentViaVariable(sub, arg.first, arg.second)
-                }
-            } else {
-                // via registers
-                if(sub.parameters.size==1) {
-                    // just a single parameter, no risk of clobbering registers
-                    argumentViaRegister(sub, IndexedValue(0, sub.parameters.single()), stmt.args[0])
-                } else {
-                    // multiple register arguments, risk of register clobbering.
-                    // evaluate arguments onto the stack, and load the registers from the evaluated values on the stack.
-                    when {
-                        stmt.args.all {it is AddressOf ||
-                                it is NumericLiteralValue ||
-                                it is StringLiteralValue ||
-                                it is ArrayLiteralValue ||
-                                it is IdentifierReference} -> {
-                            // no risk of clobbering for these simple argument types. Optimize the register loading.
-                            for(arg in sub.parameters.withIndex().zip(stmt.args)) {
-                                argumentViaRegister(sub, arg.first, arg.second)
-                            }
-                        }
-                        else -> {
-                            // Risk of clobbering due to complex expression args. Work via the stack.
-                            registerArgsViaStackEvaluation(stmt, sub)
-                        }
-                    }
-                }
-            }
-        }
-        asmgen.out("  jsr  $subName")
-
-        if(preserveStatusRegisterAfterCall) {
-            asmgen.out("  php\t; save status flags from call")
-            // note: the containing statement (such as the FunctionCallStatement or the Assignment or the Expression)
-            //       must take care of popping this value again at the end!
-        }
-
-        if(saveX)
-            asmgen.restoreRegister(CpuRegister.X, preserveStatusRegisterAfterCall)
-    }
-
-    private fun registerArgsViaStackEvaluation(stmt: IFunctionCall, sub: Subroutine) {
-        // this is called when one or more of the arguments are 'complex' and
-        // cannot be assigned to a register easily or risk clobbering other registers.
-
-        if(sub.parameters.isEmpty())
-            return
-
-        // 1. load all arguments reversed onto the stack: first arg goes last (is on top).
-        for (arg in stmt.args.reversed())
-            asmgen.translateExpression(arg)
-
-        var argForCarry: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
-        var argForXregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
-        var argForAregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
-
-        asmgen.out("  inx")     // align estack pointer
-
-        for(argi in stmt.args.zip(sub.asmParameterRegisters).withIndex()) {
-            when {
-                argi.value.second.statusflag == Statusflag.Pc -> {
-                    require(argForCarry == null)
-                    argForCarry = argi
-                }
-                argi.value.second.statusflag != null -> throw AssemblyError("can only use Carry as status flag parameter")
-                argi.value.second.registerOrPair in setOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) -> {
-                    require(argForXregister==null)
-                    argForXregister = argi
-                }
-                argi.value.second.registerOrPair in setOf(RegisterOrPair.A, RegisterOrPair.AY) -> {
-                    require(argForAregister == null)
-                    argForAregister = argi
-                }
-                argi.value.second.registerOrPair == RegisterOrPair.Y -> {
-                    asmgen.out("  ldy  P8ESTACK_LO+${argi.index},x")
-                }
-                else -> throw AssemblyError("weird argument")
-            }
-        }
-
-        if(argForCarry!=null) {
-            asmgen.out("""
-                lda  P8ESTACK_LO+${argForCarry.index},x
-                beq  +
-                sec
-                bcs  ++
-+               clc
-+               php""")             // push the status flags
-        }
-
-        if(argForAregister!=null) {
-            when(argForAregister.value.second.registerOrPair) {
-                RegisterOrPair.A -> asmgen.out("  lda  P8ESTACK_LO+${argForAregister.index},x")
-                RegisterOrPair.AY -> asmgen.out("  lda  P8ESTACK_LO+${argForAregister.index},x |  ldy  P8ESTACK_HI+${argForAregister.index},x")
-                else -> throw AssemblyError("weird arg")
-            }
-        }
-
-        if(argForXregister!=null) {
-
-            if(argForAregister!=null)
-                asmgen.out("  pha")
-            when(argForXregister.value.second.registerOrPair) {
-                RegisterOrPair.X -> asmgen.out("  lda  P8ESTACK_LO+${argForXregister.index},x |  tax")
-                RegisterOrPair.AX -> asmgen.out("  ldy  P8ESTACK_LO+${argForXregister.index},x |  lda  P8ESTACK_HI+${argForXregister.index},x |  tax |  tya")
-                RegisterOrPair.XY -> asmgen.out("  ldy  P8ESTACK_HI+${argForXregister.index},x |  lda  P8ESTACK_LO+${argForXregister.index},x |  tax")
-                else -> throw AssemblyError("weird arg")
-            }
-            if(argForAregister!=null)
-                asmgen.out("  pla")
-        } else {
-            repeat(sub.parameters.size - 1) { asmgen.out("  inx") }       // unwind stack
-        }
-
-        if(argForCarry!=null)
-            asmgen.out("  plp")       // set the carry flag back to correct value
-    }
-
-    private fun argumentViaVariable(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
-        // pass parameter via a regular variable (not via registers)
-        val valueIDt = value.inferType(program)
-        if(!valueIDt.isKnown)
-            throw AssemblyError("unknown dt")
-        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
-        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
-            throw AssemblyError("argument type incompatible")
-
-        val varName = asmgen.asmVariableName(sub.scopedname+"."+parameter.value.name)
-        asmgen.assignExpressionToVariable(value, varName, parameter.value.type, sub)
-    }
-
-    private fun argumentViaRegister(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
-        // pass argument via a register parameter
-        val valueIDt = value.inferType(program)
-        if(!valueIDt.isKnown)
-            throw AssemblyError("unknown dt")
-        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
-        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
-            throw AssemblyError("argument type incompatible")
-
-        val paramRegister = sub.asmParameterRegisters[parameter.index]
-        val statusflag = paramRegister.statusflag
-        val register = paramRegister.registerOrPair
-        val requiredDt = parameter.value.type
-        if(requiredDt!=valueDt) {
-            if(valueDt largerThan requiredDt)
-                throw AssemblyError("can only convert byte values to word param types")
-        }
-        when {
-            statusflag!=null -> {
-                if(requiredDt!=valueDt)
-                    throw AssemblyError("for statusflag, byte value is required")
-                if (statusflag == Statusflag.Pc) {
-                    // this param needs to be set last, right before the jsr
-                    // for now, this is already enforced on the subroutine definition by the Ast Checker
-                    when(value) {
-                        is NumericLiteralValue -> {
-                            val carrySet = value.number.toInt() != 0
-                            asmgen.out(if(carrySet) "  sec" else "  clc")
-                        }
-                        is IdentifierReference -> {
-                            val sourceName = asmgen.asmVariableName(value)
-                            asmgen.out("""
-            pha
-            lda  $sourceName
-            beq  +
-            sec  
-            bcs  ++
-+           clc
-+           pla
-""")
-                        }
-                        else -> {
-                            asmgen.assignExpressionToRegister(value, RegisterOrPair.A)
-                            asmgen.out("""
-                                beq  +
-                                sec
-                                bcs  ++
-+                               clc
-+""")
-                        }
-                    }
-                }
-                else throw AssemblyError("can only use Carry as status flag parameter")
-            }
-            else -> {
-                // via register or register pair
-                register!!
-                if(requiredDt largerThan valueDt) {
-                    // we need to sign extend the source, do this via temporary word variable
-                    val scratchVar = asmgen.asmVariableName("P8ZP_SCRATCH_W1")
-                    asmgen.assignExpressionToVariable(value, scratchVar, DataType.UBYTE, sub)
-                    asmgen.signExtendVariableLsb(scratchVar, valueDt)
-                    asmgen.assignVariableToRegister(scratchVar, register)
-                }
-                else {
-                    val target = AsmAssignTarget.fromRegisters(register, sub, program, asmgen)
-                    val src = if(valueDt in PassByReferenceDatatypes) {
-                        if(value is IdentifierReference) {
-                            val addr = AddressOf(value, Position.DUMMY)
-                            AsmAssignSource.fromAstSource(addr, program, asmgen).adjustSignedUnsigned(target)
-                        } else {
-                            AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
-                        }
-                    } else {
-                        AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
-                    }
-                    asmgen.translateNormalAssignment(AsmAssignment(src, target, false, Position.DUMMY))
-                }
-            }
-        }
-    }
-
-    private fun isArgumentTypeCompatible(argType: DataType, paramType: DataType): Boolean {
-        if(argType isAssignableTo paramType)
-            return true
-        if(argType in ByteDatatypes && paramType in ByteDatatypes)
-            return true
-        if(argType in WordDatatypes && paramType in WordDatatypes)
-            return true
-
-        // we have a special rule for some types.
-        // strings are assignable to UWORD, for example, and vice versa
-        if(argType==DataType.STR && paramType==DataType.UWORD)
-            return true
-        if(argType==DataType.UWORD && paramType == DataType.STR)
-            return true
-
-        return false
-    }
-}

compiler/src/prog8/compiler/target/cpu6502/codegen/AsmOptimizer.kt

@@ -1,4 +1,4 @@
-package prog8.compiler.target.c64.codegen
+package prog8.compiler.target.cpu6502.codegen
 
 
 // note: see https://wiki.nesdev.com/w/index.php/6502_assembly_optimisations
@@ -75,11 +75,11 @@ private fun getLinesBy(lines: MutableList<String>, windowSize: Int) =
 
 private fun optimizeCmpSequence(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
     // the when statement (on bytes) generates a sequence of:
-    //	 lda $ce01,x
-    //	 cmp #$20
+    //	 lda  $ce01,x
+    //	 cmp  #$20
     //	 beq  check_prog8_s72choice_32
-    //	 lda $ce01,x
-    //	 cmp #$21
+    //	 lda  $ce01,x
+    //	 cmp  #$21
     //	 beq  check_prog8_s73choice_33
     // the repeated lda can be removed
     val mods = mutableListOf<Modification>()
@@ -179,7 +179,7 @@ private fun optimizeSameAssignments(linesByFourteen: List<List<IndexedValue<Stri
 }
 
 private fun optimizeStoreLoadSame(linesByFour: List<List<IndexedValue<String>>>): List<Modification> {
-    // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can be eliminated
+    // sta X + lda X,  sty X + ldy X,   stx X + ldx X  -> the second instruction can OFTEN be eliminated
     // TODO this is not true if X is not a regular RAM memory address (but instead mapped I/O or ROM)
     val mods = mutableListOf<Modification>()
     for (pair in linesByFour) {
@@ -196,10 +196,14 @@ private fun optimizeStoreLoadSame(linesByFour: List<List<IndexedValue<String>>>)
                 (first.startsWith("sty ") && second.startsWith("ldy ")) ||
                 (first.startsWith("stx ") && second.startsWith("ldx "))
         ) {
-            val firstLoc = first.substring(4).trimStart()
-            val secondLoc = second.substring(4).trimStart()
-            if (firstLoc == secondLoc) {
-                mods.add(Modification(pair[1].index, true, null))
+            val third = pair[2].value.trimStart()
+            if(!third.startsWith("b")) {
+                // no branch instruction follows, we can potentiall remove the load instruction
+                val firstLoc = first.substring(4).trimStart()
+                val secondLoc = second.substring(4).trimStart()
+                if (firstLoc == secondLoc) {
+                    mods.add(Modification(pair[1].index, true, null))
+                }
             }
         }
     }

compiler/src/prog8/compiler/target/cpu6502/codegen/BuiltinFunctionsAsmGen.kt

@@ -1,4 +1,4 @@
-package prog8.compiler.target.c64.codegen
+package prog8.compiler.target.cpu6502.codegen
 
 import prog8.ast.IFunctionCall
 import prog8.ast.Node
@@ -9,176 +9,190 @@ import prog8.ast.statements.ArrayIndex
 import prog8.ast.statements.DirectMemoryWrite
 import prog8.ast.statements.FunctionCallStatement
 import prog8.ast.statements.Subroutine
+import prog8.ast.toHex
 import prog8.compiler.AssemblyError
-import prog8.compiler.target.CompilationTarget
-import prog8.compiler.target.Cx16Target
-import prog8.compiler.target.c64.codegen.assignment.AsmAssignSource
-import prog8.compiler.target.c64.codegen.assignment.AsmAssignTarget
-import prog8.compiler.target.c64.codegen.assignment.AsmAssignment
-import prog8.compiler.target.c64.codegen.assignment.SourceStorageKind
-import prog8.compiler.target.c64.codegen.assignment.TargetStorageKind
-import prog8.compiler.toHex
-import prog8.functions.FSignature
+import prog8.compiler.functions.FSignature
+import prog8.compiler.target.CpuType
+import prog8.compiler.target.cpu6502.codegen.assignment.*
+import prog8.compiler.target.subroutineFloatEvalResultVar2
 
-internal class BuiltinFunctionsAsmGen(private val program: Program, private val asmgen: AsmGen) {
+internal class BuiltinFunctionsAsmGen(private val program: Program, private val asmgen: AsmGen, private val assignAsmGen: AssignmentAsmGen) {
 
-    internal fun translateFunctioncallExpression(fcall: FunctionCall, func: FSignature, resultToStack: Boolean) {
-        translateFunctioncall(fcall, func, discardResult = false, resultToStack = resultToStack)
+    internal fun translateFunctioncallExpression(fcall: FunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+        translateFunctioncall(fcall, func, discardResult = false, resultToStack = resultToStack, resultRegister = resultRegister)
     }
 
     internal fun translateFunctioncallStatement(fcall: FunctionCallStatement, func: FSignature) {
-        translateFunctioncall(fcall, func, discardResult = true, resultToStack = false)
+        translateFunctioncall(fcall, func, discardResult = true, resultToStack = false, resultRegister = null)
     }
 
-    private fun translateFunctioncall(fcall: IFunctionCall, func: FSignature, discardResult: Boolean, resultToStack: Boolean) {
+    private fun translateFunctioncall(fcall: IFunctionCall, func: FSignature, discardResult: Boolean, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
         if (discardResult && func.pure)
             return  // can just ignore the whole function call altogether
 
         if(discardResult && resultToStack)
             throw AssemblyError("cannot both discard the result AND put it onto stack")
 
-        val scope = (fcall as Node).definingSubroutine()!!
+        val sscope = (fcall as Node).definingSubroutine()
+
         when (func.name) {
-            "msb" -> funcMsb(fcall, resultToStack)
-            "lsb" -> funcLsb(fcall, resultToStack)
-            "mkword" -> funcMkword(fcall, resultToStack)
-            "abs" -> funcAbs(fcall, func, resultToStack, scope)
+            "msb" -> funcMsb(fcall, resultToStack, resultRegister)
+            "lsb" -> funcLsb(fcall, resultToStack, resultRegister)
+            "mkword" -> funcMkword(fcall, resultToStack, resultRegister)
+            "abs" -> funcAbs(fcall, func, resultToStack, resultRegister, sscope)
             "swap" -> funcSwap(fcall)
-            "min", "max" -> funcMinMax(fcall, func, resultToStack)
-            "sum" -> funcSum(fcall, resultToStack)
-            "any", "all" -> funcAnyAll(fcall, func, resultToStack)
+            "min", "max" -> funcMinMax(fcall, func, resultToStack, resultRegister, sscope)
+            "sum" -> funcSum(fcall, resultToStack, resultRegister, sscope)
+            "any", "all" -> funcAnyAll(fcall, func, resultToStack, resultRegister, sscope)
             "sin8", "sin8u", "sin16", "sin16u",
-            "cos8", "cos8u", "cos16", "cos16u" -> funcSinCosInt(fcall, func, resultToStack, scope)
-            "sgn" -> funcSgn(fcall, func, resultToStack, scope)
+            "cos8", "cos8u", "cos16", "cos16u" -> funcSinCosInt(fcall, func, resultToStack, resultRegister, sscope)
+            "sgn" -> funcSgn(fcall, func, resultToStack, resultRegister, sscope)
             "sin", "cos", "tan", "atan",
             "ln", "log2", "sqrt", "rad",
             "deg", "round", "floor", "ceil",
-            "rndf" -> funcVariousFloatFuncs(fcall, func, resultToStack, scope)
-            "rnd", "rndw" -> funcRnd(func, resultToStack)
-            "sqrt16" -> funcSqrt16(fcall, func, resultToStack, scope)
+            "rndf" -> funcVariousFloatFuncs(fcall, func, resultToStack, resultRegister, sscope)
+            "fastrnd8", "rnd", "rndw" -> funcRnd(func, resultToStack, resultRegister, sscope)
+            "sqrt16" -> funcSqrt16(fcall, func, resultToStack, resultRegister, sscope)
             "rol" -> funcRol(fcall)
             "rol2" -> funcRol2(fcall)
             "ror" -> funcRor(fcall)
             "ror2" -> funcRor2(fcall)
             "sort" -> funcSort(fcall)
             "reverse" -> funcReverse(fcall)
-            "rsave" -> {
-                // save cpu status flag and all registers A, X, Y.
-                // see http://6502.org/tutorials/register_preservation.html
-                asmgen.out(" php |  sta  P8ZP_SCRATCH_REG | pha  | txa  | pha  | tya  | pha  | lda  P8ZP_SCRATCH_REG")
-            }
-            "rrestore" -> {
-                // restore all registers and cpu status flag
-                asmgen.out(" pla |  tay |  pla |  tax |  pla |  plp")
-            }
-            "read_flags" -> {
-                if(resultToStack)
-                    asmgen.out("  jsr  prog8_lib.func_read_flags_stack")
-                else
-                    asmgen.out("  php |  pla")
-            }
-            "clear_carry" -> asmgen.out("  clc")
-            "set_carry" -> asmgen.out("  sec")
-            "clear_irqd" -> asmgen.out("  cli")
-            "set_irqd" -> asmgen.out("  sei")
-            "strlen" -> funcStrlen(fcall, resultToStack)
-            "strcmp" -> funcStrcmp(fcall, func, resultToStack, scope)
-            "memcopy", "memset", "memsetw" -> funcMemSetCopy(fcall, func, scope)
-            "substr", "leftstr", "rightstr" -> {
-                translateArguments(fcall.args, func, scope)
-                asmgen.out("  jsr  prog8_lib.func_${func.name}")
-            }
-            "exit" -> {
-                translateArguments(fcall.args, func, scope)
-                asmgen.out("  jmp  prog8_lib.func_exit")
-            }
-            else -> TODO("missing asmgen for builtin func ${func.name}")
+            "memory" -> funcMemory(fcall, discardResult, resultToStack, resultRegister)
+            "peekw" -> funcPeekW(fcall, resultToStack, resultRegister)
+            "peek" -> throw AssemblyError("peek() should have been replaced by @()")
+            "pokew" -> funcPokeW(fcall)
+            "poke" -> throw AssemblyError("poke() should have been replaced by @()")
+            "cmp" -> funcCmp(fcall)
+            else -> throw AssemblyError("missing asmgen for builtin func ${func.name}")
         }
     }
 
-    private fun funcMemSetCopy(fcall: IFunctionCall, func: FSignature, scope: Subroutine) {
-        if(CompilationTarget.instance is Cx16Target) {
-            when(func.name) {
-                "memset" -> {
-                    // use the ROM function of the Cx16
-                    asmgen.assignExpressionToVariable(fcall.args[0], "cx16.r0", DataType.UWORD, scope)
-                    asmgen.assignExpressionToVariable(fcall.args[1], "cx16.r1", DataType.UWORD, scope)
-                    asmgen.assignExpressionToRegister(fcall.args[2], RegisterOrPair.A)
-                    val sub = (fcall as FunctionCallStatement).definingSubroutine()!!
-                    asmgen.saveRegister(CpuRegister.X, false, sub)
-                    asmgen.out("  jsr  cx16.memory_fill")
-                    asmgen.restoreRegister(CpuRegister.X, false)
-                }
-                "memcopy" -> {
-                    val count = fcall.args[2].constValue(program)?.number?.toInt()
-                    val countDt = fcall.args[2].inferType(program)
-                    if((count!=null && count <= 255) || countDt.istype(DataType.UBYTE) || countDt.istype(DataType.BYTE)) {
-                        // fast memcopy of up to 255
-                        translateArguments(fcall.args, func, scope)
-                        asmgen.out("  jsr  prog8_lib.func_memcopy255")
-                        return
+    private fun funcCmp(fcall: IFunctionCall) {
+        val arg1 = fcall.args[0]
+        val arg2 = fcall.args[1]
+        val dt1 = arg1.inferType(program).typeOrElse(DataType.STRUCT)
+        val dt2 = arg2.inferType(program).typeOrElse(DataType.STRUCT)
+        if(dt1 in ByteDatatypes) {
+            if(dt2 in ByteDatatypes) {
+                when (arg2) {
+                    is IdentifierReference -> {
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.A)
+                        asmgen.out("  cmp  ${asmgen.asmVariableName(arg2)}")
+                    }
+                    is NumericLiteralValue -> {
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.A)
+                        asmgen.out("  cmp  #${arg2.number}")
+                    }
+                    is DirectMemoryRead -> {
+                        if(arg2.addressExpression is NumericLiteralValue) {
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.A)
+                            asmgen.out("  cmp  ${arg2.addressExpression.constValue(program)!!.number.toHex()}")
+                        } else {
+                            asmgen.assignExpressionToVariable(arg2, "P8ZP_SCRATCH_B1", DataType.UBYTE, (fcall as Node).definingSubroutine())
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.A)
+                            asmgen.out("  cmp  P8ZP_SCRATCH_B1")
+                        }
+                    }
+                    else -> {
+                        asmgen.assignExpressionToVariable(arg2, "P8ZP_SCRATCH_B1", DataType.UBYTE, (fcall as Node).definingSubroutine())
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.A)
+                        asmgen.out("  cmp  P8ZP_SCRATCH_B1")
                     }
-
-                    // use the ROM function of the Cx16
-                    asmgen.assignExpressionToVariable(fcall.args[0], "cx16.r0", DataType.UWORD, scope)
-                    asmgen.assignExpressionToVariable(fcall.args[1], "cx16.r1", DataType.UWORD, scope)
-                    asmgen.assignExpressionToVariable(fcall.args[2], "cx16.r2", DataType.UWORD, scope)
-                    val sub = (fcall as FunctionCallStatement).definingSubroutine()!!
-                    asmgen.saveRegister(CpuRegister.X, false, sub)
-                    asmgen.out("  jsr  cx16.memory_copy")
-                    asmgen.restoreRegister(CpuRegister.X, false)
                 }
-                "memsetw" -> {
-                    translateArguments(fcall.args, func, scope)
-                    asmgen.out("  jsr  prog8_lib.func_memsetw")
-                }
-            }
+            } else
+                throw AssemblyError("args for cmp() should have same dt")
         } else {
-            if(func.name=="memcopy") {
-                val count = fcall.args[2].constValue(program)?.number?.toInt()
-                val countDt = fcall.args[2].inferType(program)
-                if((count!=null && count <= 255) || countDt.istype(DataType.UBYTE) || countDt.istype(DataType.BYTE)) {
-                    translateArguments(fcall.args, func, scope)
-                    asmgen.out("  jsr  prog8_lib.func_memcopy255")
-                    return
+            // dt1 is a word
+            if(dt2 in WordDatatypes) {
+                when (arg2) {
+                    is IdentifierReference -> {
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                        asmgen.out("""
+                            cpy  ${asmgen.asmVariableName(arg2)}+1
+                            bne  +
+                            cmp  ${asmgen.asmVariableName(arg2)}
++""")
+                    }
+                    is NumericLiteralValue -> {
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                        asmgen.out("""
+                            cpy  #>${arg2.number}
+                            bne  +
+                            cmp  #<${arg2.number}
++""")
+                    }
+                    else -> {
+                        asmgen.assignExpressionToVariable(arg2, "P8ZP_SCRATCH_W1", DataType.UWORD, (fcall as Node).definingSubroutine())
+                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                        asmgen.out("""
+                            cpy  P8ZP_SCRATCH_W1+1
+                            bne  +
+                            cmp  P8ZP_SCRATCH_W1
++""")
+                    }
                 }
-            }
-            translateArguments(fcall.args, func, scope)
-            asmgen.out("  jsr  prog8_lib.func_${func.name}")
+            } else
+                throw AssemblyError("args for cmp() should have same dt")
         }
     }
 
-    private fun funcStrcmp(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
-        translateArguments(fcall.args, func, scope)
-        if(resultToStack)
-            asmgen.out("  jsr  prog8_lib.func_strcmp_stack")
-        else
-            asmgen.out("  jsr  prog8_lib.func_strcmp")
+    private fun funcMemory(fcall: IFunctionCall, discardResult: Boolean, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+        if(discardResult || fcall !is FunctionCall)
+            throw AssemblyError("should not discard result of memory allocation at $fcall")
+        val nameRef = fcall.args[0] as IdentifierReference
+        val name = (nameRef.targetVarDecl(program)!!.value as StringLiteralValue).value
+        val size = (fcall.args[1] as NumericLiteralValue).number.toInt()
+
+        val existingSize = asmgen.slabs[name]
+        if(existingSize!=null && existingSize!=size)
+            throw AssemblyError("memory slab '$name' already exists with a different size ($size) at ${fcall.position}")
+
+        val slabname = IdentifierReference(listOf("prog8_slabs", name), fcall.position)
+        slabname.linkParents(fcall)
+        val src = AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, DataType.UWORD, expression = AddressOf(slabname, fcall.position))
+        val target =
+            if(resultToStack)
+                AsmAssignTarget(TargetStorageKind.STACK, program, asmgen, DataType.UWORD, null)
+            else
+                AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, null, program, asmgen)
+        val assign = AsmAssignment(src, target, false, program.memsizer, fcall.position)
+        asmgen.translateNormalAssignment(assign)
+        asmgen.slabs[name] = size
     }
 
-    private fun funcSqrt16(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
+    private fun funcSqrt16(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         translateArguments(fcall.args, func, scope)
         if(resultToStack)
             asmgen.out("  jsr  prog8_lib.func_sqrt16_stack")
-        else
+        else {
             asmgen.out("  jsr  prog8_lib.func_sqrt16_into_A")
+            assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+        }
     }
 
-    private fun funcSinCosInt(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
+    private fun funcSinCosInt(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         translateArguments(fcall.args, func, scope)
         if(resultToStack)
             asmgen.out("  jsr  prog8_lib.func_${func.name}_stack")
         else
             when(func.name) {
-                "sin8", "sin8u", "cos8", "cos8u" -> asmgen.out("  jsr  prog8_lib.func_${func.name}_into_A")
-                "sin16", "sin16u", "cos16", "cos16u" -> asmgen.out("  jsr  prog8_lib.func_${func.name}_into_AY")
+                "sin8", "sin8u", "cos8", "cos8u" -> {
+                    asmgen.out("  jsr  prog8_lib.func_${func.name}_into_A")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
+                "sin16", "sin16u", "cos16", "cos16u" -> {
+                    asmgen.out("  jsr  prog8_lib.func_${func.name}_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
             }
     }
 
     private fun funcReverse(fcall: IFunctionCall) {
         val variable = fcall.args.single()
         if (variable is IdentifierReference) {
-            val decl = variable.targetVarDecl(program.namespace)!!
+            val decl = variable.targetVarDecl(program)!!
             val varName = asmgen.asmVariableName(variable)
             val numElements = decl.arraysize!!.constIndex()
             when (decl.datatype) {
@@ -217,7 +231,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
     private fun funcSort(fcall: IFunctionCall) {
         val variable = fcall.args.single()
         if (variable is IdentifierReference) {
-            val decl = variable.targetVarDecl(program.namespace)!!
+            val decl = variable.targetVarDecl(program)!!
             val varName = asmgen.asmVariableName(variable)
             val numElements = decl.arraysize!!.constIndex()
             when (decl.datatype) {
@@ -304,11 +318,23 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                             val number = (what.addressExpression as NumericLiteralValue).number
                             asmgen.out("  ror  ${number.toHex()}")
                         } else {
-                            asmgen.assignExpressionToRegister(what.addressExpression, RegisterOrPair.AY)
-                            asmgen.out("""
-                                sta  (+) + 1
-                                sty  (+) + 2
-+                               ror  ${'$'}ffff            ; modified""")
+                            val ptrAndIndex = asmgen.pointerViaIndexRegisterPossible(what.addressExpression)
+                            if(ptrAndIndex!=null) {
+                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.X)
+                                asmgen.saveRegisterLocal(CpuRegister.X, (fcall as FunctionCallStatement).definingSubroutine()!!)
+                                asmgen.assignExpressionToRegister(ptrAndIndex.first, RegisterOrPair.AY)
+                                asmgen.restoreRegisterLocal(CpuRegister.X)
+                                asmgen.out("""
+                                    sta  (+) + 1
+                                    sty  (+) + 2
++                                   ror  ${'$'}ffff,x           ; modified""")
+                            } else {
+                                asmgen.assignExpressionToRegister(what.addressExpression, RegisterOrPair.AY)
+                                asmgen.out("""
+                                    sta  (+) + 1
+                                    sty  (+) + 2
++                                   ror  ${'$'}ffff            ; modified""")
+                            }
                         }
                     }
                     is IdentifierReference -> {
@@ -393,11 +419,23 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                             val number = (what.addressExpression as NumericLiteralValue).number
                             asmgen.out("  rol  ${number.toHex()}")
                         } else {
-                            asmgen.assignExpressionToRegister(what.addressExpression, RegisterOrPair.AY)
-                            asmgen.out("""
-                                sta  (+) + 1
-                                sty  (+) + 2
-+                               rol  ${'$'}ffff            ; modified""")
+                            val ptrAndIndex = asmgen.pointerViaIndexRegisterPossible(what.addressExpression)
+                            if(ptrAndIndex!=null) {
+                                asmgen.assignExpressionToRegister(ptrAndIndex.second, RegisterOrPair.X)
+                                asmgen.saveRegisterLocal(CpuRegister.X, (fcall as FunctionCallStatement).definingSubroutine()!!)
+                                asmgen.assignExpressionToRegister(ptrAndIndex.first, RegisterOrPair.AY)
+                                asmgen.restoreRegisterLocal(CpuRegister.X)
+                                asmgen.out("""
+                                    sta  (+) + 1
+                                    sty  (+) + 2
++                                   rol  ${'$'}ffff,x           ; modified""")
+                            } else {
+                                asmgen.assignExpressionToRegister(what.addressExpression, RegisterOrPair.AY)
+                                asmgen.out("""
+                                    sta  (+) + 1
+                                    sty  (+) + 2
++                                   rol  ${'$'}ffff            ; modified""")
+                            }
                         }
                     }
                     is IdentifierReference -> {
@@ -426,19 +464,20 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
 
     private fun translateRolRorArrayArgs(arrayvar: IdentifierReference, indexer: ArrayIndex, operation: String, dt: Char) {
         asmgen.assignExpressionToVariable(AddressOf(arrayvar, arrayvar.position), "prog8_lib.${operation}_array_u${dt}._arg_target", DataType.UWORD, null)
-        val indexerExpr = if(indexer.indexVar!=null) indexer.indexVar!! else indexer.indexNum!!
-        asmgen.assignExpressionToVariable(indexerExpr, "prog8_lib.${operation}_array_u${dt}._arg_index", DataType.UBYTE, null)
+        asmgen.assignExpressionToVariable(indexer.indexExpr, "prog8_lib.${operation}_array_u${dt}._arg_index", DataType.UBYTE, null)
     }
 
-    private fun funcVariousFloatFuncs(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
+    private fun funcVariousFloatFuncs(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         translateArguments(fcall.args, func, scope)
         if(resultToStack)
             asmgen.out("  jsr  floats.func_${func.name}_stack")
-        else
+        else {
             asmgen.out("  jsr  floats.func_${func.name}_fac1")
+            assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, scope, program, asmgen))
+        }
     }
 
-    private fun funcSgn(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
+    private fun funcSgn(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         translateArguments(fcall.args, func, scope)
         val dt = fcall.args.single().inferType(program)
         if(resultToStack) {
@@ -459,10 +498,11 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                 DataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_into_A")
                 else -> throw AssemblyError("weird type $dt")
             }
+            assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
         }
     }
 
-    private fun funcAnyAll(fcall: IFunctionCall, function: FSignature, resultToStack: Boolean) {
+    private fun funcAnyAll(fcall: IFunctionCall, function: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         outputAddressAndLenghtOfArray(fcall.args[0])
         val dt = fcall.args.single().inferType(program)
         if(resultToStack) {
@@ -479,10 +519,11 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                 DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_${function.name}_f_into_A")
                 else -> throw AssemblyError("weird type $dt")
             }
+            assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
         }
     }
 
-    private fun funcMinMax(fcall: IFunctionCall, function: FSignature, resultToStack: Boolean) {
+    private fun funcMinMax(fcall: IFunctionCall, function: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         outputAddressAndLenghtOfArray(fcall.args[0])
         val dt = fcall.args.single().inferType(program)
         if(resultToStack) {
@@ -496,17 +537,32 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
             }
         } else {
             when (dt.typeOrElse(DataType.STRUCT)) {
-                DataType.ARRAY_UB, DataType.STR -> asmgen.out("  jsr  prog8_lib.func_${function.name}_ub_into_A")
-                DataType.ARRAY_B -> asmgen.out("  jsr  prog8_lib.func_${function.name}_b_into_A")
-                DataType.ARRAY_UW -> asmgen.out("  jsr  prog8_lib.func_${function.name}_uw_into_AY")
-                DataType.ARRAY_W -> asmgen.out("  jsr  prog8_lib.func_${function.name}_w_into_AY")
-                DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_${function.name}_f_fac1")
+                DataType.ARRAY_UB, DataType.STR -> {
+                    asmgen.out("  jsr  prog8_lib.func_${function.name}_ub_into_A")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
+                DataType.ARRAY_B -> {
+                    asmgen.out("  jsr  prog8_lib.func_${function.name}_b_into_A")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
+                DataType.ARRAY_UW -> {
+                    asmgen.out("  jsr  prog8_lib.func_${function.name}_uw_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_W -> {
+                    asmgen.out("  jsr  prog8_lib.func_${function.name}_w_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_F -> {
+                    asmgen.out("  jsr  floats.func_${function.name}_f_fac1")
+                    assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, scope, program, asmgen))
+                }
                 else -> throw AssemblyError("weird type $dt")
             }
         }
     }
 
-    private fun funcSum(fcall: IFunctionCall, resultToStack: Boolean) {
+    private fun funcSum(fcall: IFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         outputAddressAndLenghtOfArray(fcall.args[0])
         val dt = fcall.args.single().inferType(program)
         if(resultToStack) {
@@ -520,30 +576,31 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
             }
         } else {
             when (dt.typeOrElse(DataType.STRUCT)) {
-                DataType.ARRAY_UB, DataType.STR -> asmgen.out("  jsr  prog8_lib.func_sum_ub_into_AY")
-                DataType.ARRAY_B -> asmgen.out("  jsr  prog8_lib.func_sum_b_into_AY")
-                DataType.ARRAY_UW -> asmgen.out("  jsr  prog8_lib.func_sum_uw_into_AY")
-                DataType.ARRAY_W -> asmgen.out("  jsr  prog8_lib.func_sum_w_into_AY")
-                DataType.ARRAY_F -> asmgen.out("  jsr  floats.func_sum_f_fac1")
+                DataType.ARRAY_UB, DataType.STR -> {
+                    asmgen.out("  jsr  prog8_lib.func_sum_ub_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_B -> {
+                    asmgen.out("  jsr  prog8_lib.func_sum_b_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_UW -> {
+                    asmgen.out("  jsr  prog8_lib.func_sum_uw_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_W -> {
+                    asmgen.out("  jsr  prog8_lib.func_sum_w_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.ARRAY_F -> {
+                    asmgen.out("  jsr  floats.func_sum_f_fac1")
+                    assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, scope, program, asmgen))
+                }
                 else -> throw AssemblyError("weird type $dt")
             }
         }
     }
 
-    private fun funcStrlen(fcall: IFunctionCall, resultToStack: Boolean) {
-        val name = asmgen.asmVariableName(fcall.args[0] as IdentifierReference)
-        val type = fcall.args[0].inferType(program)
-        when {
-            type.istype(DataType.STR) -> asmgen.out("  lda  #<$name |  ldy  #>$name")
-            type.istype(DataType.UWORD) -> asmgen.out("  lda  $name |  ldy  $name+1")
-            else -> throw AssemblyError("strlen requires str or uword arg")
-        }
-        if(resultToStack)
-            asmgen.out("  jsr  prog8_lib.func_strlen_stack")
-        else
-            asmgen.out("  jsr  prog8_lib.func_strlen_into_A")
-    }
-
     private fun funcSwap(fcall: IFunctionCall) {
         val first = fcall.args[0]
         val second = fcall.args[1]
@@ -588,6 +645,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
 
         // optimized simple case: swap two memory locations
         if(first is DirectMemoryRead && second is DirectMemoryRead) {
+            // TODO optimize swap of two memread values with index, using the same pointer expression/variable, like swap(@(ptr+1), @(ptr+2))
             val addr1 = (first.addressExpression as? NumericLiteralValue)?.number?.toHex()
             val addr2 = (second.addressExpression as? NumericLiteralValue)?.number?.toHex()
             val name1 = if(first.addressExpression is IdentifierReference) asmgen.asmVariableName(first.addressExpression as IdentifierReference) else null
@@ -621,10 +679,10 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                 throw AssemblyError("unknown dt")
             val elementDt = elementIDt.typeOrElse(DataType.STRUCT)
 
-            val firstNum = first.indexer.indexNum
-            val firstVar = first.indexer.indexVar
-            val secondNum = second.indexer.indexNum
-            val secondVar = second.indexer.indexVar
+            val firstNum = first.indexer.indexExpr as? NumericLiteralValue
+            val firstVar = first.indexer.indexExpr as? IdentifierReference
+            val secondNum = second.indexer.indexExpr as? NumericLiteralValue
+            val secondVar = second.indexer.indexExpr as? IdentifierReference
 
             if(firstNum!=null && secondNum!=null) {
                 swapArrayValues(elementDt, arrayVarName1, firstNum, arrayVarName2, secondNum)
@@ -660,12 +718,12 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                 val assignFirst = AsmAssignment(
                         AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, datatype, variableAsmName = "P8ZP_SCRATCH_W2"),
                         targetFromExpr(first, datatype),
-                        false, first.position
+                        false, program.memsizer, first.position
                 )
                 val assignSecond = AsmAssignment(
                         AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, datatype, variableAsmName = "P8ZP_SCRATCH_W1"),
                         targetFromExpr(second, datatype),
-                        false, second.position
+                        false, program.memsizer, second.position
                 )
                 asmgen.translateNormalAssignment(assignFirst)
                 asmgen.translateNormalAssignment(assignSecond)
@@ -677,12 +735,12 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                 val assignFirst = AsmAssignment(
                         AsmAssignSource(SourceStorageKind.STACK, program, asmgen, DataType.FLOAT),
                         targetFromExpr(first, datatype),
-                        false, first.position
+                        false, program.memsizer, first.position
                 )
                 val assignSecond = AsmAssignment(
                         AsmAssignSource(SourceStorageKind.STACK, program, asmgen, DataType.FLOAT),
                         targetFromExpr(second, datatype),
-                        false, second.position
+                        false, program.memsizer, second.position
                 )
                 asmgen.translateNormalAssignment(assignFirst)
                 asmgen.translateNormalAssignment(assignSecond)
@@ -692,8 +750,8 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
     }
 
     private fun swapArrayValues(elementDt: DataType, arrayVarName1: String, indexValue1: NumericLiteralValue, arrayVarName2: String, indexValue2: NumericLiteralValue) {
-        val index1 = indexValue1.number.toInt() * elementDt.memorySize()
-        val index2 = indexValue2.number.toInt() * elementDt.memorySize()
+        val index1 = indexValue1.number.toInt() * program.memsizer.memorySize(elementDt)
+        val index2 = indexValue2.number.toInt() * program.memsizer.memorySize(elementDt)
         when(elementDt) {
             DataType.UBYTE, DataType.BYTE -> {
                 asmgen.out("""
@@ -806,7 +864,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
     }
 
     private fun swapArrayValues(elementDt: DataType, arrayVarName1: String, indexValue1: NumericLiteralValue, arrayVarName2: String, indexName2: IdentifierReference) {
-        val index1 = indexValue1.number.toInt() * elementDt.memorySize()
+        val index1 = indexValue1.number.toInt() * program.memsizer.memorySize(elementDt)
         val idxAsmName2 = asmgen.asmVariableName(indexName2)
         when(elementDt) {
             DataType.UBYTE, DataType.BYTE -> {
@@ -865,7 +923,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
 
     private fun swapArrayValues(elementDt: DataType, arrayVarName1: String, indexName1: IdentifierReference, arrayVarName2: String, indexValue2: NumericLiteralValue) {
         val idxAsmName1 = asmgen.asmVariableName(indexName1)
-        val index2 = indexValue2.number.toInt() * elementDt.memorySize()
+        val index2 = indexValue2.number.toInt() * program.memsizer.memorySize(elementDt)
         when(elementDt) {
             DataType.UBYTE, DataType.BYTE -> {
                 asmgen.out("""
@@ -921,7 +979,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
         }
     }
 
-    private fun funcAbs(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, scope: Subroutine) {
+    private fun funcAbs(fcall: IFunctionCall, func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         translateArguments(fcall.args, func, scope)
         val dt = fcall.args.single().inferType(program).typeOrElse(DataType.STRUCT)
         if(resultToStack) {
@@ -933,40 +991,229 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
             }
         } else {
             when (dt) {
-                in ByteDatatypes -> asmgen.out("  jsr  prog8_lib.abs_b_into_A")
-                in WordDatatypes -> asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
-                DataType.FLOAT -> asmgen.out("  jsr  floats.abs_f_fac1")
+                in ByteDatatypes -> {
+                    asmgen.out("  jsr  prog8_lib.abs_b_into_A")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
+                in WordDatatypes -> {
+                    asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
+                DataType.FLOAT -> {
+                    asmgen.out("  jsr  floats.abs_f_fac1")
+                    assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, scope, program, asmgen))
+                }
                 else -> throw AssemblyError("weird type")
             }
         }
     }
 
-    private fun funcRnd(func: FSignature, resultToStack: Boolean) {
+    private fun funcRnd(func: FSignature, resultToStack: Boolean, resultRegister: RegisterOrPair?, scope: Subroutine?) {
         when(func.name) {
+            "fastrnd8" -> {
+                if(resultToStack)
+                    asmgen.out("  jsr  prog8_lib.func_fastrnd8_stack")
+                else {
+                    asmgen.out("  jsr  math.fast_randbyte")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
+            }
             "rnd" -> {
                 if(resultToStack)
                     asmgen.out("  jsr  prog8_lib.func_rnd_stack")
-                else
+                else {
                     asmgen.out("  jsr  math.randbyte")
+                    assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, scope, program, asmgen), CpuRegister.A)
+                }
             }
             "rndw" -> {
                 if(resultToStack)
                     asmgen.out("  jsr  prog8_lib.func_rndw_stack")
-                else
+                else {
                     asmgen.out("  jsr  math.randword")
+                    assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, scope, program, asmgen), RegisterOrPair.AY)
+                }
             }
             else -> throw AssemblyError("wrong func")
         }
     }
 
-    private fun funcMkword(fcall: IFunctionCall, resultToStack: Boolean) {
-        asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
-        asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
-        if(resultToStack)
-            asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
+    private fun funcPokeW(fcall: IFunctionCall) {
+        when(val addrExpr = fcall.args[0]) {
+            is NumericLiteralValue -> {
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)
+                val addr = addrExpr.number.toHex()
+                asmgen.out("  sta  $addr |  sty  ${addr}+1")
+                return
+            }
+            is IdentifierReference -> {
+                val varname = asmgen.asmVariableName(addrExpr)
+                if(asmgen.isZpVar(addrExpr)) {
+                    // pointervar is already in the zero page, no need to copy
+                    asmgen.saveRegisterLocal(CpuRegister.X, (fcall as Node).definingSubroutine()!!)
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AX)
+                    if (asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                        asmgen.out("""
+                            sta  ($varname)
+                            txa
+                            ldy  #1
+                            sta  ($varname),y""")
+                    } else {
+                        asmgen.out("""
+                            ldy  #0
+                            sta  ($varname),y
+                            txa
+                            iny
+                            sta  ($varname),y""")
+                    }
+                    asmgen.restoreRegisterLocal(CpuRegister.X)
+                    return
+                }
+            }
+            is BinaryExpression -> {
+                if(addrExpr.operator=="+" && addrExpr.left is IdentifierReference && addrExpr.right is NumericLiteralValue) {
+                    asmgen.saveRegisterLocal(CpuRegister.X, (fcall as Node).definingSubroutine()!!)
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AX)
+                    val varname = asmgen.asmVariableName(addrExpr.left as IdentifierReference)
+                    val index = (addrExpr.right as NumericLiteralValue).number.toHex()
+                    asmgen.out("""
+                        ldy  #$index
+                        sta  ($varname),y
+                        txa
+                        iny
+                        sta  ($varname),y""")
+                    asmgen.restoreRegisterLocal(CpuRegister.X)
+                    return
+                }
+            }
+        }
+
+        asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", DataType.UWORD, null)
+        asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)
+        asmgen.out("  jsr  prog8_lib.func_pokew")
     }
 
-    private fun funcMsb(fcall: IFunctionCall, resultToStack: Boolean) {
+    private fun funcPeekW(fcall: IFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+        when(val addrExpr = fcall.args[0]) {
+            is NumericLiteralValue -> {
+                val addr = addrExpr.number.toHex()
+                asmgen.out("  lda  $addr |  ldy  ${addr}+1")
+            }
+            is IdentifierReference -> {
+                val varname = asmgen.asmVariableName(addrExpr)
+                if(asmgen.isZpVar(addrExpr)) {
+                    // pointervar is already in the zero page, no need to copy
+                    if (asmgen.isTargetCpu(CpuType.CPU65c02)) {
+                        asmgen.out("""
+                            ldy  #1
+                            lda  ($varname),y
+                            tay
+                            lda  ($varname)""")
+                    } else {
+                        asmgen.out("""
+                            ldy  #0
+                            lda  ($varname),y
+                            pha
+                            iny
+                            lda  ($varname),y
+                            tay
+                            pla""")
+                    }
+                } else {
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+                    asmgen.out("  jsr  prog8_lib.func_peekw")
+                }
+            }
+            is BinaryExpression -> {
+                if(addrExpr.operator=="+" && addrExpr.left is IdentifierReference && addrExpr.right is NumericLiteralValue) {
+                    val varname = asmgen.asmVariableName(addrExpr.left as IdentifierReference)
+                    val index = (addrExpr.right as NumericLiteralValue).number.toHex()
+                    asmgen.out("""
+                        ldy  #$index
+                        lda  ($varname),y
+                        pha
+                        iny
+                        lda  ($varname),y
+                        tay
+                        pla""")
+                } else {
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+                    asmgen.out("  jsr  prog8_lib.func_peekw")
+                }
+            }
+            else -> {
+                asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+                asmgen.out("  jsr  prog8_lib.func_peekw")
+            }
+        }
+
+        if(resultToStack){
+            asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
+        } else {
+            when(resultRegister ?: RegisterOrPair.AY) {
+                RegisterOrPair.AY -> {}
+                RegisterOrPair.AX -> asmgen.out("  sty  P8ZP_SCRATCH_REG |  ldx  P8ZP_SCRATCH_REG")
+                RegisterOrPair.XY -> asmgen.out("  tax")
+                in Cx16VirtualRegisters -> asmgen.out("  sta  cx16.${resultRegister.toString().toLowerCase()} |  sty  cx16.${resultRegister.toString().toLowerCase()}+1")
+                else -> throw AssemblyError("invalid reg")
+            }
+        }
+    }
+
+    private fun funcMkword(fcall: IFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
+        if(resultToStack) {
+            asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
+            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+            asmgen.out("  sta  P8ESTACK_LO,x |  tya |  sta  P8ESTACK_HI,x |  dex")
+        } else {
+            val reg = resultRegister ?: RegisterOrPair.AY
+            var needAsave = !(fcall.args[0] is DirectMemoryRead || fcall.args[0] is NumericLiteralValue || fcall.args[0] is IdentifierReference)
+            if(!needAsave) {
+                val mr0 = fcall.args[0] as? DirectMemoryRead
+                val mr1 = fcall.args[1] as? DirectMemoryRead
+                if (mr0 != null)
+                    needAsave = mr0.addressExpression !is NumericLiteralValue && mr0.addressExpression !is IdentifierReference
+                if (mr1 != null)
+                    needAsave = needAsave or (mr1.addressExpression !is NumericLiteralValue && mr1.addressExpression !is IdentifierReference)
+            }
+            when(reg) {
+                RegisterOrPair.AX -> {
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                    if(needAsave)
+                        asmgen.out("  pha")
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.X)      // msb
+                    if(needAsave)
+                        asmgen.out("  pla")
+                }
+                RegisterOrPair.AY -> {
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                    if(needAsave)
+                        asmgen.out("  pha")
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
+                    if(needAsave)
+                        asmgen.out("  pla")
+                }
+                RegisterOrPair.XY -> {
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                    if(needAsave)
+                        asmgen.out("  pha")
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.Y)      // msb
+                    if(needAsave)
+                        asmgen.out("  pla")
+                    asmgen.out("  tax")
+                }
+                in Cx16VirtualRegisters -> {
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // lsb
+                    asmgen.out("  sta  cx16.${reg.toString().toLowerCase()}")
+                    asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.A)      // msb
+                    asmgen.out("  sta  cx16.${reg.toString().toLowerCase()}+1")
+                }
+                else -> throw AssemblyError("invalid mkword target reg")
+            }
+        }
+    }
+
+    private fun funcMsb(fcall: IFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
         val arg = fcall.args.single()
         if (arg.inferType(program).typeOrElse(DataType.STRUCT) !in WordDatatypes)
             throw AssemblyError("msb required word argument")
@@ -974,19 +1221,43 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
             throw AssemblyError("msb(const) should have been const-folded away")
         if (arg is IdentifierReference) {
             val sourceName = asmgen.asmVariableName(arg)
-            asmgen.out("  lda  $sourceName+1")
-            if (resultToStack)
-                asmgen.out("  sta  P8ESTACK_LO,x |  dex")
+            if(resultToStack) {
+                asmgen.out("  lda  $sourceName+1 |  sta  P8ESTACK_LO,x |  dex")
+            } else {
+                when(resultRegister) {
+                    null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName+1")
+                    RegisterOrPair.X -> asmgen.out("  ldx  $sourceName+1")
+                    RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName+1")
+                    else -> throw AssemblyError("invalid reg")
+                }
+            }
         } else {
-            asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-            if (resultToStack)
+            if(resultToStack) {
+                asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
                 asmgen.out("  tya |  sta  P8ESTACK_LO,x |  dex")
-            else
-                asmgen.out("  tya")
+            } else {
+                when(resultRegister) {
+                    null, RegisterOrPair.A -> {
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                        asmgen.out("  tya")
+                    }
+                    RegisterOrPair.X -> {
+                        asmgen.out("  pha")
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AX)
+                        asmgen.out("  pla")
+                    }
+                    RegisterOrPair.Y -> {
+                        asmgen.out("  pha")
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                        asmgen.out("  pla")
+                    }
+                    else -> throw AssemblyError("invalid reg")
+                }
+            }
         }
     }
 
-    private fun funcLsb(fcall: IFunctionCall, resultToStack: Boolean) {
+    private fun funcLsb(fcall: IFunctionCall, resultToStack: Boolean, resultRegister: RegisterOrPair?) {
         val arg = fcall.args.single()
         if (arg.inferType(program).typeOrElse(DataType.STRUCT) !in WordDatatypes)
             throw AssemblyError("lsb required word argument")
@@ -995,13 +1266,45 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
 
         if (arg is IdentifierReference) {
             val sourceName = asmgen.asmVariableName(arg)
-            asmgen.out("  lda  $sourceName")
-            if (resultToStack)
-                asmgen.out("  sta  P8ESTACK_LO,x |  dex")
+            if(resultToStack) {
+                asmgen.out("  lda  $sourceName |  sta  P8ESTACK_LO,x |  dex")
+            } else {
+                when(resultRegister) {
+                    null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName")
+                    RegisterOrPair.X -> asmgen.out("  ldx  $sourceName")
+                    RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName")
+                    else -> throw AssemblyError("invalid reg")
+                }
+            }
         } else {
-            asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
-            if (resultToStack)
+            if(resultToStack) {
+                asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
+                //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
+                //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
                 asmgen.out("  sta  P8ESTACK_LO,x |  dex")
+            } else {
+                when(resultRegister) {
+                    null, RegisterOrPair.A -> {
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                        // NOTE: we rely on the fact that the above assignment to AY, assigns the Lsb to A as the last instruction.
+                        //       this is required because the compiler assumes the status bits are set according to what A is (lsb)
+                        //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
+                    }
+                    RegisterOrPair.X -> {
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.XY)
+                        // NOTE: we rely on the fact that the above assignment to XY, assigns the Lsb to X as the last instruction.
+                        //       this is required because the compiler assumes the status bits are set according to what X is (lsb)
+                        //       and will not generate another cmp when lsb() is directly used inside a comparison expression.
+                    }
+                    RegisterOrPair.Y -> {
+                        asmgen.out("  pha")
+                        asmgen.assignExpressionToRegister(fcall.args.single(), RegisterOrPair.AY)
+                        asmgen.out("  tay |  pla |  cpy  #0")
+                    }
+                    else -> throw AssemblyError("invalid reg")
+                }
+            }
         }
     }
 
@@ -1009,7 +1312,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
         // address in P8ZP_SCRATCH_W1,  number of elements in A
         arg as IdentifierReference
         val identifierName = asmgen.asmVariableName(arg)
-        val size = arg.targetVarDecl(program.namespace)!!.arraysize!!.constIndex()!!
+        val size = arg.targetVarDecl(program)!!.arraysize!!.constIndex()!!
         asmgen.out("""
                     lda  #<$identifierName
                     ldy  #>$identifierName
@@ -1019,7 +1322,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                     """)
     }
 
-    private fun translateArguments(args: MutableList<Expression>, signature: FSignature, scope: Subroutine) {
+    private fun translateArguments(args: MutableList<Expression>, signature: FSignature, scope: Subroutine?) {
         val callConv = signature.callConvention(args.map { it.inferType(program).typeOrElse(DataType.STRUCT) })
 
         fun getSourceForFloat(value: Expression): AsmAssignSource {
@@ -1032,8 +1335,11 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                     throw AssemblyError("float literals should have been converted into autovar")
                 }
                 else -> {
-                    scope.asmGenInfo.usedFloatEvalResultVar = true
-                    val variable = IdentifierReference(listOf("_prog8_float_eval_result"), value.position)
+                    if(scope==null)
+                        throw AssemblyError("cannot use float arguments outside of a subroutine scope")
+
+                    scope.asmGenInfo.usedFloatEvalResultVar2 = true
+                    val variable = IdentifierReference(listOf(subroutineFloatEvalResultVar2), value.position)
                     val addr = AddressOf(variable, value.position)
                     addr.linkParents(value)
                     asmgen.assignExpressionToVariable(value, asmgen.asmVariableName(variable), DataType.FLOAT, scope)
@@ -1061,7 +1367,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                         }
                     }
                     val tgt = AsmAssignTarget(TargetStorageKind.VARIABLE, program, asmgen, conv.dt, null, variableAsmName = varname)
-                    val assign = AsmAssignment(src, tgt, false, value.position)
+                    val assign = AsmAssignment(src, tgt, false, program.memsizer, value.position)
                     asmgen.translateNormalAssignment(assign)
                 }
                 conv.reg != null -> {
@@ -1077,7 +1383,7 @@ internal class BuiltinFunctionsAsmGen(private val program: Program, private val
                         }
                     }
                     val tgt = AsmAssignTarget.fromRegisters(conv.reg, null, program, asmgen)
-                    val assign = AsmAssignment(src, tgt, false, value.position)
+                    val assign = AsmAssignment(src, tgt, false, program.memsizer, value.position)
                     asmgen.translateNormalAssignment(assign)
                 }
                 else -> throw AssemblyError("callconv")

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

@@ -1,4 +1,4 @@
-package prog8.compiler.target.c64.codegen
+package prog8.compiler.target.cpu6502.codegen
 
 import prog8.ast.Program
 import prog8.ast.base.ArrayElementTypes
@@ -7,8 +7,8 @@ import prog8.ast.base.RegisterOrPair
 import prog8.ast.expressions.IdentifierReference
 import prog8.ast.expressions.RangeExpr
 import prog8.ast.statements.ForLoop
+import prog8.ast.toHex
 import prog8.compiler.AssemblyError
-import prog8.compiler.toHex
 import kotlin.math.absoluteValue
 
 internal class ForLoopsAsmGen(private val program: Program, private val asmgen: AsmGen) {
@@ -57,9 +57,9 @@ internal class ForLoopsAsmGen(private val program: Program, private val asmgen:
                         lda  $varname
 $modifiedLabel          cmp  #0         ; modified 
                         beq  $endLabel
-                        $incdec  $varname
-                        jmp  $loopLabel
-$endLabel""")
+                        $incdec  $varname""")
+                    asmgen.jmp(loopLabel)
+                    asmgen.out(endLabel)
 
                 } else {
 
@@ -117,16 +117,15 @@ $modifiedLabel2 cmp  #0    ; modified
                             asmgen.out("""
 +               inc  $varname
                 bne  $loopLabel
-                inc  $varname+1
-                jmp  $loopLabel
-                            """)
+                inc  $varname+1""")
+                            asmgen.jmp(loopLabel)
                         } else {
                             asmgen.out("""
 +               lda  $varname
                 bne  +
                 dec  $varname+1
-+               dec  $varname
-                jmp  $loopLabel""")
++               dec  $varname""")
+                            asmgen.jmp(loopLabel)
                         }
                         asmgen.out(endLabel)
                     }
@@ -239,7 +238,7 @@ $endLabel""")
         val endLabel = asmgen.makeLabel("for_end")
         asmgen.loopEndLabels.push(endLabel)
         val iterableName = asmgen.asmVariableName(ident)
-        val decl = ident.targetVarDecl(program.namespace)!!
+        val decl = ident.targetVarDecl(program)!!
         when(iterableDt) {
             DataType.STR -> {
                 asmgen.out("""
@@ -369,7 +368,7 @@ $loopLabel""")
                         throw AssemblyError("step 0, 1 and -1 should have been handled specifically  $stmt")
                     }
                     2 -> {
-                        if(range.last==255) {
+                        if(range.last==255 || range.last==254) {
                             asmgen.out("""
                                 inc  $varname
                                 beq  $endLabel
@@ -377,34 +376,34 @@ $loopLabel""")
                                 bne  $loopLabel""")
                         } else {
                             asmgen.out("""
+                                inc  $varname
+                                inc  $varname
                                 lda  $varname
-                                cmp  #${range.last}
-                                beq  $endLabel
-                                inc  $varname
-                                inc  $varname
-                                jmp  $loopLabel""")
+                                cmp  #${range.last+2}
+                                bne  $loopLabel""")
                         }
                     }
                     -2 -> {
                         when (range.last) {
-                            0 -> asmgen.out("""
-                                lda  $varname
-                                beq  $endLabel
-                                dec  $varname
-                                dec  $varname
-                                jmp  $loopLabel""")
+                            0 -> {
+                                asmgen.out("""
+                                    lda  $varname
+                                    beq  $endLabel
+                                    dec  $varname
+                                    dec  $varname""")
+                                asmgen.jmp(loopLabel)
+                            }
                             1 -> asmgen.out("""
-                                dec  $varname
-                                beq  $endLabel
-                                dec  $varname
-                                bne  $loopLabel""")
+                                    dec  $varname
+                                    beq  $endLabel
+                                    dec  $varname
+                                    bne  $loopLabel""")
                             else -> asmgen.out("""
-                                lda  $varname
-                                cmp  #${range.last}
-                                beq  $endLabel
-                                dec  $varname
-                                dec  $varname
-                                jmp  $loopLabel""")
+                                    dec  $varname
+                                    dec  $varname
+                                    lda  $varname
+                                    cmp  #${range.last-2}
+                                    bne  $loopLabel""")
                         }
                     }
                     else -> {
@@ -415,8 +414,8 @@ $loopLabel""")
                             beq  $endLabel
                             clc
                             adc  #${range.step}
-                            sta  $varname
-                            jmp  $loopLabel""")
+                            sta  $varname""")
+                        asmgen.jmp(loopLabel)
                     }
                 }
                 asmgen.out(endLabel)
@@ -452,9 +451,9 @@ $loopLabel""")
                             sta  $varname
                             lda  $varname+1
                             adc  #>${range.step}
-                            sta  $varname+1
-                            jmp  $loopLabel
-$endLabel""")
+                            sta  $varname+1""")
+                        asmgen.jmp(loopLabel)
+                        asmgen.out(endLabel)
                     }
                 }
             }
@@ -480,11 +479,10 @@ $loopLabel""")
 $endLabel""")
         } else {
             asmgen.out("""
-                lda  $varname
-                cmp  #${range.last}
-                beq  $endLabel
                 inc  $varname
-                jmp  $loopLabel
+                lda  $varname
+                cmp  #${range.last+1}
+                bne  $loopLabel
 $endLabel""")
         }
         asmgen.loopEndLabels.pop()
@@ -505,23 +503,22 @@ $loopLabel""")
                 asmgen.out("""
                     lda  $varname
                     beq  $endLabel
-                    dec  $varname
-                    jmp  $loopLabel
-$endLabel""")
+                    dec  $varname""")
+                asmgen.jmp(loopLabel)
+                asmgen.out(endLabel)
             }
             1 -> {
                 asmgen.out("""
                     dec  $varname
-                    jmp  $loopLabel
+                    bne  $loopLabel
 $endLabel""")
             }
             else -> {
                 asmgen.out("""
-                    lda  $varname
-                    cmp  #${range.last}
-                    beq  $endLabel
                     dec  $varname
-                    jmp  $loopLabel
+                    lda  $varname
+                    cmp  #${range.last-1}
+                    bne  $loopLabel
 $endLabel""")
             }
         }
@@ -546,13 +543,12 @@ $loopLabel""")
             bne  +
             lda  $varname+1
             cmp  #>${range.last}
-            bne  +
             beq  $endLabel
 +           inc  $varname
             bne  $loopLabel
-            inc  $varname+1
-            jmp  $loopLabel
-$endLabel""")
+            inc  $varname+1""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
         asmgen.loopEndLabels.pop()
     }
 
@@ -574,14 +570,13 @@ $loopLabel""")
             bne  +
             lda  $varname+1
             cmp  #>${range.last}
-            bne  +
             beq  $endLabel
 +           lda  $varname
             bne  +
             dec  $varname+1
-+           dec  $varname
-            jmp  $loopLabel
-$endLabel""")
++           dec  $varname""")
+        asmgen.jmp(loopLabel)
+        asmgen.out(endLabel)
         asmgen.loopEndLabels.pop()
     }
 

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

@@ -0,0 +1,347 @@
+package prog8.compiler.target.cpu6502.codegen
+
+import prog8.ast.IFunctionCall
+import prog8.ast.Node
+import prog8.ast.Program
+import prog8.ast.base.*
+import prog8.ast.expressions.*
+import prog8.ast.statements.*
+import prog8.compiler.AssemblyError
+import prog8.compiler.target.CpuType
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignSource
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignTarget
+import prog8.compiler.target.cpu6502.codegen.assignment.AsmAssignment
+import prog8.compiler.target.cpu6502.codegen.assignment.TargetStorageKind
+
+
+internal class FunctionCallAsmGen(private val program: Program, private val asmgen: AsmGen) {
+
+    internal fun translateFunctionCallStatement(stmt: IFunctionCall) {
+        saveXbeforeCall(stmt)
+        translateFunctionCall(stmt)
+        restoreXafterCall(stmt)
+        // just ignore any result values from the function call.
+    }
+
+    internal fun saveXbeforeCall(stmt: IFunctionCall) {
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        if(sub.shouldSaveX()) {
+            val regSaveOnStack = sub.asmAddress==null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
+            val (keepAonEntry: Boolean, keepAonReturn: Boolean) = sub.shouldKeepA()
+            if(regSaveOnStack)
+                asmgen.saveRegisterStack(CpuRegister.X, keepAonEntry)
+            else
+                asmgen.saveRegisterLocal(CpuRegister.X, (stmt as Node).definingSubroutine()!!)
+        }
+    }
+
+    internal fun restoreXafterCall(stmt: IFunctionCall) {
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        if(sub.shouldSaveX()) {
+            val regSaveOnStack = sub.asmAddress==null       // rom-routines don't require registers to be saved on stack, normal subroutines do because they can contain nested calls
+            val (keepAonEntry: Boolean, keepAonReturn: Boolean) = sub.shouldKeepA()
+
+            if(regSaveOnStack)
+                asmgen.restoreRegisterStack(CpuRegister.X, keepAonReturn)
+            else
+                asmgen.restoreRegisterLocal(CpuRegister.X)
+        }
+    }
+
+    internal fun translateFunctionCall(stmt: IFunctionCall) {
+        // Output only the code to setup the parameters and perform the actual call
+        // NOTE: does NOT output the code to deal with the result values!
+        // NOTE: does NOT output code to save/restore the X register for this call! Every caller should deal with this in their own way!!
+        //       (you can use subroutine.shouldSaveX() and saveX()/restoreX() routines as a help for this)
+
+        val sub = stmt.target.targetSubroutine(program) ?: throw AssemblyError("undefined subroutine ${stmt.target}")
+        val subName = asmgen.asmSymbolName(stmt.target)
+        if(stmt.args.isNotEmpty()) {
+
+            if(sub.asmParameterRegisters.isEmpty()) {
+                // via variables
+                for(arg in sub.parameters.withIndex().zip(stmt.args)) {
+                    argumentViaVariable(sub, arg.first, arg.second)
+                }
+            } else {
+                // via registers
+                if(sub.parameters.size==1) {
+                    // just a single parameter, no risk of clobbering registers
+                    argumentViaRegister(sub, IndexedValue(0, sub.parameters.single()), stmt.args[0])
+                } else {
+
+                    fun isNoClobberRisk(expr: Expression): Boolean {
+                        if(expr is AddressOf ||
+                                expr is NumericLiteralValue ||
+                                expr is StringLiteralValue ||
+                                expr is ArrayLiteralValue ||
+                                expr is IdentifierReference)
+                            return true
+
+                        if(expr is FunctionCall) {
+                            if(expr.target.nameInSource==listOf("lsb") || expr.target.nameInSource==listOf("msb"))
+                                return isNoClobberRisk(expr.args[0])
+                            if(expr.target.nameInSource==listOf("mkword"))
+                                return isNoClobberRisk(expr.args[0]) && isNoClobberRisk(expr.args[1])
+                        }
+
+                        return false
+                    }
+
+                    when {
+                        stmt.args.all {isNoClobberRisk(it)} -> {
+                            // There's no risk of clobbering for these simple argument types. Optimize the register loading directly from these values.
+                            // register assignment order: 1) cx16 virtual word registers, 2) actual CPU registers, 3) CPU Carry status flag.
+                            val argsInfo = sub.parameters.withIndex().zip(stmt.args).zip(sub.asmParameterRegisters)
+                            val (cx16virtualRegs, args2) = argsInfo.partition { it.second.registerOrPair in Cx16VirtualRegisters }
+                            val (cpuRegs, statusRegs) = args2.partition { it.second.registerOrPair!=null }
+                            for(arg in cx16virtualRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                            for(arg in cpuRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                            for(arg in statusRegs)
+                                argumentViaRegister(sub, arg.first.first, arg.first.second)
+                        }
+                        else -> {
+                            // Risk of clobbering due to complex expression args. Evaluate first, then assign registers.
+                            registerArgsViaStackEvaluation(stmt, sub)
+                        }
+                    }
+                }
+            }
+        }
+
+        if(sub.inline && asmgen.options.optimize) {
+            if(sub.containsDefinedVariables())
+                throw AssemblyError("can't inline sub with vars")
+            if(!sub.isAsmSubroutine && sub.parameters.isNotEmpty())
+                throw AssemblyError("can't inline a non-asm subroutine with parameters")
+            asmgen.out("  \t; inlined routine follows: ${sub.name} from ${sub.position}")
+            val statements = sub.statements.filter { it !is ParameterVarDecl && it !is Directive }
+            statements.forEach { asmgen.translate(it) }
+        }
+        else {
+            asmgen.out("  jsr  $subName")
+        }
+
+        // remember: dealing with the X register and/or dealing with return values is the responsibility of the caller
+    }
+
+    private fun registerArgsViaStackEvaluation(stmt: IFunctionCall, sub: Subroutine) {
+        // this is called when one or more of the arguments are 'complex' and
+        // cannot be assigned to a register easily or risk clobbering other registers.
+
+        if(sub.parameters.isEmpty())
+            return
+
+        // 1. load all arguments reversed onto the stack: first arg goes last (is on top).
+        for (arg in stmt.args.reversed())
+            asmgen.translateExpression(arg)
+
+        var argForCarry: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForXregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+        var argForAregister: IndexedValue<Pair<Expression, RegisterOrStatusflag>>? = null
+
+        asmgen.out("  inx")     // align estack pointer
+
+        for(argi in stmt.args.zip(sub.asmParameterRegisters).withIndex()) {
+            val plusIdxStr = if(argi.index==0) "" else "+${argi.index}"
+            when {
+                argi.value.second.statusflag == Statusflag.Pc -> {
+                    require(argForCarry == null)
+                    argForCarry = argi
+                }
+                argi.value.second.statusflag != null -> throw AssemblyError("can only use Carry as status flag parameter")
+                argi.value.second.registerOrPair in setOf(RegisterOrPair.X, RegisterOrPair.AX, RegisterOrPair.XY) -> {
+                    require(argForXregister==null)
+                    argForXregister = argi
+                }
+                argi.value.second.registerOrPair in setOf(RegisterOrPair.A, RegisterOrPair.AY) -> {
+                    require(argForAregister == null)
+                    argForAregister = argi
+                }
+                argi.value.second.registerOrPair == RegisterOrPair.Y -> {
+                    asmgen.out("  ldy  P8ESTACK_LO$plusIdxStr,x")
+                }
+                argi.value.second.registerOrPair in Cx16VirtualRegisters -> {
+                    // immediately output code to load the virtual register, to avoid clobbering the A register later
+                    when (sub.parameters[argi.index].type) {
+                        in ByteDatatypes -> {
+                            // only load the lsb of the virtual register
+                            asmgen.out("""
+                                lda  P8ESTACK_LO$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}
+                            """)
+                            if (asmgen.isTargetCpu(CpuType.CPU65c02))
+                                asmgen.out("  stz  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1")
+                            else
+                                asmgen.out("  lda  #0 |  sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1")
+                        }
+                        in WordDatatypes, in IterableDatatypes ->
+                            asmgen.out("""
+                                lda  P8ESTACK_LO$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}
+                                lda  P8ESTACK_HI$plusIdxStr,x
+                                sta  cx16.${argi.value.second.registerOrPair.toString().toLowerCase()}+1
+                            """)
+                        else -> throw AssemblyError("weird dt")
+                    }
+                }
+                else -> throw AssemblyError("weird argument")
+            }
+        }
+
+        if(argForCarry!=null) {
+            val plusIdxStr = if(argForCarry.index==0) "" else "+${argForCarry.index}"
+            asmgen.out("""
+                lda  P8ESTACK_LO$plusIdxStr,x
+                beq  +
+                sec
+                bcs  ++
++               clc
++               php""")             // push the status flags
+        }
+
+        if(argForAregister!=null) {
+            val plusIdxStr = if(argForAregister.index==0) "" else "+${argForAregister.index}"
+            when(argForAregister.value.second.registerOrPair) {
+                RegisterOrPair.A -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x")
+                RegisterOrPair.AY -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x |  ldy  P8ESTACK_HI$plusIdxStr,x")
+                else -> throw AssemblyError("weird arg")
+            }
+        }
+
+        if(argForXregister!=null) {
+            val plusIdxStr = if(argForXregister.index==0) "" else "+${argForXregister.index}"
+
+            if(argForAregister!=null)
+                asmgen.out("  pha")
+            when(argForXregister.value.second.registerOrPair) {
+                RegisterOrPair.X -> asmgen.out("  lda  P8ESTACK_LO$plusIdxStr,x |  tax")
+                RegisterOrPair.AX -> asmgen.out("  ldy  P8ESTACK_LO$plusIdxStr,x |  lda  P8ESTACK_HI$plusIdxStr,x |  tax |  tya")
+                RegisterOrPair.XY -> asmgen.out("  ldy  P8ESTACK_HI$plusIdxStr,x |  lda  P8ESTACK_LO$plusIdxStr,x |  tax")
+                else -> throw AssemblyError("weird arg")
+            }
+            if(argForAregister!=null)
+                asmgen.out("  pla")
+        } else {
+            repeat(sub.parameters.size - 1) { asmgen.out("  inx") }       // unwind stack
+        }
+
+        if(argForCarry!=null)
+            asmgen.out("  plp")       // set the carry flag back to correct value
+    }
+
+    private fun argumentViaVariable(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
+        // pass parameter via a regular variable (not via registers)
+        val valueIDt = value.inferType(program)
+        if(!valueIDt.isKnown)
+            throw AssemblyError("unknown dt")
+        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
+        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
+            throw AssemblyError("argument type incompatible")
+
+        val varName = asmgen.asmVariableName(sub.scopedname+"."+parameter.value.name)
+        asmgen.assignExpressionToVariable(value, varName, parameter.value.type, sub)
+    }
+
+    private fun argumentViaRegister(sub: Subroutine, parameter: IndexedValue<SubroutineParameter>, value: Expression) {
+        // pass argument via a register parameter
+        val valueIDt = value.inferType(program)
+        if(!valueIDt.isKnown)
+            throw AssemblyError("unknown dt")
+        val valueDt = valueIDt.typeOrElse(DataType.STRUCT)
+        if(!isArgumentTypeCompatible(valueDt, parameter.value.type))
+            throw AssemblyError("argument type incompatible")
+
+        val paramRegister = sub.asmParameterRegisters[parameter.index]
+        val statusflag = paramRegister.statusflag
+        val register = paramRegister.registerOrPair
+        val requiredDt = parameter.value.type
+        if(requiredDt!=valueDt) {
+            if(valueDt largerThan requiredDt)
+                throw AssemblyError("can only convert byte values to word param types")
+        }
+        if (statusflag!=null) {
+            if(requiredDt!=valueDt)
+                throw AssemblyError("for statusflag, byte value is required")
+            if (statusflag == Statusflag.Pc) {
+                // this param needs to be set last, right before the jsr
+                // for now, this is already enforced on the subroutine definition by the Ast Checker
+                when(value) {
+                    is NumericLiteralValue -> {
+                        val carrySet = value.number.toInt() != 0
+                        asmgen.out(if(carrySet) "  sec" else "  clc")
+                    }
+                    is IdentifierReference -> {
+                        val sourceName = asmgen.asmVariableName(value)
+                        asmgen.out("""
+                pha
+                lda  $sourceName
+                beq  +
+                sec  
+                bcs  ++
+    +           clc
+    +           pla
+    """)
+                    }
+                    else -> {
+                        asmgen.assignExpressionToRegister(value, RegisterOrPair.A)
+                        asmgen.out("""
+                                    beq  +
+                                    sec
+                                    bcs  ++
+    +                               clc
+    +""")
+                    }
+                }
+            } else throw AssemblyError("can only use Carry as status flag parameter")
+        }
+        else {
+            // via register or register pair
+            register!!
+            if(requiredDt largerThan valueDt) {
+                // we need to sign extend the source, do this via temporary word variable
+                val scratchVar = asmgen.asmVariableName("P8ZP_SCRATCH_W1")
+                asmgen.assignExpressionToVariable(value, scratchVar, DataType.UBYTE, sub)
+                asmgen.signExtendVariableLsb(scratchVar, valueDt)
+                asmgen.assignVariableToRegister(scratchVar, register)
+            } else {
+                val target: AsmAssignTarget =
+                    if(parameter.value.type in ByteDatatypes && (register==RegisterOrPair.AX || register == RegisterOrPair.AY || register==RegisterOrPair.XY || register in Cx16VirtualRegisters))
+                        AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, parameter.value.type, sub, register = register)
+                    else
+                        AsmAssignTarget.fromRegisters(register, sub, program, asmgen)
+                val src = if(valueDt in PassByReferenceDatatypes) {
+                    if(value is IdentifierReference) {
+                        val addr = AddressOf(value, Position.DUMMY)
+                        AsmAssignSource.fromAstSource(addr, program, asmgen).adjustSignedUnsigned(target)
+                    } else {
+                        AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
+                    }
+                } else {
+                    AsmAssignSource.fromAstSource(value, program, asmgen).adjustSignedUnsigned(target)
+                }
+                asmgen.translateNormalAssignment(AsmAssignment(src, target, false, program.memsizer, Position.DUMMY))
+            }
+        }
+    }
+
+    private fun isArgumentTypeCompatible(argType: DataType, paramType: DataType): Boolean {
+        if(argType isAssignableTo paramType)
+            return true
+        if(argType in ByteDatatypes && paramType in ByteDatatypes)
+            return true
+        if(argType in WordDatatypes && paramType in WordDatatypes)
+            return true
+
+        // we have a special rule for some types.
+        // strings are assignable to UWORD, for example, and vice versa
+        if(argType==DataType.STR && paramType==DataType.UWORD)
+            return true
+        if(argType==DataType.UWORD && paramType == DataType.STR)
+            return true
+
+        return false
+    }
+}

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

@@ -1,12 +1,12 @@
-package prog8.compiler.target.c64.codegen
+package prog8.compiler.target.cpu6502.codegen
 
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.IdentifierReference
 import prog8.ast.expressions.NumericLiteralValue
 import prog8.ast.statements.PostIncrDecr
+import prog8.ast.toHex
 import prog8.compiler.AssemblyError
-import prog8.compiler.toHex
 
 
 internal class PostIncrDecrAsmGen(private val program: Program, private val asmgen: AsmGen) {
@@ -66,8 +66,9 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
             targetArrayIdx!=null -> {
                 val asmArrayvarname = asmgen.asmVariableName(targetArrayIdx.arrayvar)
                 val elementDt = targetArrayIdx.inferType(program).typeOrElse(DataType.STRUCT)
-                if(targetArrayIdx.indexer.indexNum!=null) {
-                    val indexValue = targetArrayIdx.indexer.constIndex()!! * elementDt.memorySize()
+                val constIndex = targetArrayIdx.indexer.constIndex()
+                if(constIndex!=null) {
+                    val indexValue = constIndex * program.memsizer.memorySize(elementDt)
                     when(elementDt) {
                         in ByteDatatypes -> asmgen.out(if (incr) "  inc  $asmArrayvarname+$indexValue" else "  dec  $asmArrayvarname+$indexValue")
                         in WordDatatypes -> {
@@ -91,7 +92,7 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
                 else
                 {
                     asmgen.loadScaledArrayIndexIntoRegister(targetArrayIdx, elementDt, CpuRegister.A)
-                    asmgen.saveRegister(CpuRegister.X, false, scope!!)
+                    asmgen.saveRegisterLocal(CpuRegister.X, scope!!)
                     asmgen.out("  tax")
                     when(elementDt) {
                         in ByteDatatypes -> {
@@ -105,7 +106,7 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
     lda  $asmArrayvarname,x
     bne  +
     dec  $asmArrayvarname+1,x
-+       dec  $asmArrayvarname 
++       dec  $asmArrayvarname,x 
 """)
                         }
                         DataType.FLOAT -> {
@@ -119,7 +120,7 @@ internal class PostIncrDecrAsmGen(private val program: Program, private val asmg
                         }
                         else -> throw AssemblyError("weird array elt dt")
                     }
-                    asmgen.restoreRegister(CpuRegister.X, false)
+                    asmgen.restoreRegisterLocal(CpuRegister.X)
                 }
             }
         }

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

@@ -1,11 +1,12 @@
-package prog8.compiler.target.c64.codegen.assignment
+package prog8.compiler.target.cpu6502.codegen.assignment
 
+import prog8.ast.IMemSizer
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
 import prog8.ast.statements.*
 import prog8.compiler.AssemblyError
-import prog8.compiler.target.c64.codegen.AsmGen
+import prog8.compiler.target.cpu6502.codegen.AsmGen
 
 
 internal enum class TargetStorageKind {
@@ -49,8 +50,8 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
     lateinit var origAssign: AsmAssignment
 
     init {
-        if(register!=null && datatype !in IntegerDatatypes)
-            throw AssemblyError("register must be integer type")
+        if(register!=null && datatype !in NumericDatatypes)
+            throw AssemblyError("register must be integer or float type")
     }
 
     companion object {
@@ -75,6 +76,24 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
                     RegisterOrPair.AX,
                     RegisterOrPair.AY,
                     RegisterOrPair.XY -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, DataType.UWORD, scope, register = registers)
+                    RegisterOrPair.FAC1,
+                    RegisterOrPair.FAC2 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, DataType.FLOAT, scope, register = registers)
+                    RegisterOrPair.R0,
+                    RegisterOrPair.R1,
+                    RegisterOrPair.R2,
+                    RegisterOrPair.R3,
+                    RegisterOrPair.R4,
+                    RegisterOrPair.R5,
+                    RegisterOrPair.R6,
+                    RegisterOrPair.R7,
+                    RegisterOrPair.R8,
+                    RegisterOrPair.R9,
+                    RegisterOrPair.R10,
+                    RegisterOrPair.R11,
+                    RegisterOrPair.R12,
+                    RegisterOrPair.R13,
+                    RegisterOrPair.R14,
+                    RegisterOrPair.R15 -> AsmAssignTarget(TargetStorageKind.REGISTER, program, asmgen, DataType.UWORD, scope, register = registers)
                 }
     }
 }
@@ -86,7 +105,7 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                                private val variableAsmName: String? = null,
                                val array: ArrayIndexedExpression? = null,
                                val memory: DirectMemoryRead? = null,
-                               val register: CpuRegister? = null,
+                               val register: RegisterOrPair? = null,
                                val number: NumericLiteralValue? = null,
                                val expression: Expression? = null
 )
@@ -101,13 +120,7 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
             asmgen.asmVariableName(array.arrayvar)
 
     companion object {
-        fun fromAstSource(indexer: ArrayIndex, program: Program, asmgen: AsmGen): AsmAssignSource {
-            return when {
-                indexer.indexNum!=null -> fromAstSource(indexer.indexNum!!, program, asmgen)
-                indexer.indexVar!=null -> fromAstSource(indexer.indexVar!!, program, asmgen)
-                else -> throw AssemblyError("weird indexer")
-            }
-        }
+        fun fromAstSource(indexer: ArrayIndex, program: Program, asmgen: AsmGen): AsmAssignSource = fromAstSource(indexer.indexExpr, program, asmgen)
 
         fun fromAstSource(value: Expression, program: Program, asmgen: AsmGen): AsmAssignSource {
             val cv = value.constValue(program)
@@ -120,7 +133,15 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                 is ArrayLiteralValue -> throw AssemblyError("array literal value should not occur anymore for asm generation")
                 is IdentifierReference -> {
                     val dt = value.inferType(program).typeOrElse(DataType.STRUCT)
-                    AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, dt, variableAsmName = asmgen.asmVariableName(value))
+                    val varName=asmgen.asmVariableName(value)
+                    // special case: "cx16.r[0-15]" are 16-bits virtual registers of the commander X16 system
+                    if(dt==DataType.UWORD && varName.toLowerCase().startsWith("cx16.r")) {
+                        val regStr = varName.toLowerCase().substring(5)
+                        val reg = RegisterOrPair.valueOf(regStr.toUpperCase())
+                        AsmAssignSource(SourceStorageKind.REGISTER, program, asmgen, dt, register = reg)
+                    } else {
+                        AsmAssignSource(SourceStorageKind.VARIABLE, program, asmgen, dt, variableAsmName = varName)
+                    }
                 }
                 is DirectMemoryRead -> {
                     AsmAssignSource(SourceStorageKind.MEMORY, program, asmgen, DataType.UBYTE, memory = value)
@@ -130,9 +151,9 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
                     AsmAssignSource(SourceStorageKind.ARRAY, program, asmgen, dt, array = value)
                 }
                 is FunctionCall -> {
-                    when (val sub = value.target.targetStatement(program.namespace)) {
+                    when (val sub = value.target.targetStatement(program)) {
                         is Subroutine -> {
-                            val returnType = sub.returntypes.zip(sub.asmReturnvaluesRegisters).firstOrNull { rr -> rr.second.registerOrPair != null }?.first
+                            val returnType = sub.returntypes.zip(sub.asmReturnvaluesRegisters).firstOrNull { rr -> rr.second.registerOrPair != null || rr.second.statusflag!=null }?.first
                                     ?: throw AssemblyError("can't translate zero return values in assignment")
 
                             AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, returnType, expression = value)
@@ -180,12 +201,13 @@ internal class AsmAssignSource(val kind: SourceStorageKind,
 internal class AsmAssignment(val source: AsmAssignSource,
                              val target: AsmAssignTarget,
                              val isAugmentable: Boolean,
+                             memsizer: IMemSizer,
                              val position: Position) {
 
     init {
         if(target.register !in setOf(RegisterOrPair.XY, RegisterOrPair.AX, RegisterOrPair.AY))
             require(source.datatype != DataType.STRUCT) { "must not be placeholder datatype" }
-            require(source.datatype.memorySize() <= target.datatype.memorySize()) {
+            require(memsizer.memorySize(source.datatype) <= memsizer.memorySize(target.datatype)) {
                 "source storage size must be less or equal to target datatype storage size"
             }
     }

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

@@ -1,11 +1,9 @@
 package prog8.compiler.target.cx16
 
-import prog8.ast.Program
 import prog8.compiler.*
 import prog8.compiler.target.CpuType
 import prog8.compiler.target.IMachineDefinition
 import prog8.compiler.target.c64.C64MachineDefinition
-import prog8.parser.ModuleImporter
 import java.io.IOException
 
 internal object CX16MachineDefinition: IMachineDefinition {
@@ -21,25 +19,23 @@ internal object CX16MachineDefinition: IMachineDefinition {
     override val RAW_LOAD_ADDRESS = 0x8000
 
     // the 2*256 byte evaluation stack (on which bytes, words, and even floats are stored during calculations)
-    // and some heavily used string constants derived from the two values above
     override val ESTACK_LO = 0x0400        //  $0400-$04ff inclusive
     override val ESTACK_HI = 0x0500        //  $0500-$05ff inclusive
 
     override lateinit var zeropage: Zeropage
 
     override fun getFloat(num: Number) = C64MachineDefinition.Mflpt5.fromNumber(num)
-
-    override fun getFloatRomConst(number: Double): String? = null       // Cx16 has no pulblic ROM float locations
-    override fun importLibs(compilerOptions: CompilationOptions, importer: ModuleImporter, program: Program) {
-        if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
-            importer.importLibraryModule(program, "syslib")
+    override fun importLibs(compilerOptions: CompilationOptions, compilationTargetName: String): List<String> {
+        return if (compilerOptions.launcher == LauncherType.BASIC || compilerOptions.output == OutputType.PRG)
+            listOf("syslib")
+        else
+            emptyList()
     }
 
     override fun launchEmulator(programName: String) {
         for(emulator in listOf("x16emu")) {
             println("\nStarting Commander X16 emulator $emulator...")
-            val cmdline = listOf(emulator, "-rom", "/usr/share/x16-rom/rom.bin", "-scale", "2",
-                    "-run", "-prg", programName + ".prg")
+            val cmdline = listOf(emulator, "-scale", "2", "-run", "-prg", "$programName.prg")
             val processb = ProcessBuilder(cmdline).inheritIO()
             val process: Process
             try {
@@ -74,8 +70,8 @@ internal object CX16MachineDefinition: IMachineDefinition {
 
     internal class CX16Zeropage(options: CompilationOptions) : Zeropage(options) {
 
-        override val SCRATCH_B1 = 0x79      // temp storage for a single byte
-        override val SCRATCH_REG = 0x7a     // temp storage for a register, must be B1+1
+        override val SCRATCH_B1 = 0x7a      // temp storage for a single byte
+        override val SCRATCH_REG = 0x7b     // temp storage for a register, must be B1+1
         override val SCRATCH_W1 = 0x7c      // temp storage 1 for a word  $7c+$7d
         override val SCRATCH_W2 = 0x7e      // temp storage 2 for a word  $7e+$7f
 

compiler/src/prog8/optimizer/BinExprSplitter.kt

@@ -1,16 +1,16 @@
 package prog8.optimizer
 
-import prog8.ast.INameScope
 import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
 import prog8.ast.statements.AssignTarget
 import prog8.ast.statements.Assignment
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.target.ICompilationTarget
 
 
-internal class BinExprSplitter(private val program: Program) : AstWalker() {
+internal class BinExprSplitter(private val program: Program, private val compTarget: ICompilationTarget) : AstWalker() {
     private val noModifications = emptyList<IAstModification>()
 
 //    override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
@@ -53,7 +53,7 @@ X =      BinExpr                                    X   =   LeftExpr
 
 
  */
-            if(binExpr.operator in augmentAssignmentOperators && isSimpleTarget(assignment.target, program.namespace)) {
+            if(binExpr.operator in augmentAssignmentOperators && isSimpleTarget(assignment.target, program)) {
                 if(assignment.target isSameAs binExpr.left || assignment.target isSameAs binExpr.right)
                     return noModifications
 
@@ -78,9 +78,9 @@ X =      BinExpr                                    X   =   LeftExpr
     private fun isSimpleExpression(expr: Expression) =
             expr is IdentifierReference || expr is NumericLiteralValue || expr is AddressOf || expr is DirectMemoryRead || expr is StringLiteralValue || expr is ArrayLiteralValue || expr is RangeExpr
 
-    private fun isSimpleTarget(target: AssignTarget, namespace: INameScope) =
+    private fun isSimpleTarget(target: AssignTarget, program: Program) =
             if (target.identifier!=null || target.memoryAddress!=null)
-                target.isInRegularRAM(namespace)
+                compTarget.isInRegularRAM(target, program)
             else
                 false
 

compiler/src/prog8/optimizer/CallGraph.kt

@@ -1,26 +1,29 @@
 package prog8.optimizer
 
-import prog8.ast.*
+import prog8.ast.INameScope
+import prog8.ast.Module
+import prog8.ast.Node
+import prog8.ast.Program
 import prog8.ast.base.DataType
-import prog8.ast.base.ErrorReporter
 import prog8.ast.base.ParentSentinel
 import prog8.ast.base.Position
+import prog8.ast.expressions.AddressOf
 import prog8.ast.expressions.FunctionCall
 import prog8.ast.expressions.IdentifierReference
-import prog8.ast.processing.IAstVisitor
 import prog8.ast.statements.*
-import prog8.compiler.loadAsmIncludeFile
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.IErrorReporter
+import java.nio.file.Path
 
 private val alwaysKeepSubroutines = setOf(
-        Pair("main", "start"),
-        Pair("irq", "irq")
+        Pair("main", "start")
 )
 
-private val asmJumpRx = Regex("""[\-+a-zA-Z0-9_ \t]+(jmp|jsr)[ \t]+(\S+).*""", RegexOption.IGNORE_CASE)
+private val asmJumpRx = Regex("""[\-+a-zA-Z0-9_ \t]+(jmp|jsr|bra)[ \t]+(\S+).*""", RegexOption.IGNORE_CASE)
 private val asmRefRx = Regex("""[\-+a-zA-Z0-9_ \t]+(...)[ \t]+(\S+).*""", RegexOption.IGNORE_CASE)
 
 
-class CallGraph(private val program: Program) : IAstVisitor {
+class CallGraph(private val program: Program, private val asmFileLoader: (filename: String, source: Path)->String) : IAstVisitor {
 
     val imports = mutableMapOf<Module, List<Module>>().withDefault { mutableListOf() }
     val importedBy = mutableMapOf<Module, List<Module>>().withDefault { mutableListOf() }
@@ -77,7 +80,7 @@ class CallGraph(private val program: Program) : IAstVisitor {
             imports[thisModule] = imports.getValue(thisModule).plus(importedModule)
             importedBy[importedModule] = importedBy.getValue(importedModule).plus(thisModule)
         } else if (directive.directive == "%asminclude") {
-            val asm = loadAsmIncludeFile(directive.args[0].str!!, thisModule.source)
+            val asm = asmFileLoader(directive.args[0].str!!, thisModule.source)
             val scope = directive.definingSubroutine()
             if(scope!=null) {
                 scanAssemblyCode(asm, directive, scope)
@@ -89,7 +92,7 @@ class CallGraph(private val program: Program) : IAstVisitor {
 
     override fun visit(identifier: IdentifierReference) {
         // track symbol usage
-        val target = identifier.targetStatement(this.program.namespace)
+        val target = identifier.targetStatement(program)
         if (target != null) {
             addNodeAndParentScopes(target)
         }
@@ -117,19 +120,16 @@ class CallGraph(private val program: Program) : IAstVisitor {
     }
 
     override fun visit(decl: VarDecl) {
-        if (decl.autogeneratedDontRemove || decl.definingModule().isLibraryModule) {
-            // make sure autogenerated vardecls are in the used symbols and are never removed as 'unused'
+        if (decl.autogeneratedDontRemove || decl.datatype==DataType.STRUCT)
             addNodeAndParentScopes(decl)
-        }
-
-        if (decl.datatype == DataType.STRUCT)
+        else if(decl.parent is Block && decl.definingModule().isLibraryModule)
             addNodeAndParentScopes(decl)
 
         super.visit(decl)
     }
 
     override fun visit(functionCall: FunctionCall) {
-        val otherSub = functionCall.target.targetSubroutine(program.namespace)
+        val otherSub = functionCall.target.targetSubroutine(program)
         if (otherSub != null) {
             functionCall.definingSubroutine()?.let { thisSub ->
                 calls[thisSub] = calls.getValue(thisSub).plus(otherSub)
@@ -140,7 +140,7 @@ class CallGraph(private val program: Program) : IAstVisitor {
     }
 
     override fun visit(functionCallStatement: FunctionCallStatement) {
-        val otherSub = functionCallStatement.target.targetSubroutine(program.namespace)
+        val otherSub = functionCallStatement.target.targetSubroutine(program)
         if (otherSub != null) {
             functionCallStatement.definingSubroutine()?.let { thisSub ->
                 calls[thisSub] = calls.getValue(thisSub).plus(otherSub)
@@ -150,8 +150,19 @@ class CallGraph(private val program: Program) : IAstVisitor {
         super.visit(functionCallStatement)
     }
 
+    override fun visit(addressOf: AddressOf) {
+        val otherSub = addressOf.identifier.targetSubroutine(program)
+        if(otherSub!=null) {
+            addressOf.definingSubroutine()?.let { thisSub ->
+                calls[thisSub] = calls.getValue(thisSub).plus(otherSub)
+                calledBy[otherSub] = calledBy.getValue(otherSub).plus(thisSub)
+            }
+        }
+        super.visit(addressOf)
+    }
+
     override fun visit(jump: Jump) {
-        val otherSub = jump.identifier?.targetSubroutine(program.namespace)
+        val otherSub = jump.identifier?.targetSubroutine(program)
         if (otherSub != null) {
             jump.definingSubroutine()?.let { thisSub ->
                 calls[thisSub] = calls.getValue(thisSub).plus(otherSub)
@@ -167,7 +178,7 @@ class CallGraph(private val program: Program) : IAstVisitor {
     }
 
     override fun visit(inlineAssembly: InlineAssembly) {
-        // parse inline asm for subroutine calls (jmp, jsr)
+        // parse inline asm for subroutine calls (jmp, jsr, bra)
         val scope = inlineAssembly.definingSubroutine()
         scanAssemblyCode(inlineAssembly.assembly, inlineAssembly, scope)
         super.visit(inlineAssembly)
@@ -213,7 +224,7 @@ class CallGraph(private val program: Program) : IAstVisitor {
         }
     }
 
-    fun checkRecursiveCalls(errors: ErrorReporter) {
+    fun checkRecursiveCalls(errors: IErrorReporter) {
         val cycles = recursionCycles()
         if(cycles.any()) {
             errors.warn("Program contains recursive subroutine calls. These only works in very specific limited scenarios!", Position.DUMMY)

compiler/src/prog8/optimizer/ConstExprEvaluator.kt

@@ -9,28 +9,32 @@ import kotlin.math.pow
 class ConstExprEvaluator {
 
     fun evaluate(left: NumericLiteralValue, operator: String, right: NumericLiteralValue): Expression {
-        return when(operator) {
-            "+" -> plus(left, right)
-            "-" -> minus(left, right)
-            "*" -> multiply(left, right)
-            "/" -> divide(left, right)
-            "%" -> remainder(left, right)
-            "**" -> power(left, right)
-            "&" -> bitwiseand(left, right)
-            "|" -> bitwiseor(left, right)
-            "^" -> bitwisexor(left, right)
-            "and" -> logicaland(left, right)
-            "or" -> logicalor(left, right)
-            "xor" -> logicalxor(left, right)
-            "<" -> NumericLiteralValue.fromBoolean(left < right, left.position)
-            ">" -> NumericLiteralValue.fromBoolean(left > right, left.position)
-            "<=" -> NumericLiteralValue.fromBoolean(left <= right, left.position)
-            ">=" -> NumericLiteralValue.fromBoolean(left >= right, left.position)
-            "==" -> NumericLiteralValue.fromBoolean(left == right, left.position)
-            "!=" -> NumericLiteralValue.fromBoolean(left != right, left.position)
-            "<<" -> shiftedleft(left, right)
-            ">>" -> shiftedright(left, right)
-            else -> throw FatalAstException("const evaluation for invalid operator $operator")
+        try {
+            return when(operator) {
+                "+" -> plus(left, right)
+                "-" -> minus(left, right)
+                "*" -> multiply(left, right)
+                "/" -> divide(left, right)
+                "%" -> remainder(left, right)
+                "**" -> power(left, right)
+                "&" -> bitwiseand(left, right)
+                "|" -> bitwiseor(left, right)
+                "^" -> bitwisexor(left, right)
+                "and" -> logicaland(left, right)
+                "or" -> logicalor(left, right)
+                "xor" -> logicalxor(left, right)
+                "<" -> NumericLiteralValue.fromBoolean(left < right, left.position)
+                ">" -> NumericLiteralValue.fromBoolean(left > right, left.position)
+                "<=" -> NumericLiteralValue.fromBoolean(left <= right, left.position)
+                ">=" -> NumericLiteralValue.fromBoolean(left >= right, left.position)
+                "==" -> NumericLiteralValue.fromBoolean(left == right, left.position)
+                "!=" -> NumericLiteralValue.fromBoolean(left != right, left.position)
+                "<<" -> shiftedleft(left, right)
+                ">>" -> shiftedright(left, right)
+                else -> throw FatalAstException("const evaluation for invalid operator $operator")
+            }
+        } catch (ax: FatalAstException) {
+            throw ExpressionError(ax.message, left.position)
         }
     }
 

compiler/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -4,12 +4,16 @@ import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
-import prog8.ast.statements.*
+import prog8.ast.statements.Assignment
+import prog8.ast.statements.ForLoop
+import prog8.ast.statements.VarDecl
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.target.ICompilationTarget
+import kotlin.math.pow
 
 
-internal class ConstantFoldingOptimizer(private val program: Program) : AstWalker() {
+internal class ConstantFoldingOptimizer(private val program: Program, private val compTarget: ICompilationTarget) : AstWalker() {
     private val noModifications = emptyList<IAstModification>()
 
     override fun before(memread: DirectMemoryRead, parent: Node): Iterable<IAstModification> {
@@ -97,21 +101,63 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
     override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
         val leftconst = expr.left.constValue(program)
         val rightconst = expr.right.constValue(program)
+        val modifications = mutableListOf<IAstModification>()
 
-        val subExpr: BinaryExpression? = when {
-            leftconst!=null -> expr.right as? BinaryExpression
-            rightconst!=null -> expr.left as? BinaryExpression
-            else -> null
+        if(expr.operator == "**" && leftconst!=null) {
+            // optimize various simple cases of ** :
+            //  optimize away 1 ** x into just 1 and 0 ** x into just 0
+            //  optimize 2 ** x into (1<<x)  if both operands are integer.
+            val leftDt = leftconst.inferType(program).typeOrElse(DataType.STRUCT)
+            when (leftconst.number.toDouble()) {
+                0.0 -> {
+                    val value = NumericLiteralValue(leftDt, 0, expr.position)
+                    modifications += IAstModification.ReplaceNode(expr, value, parent)
+                }
+                1.0 -> {
+                    val value = NumericLiteralValue(leftDt, 1, expr.position)
+                    modifications += IAstModification.ReplaceNode(expr, value, parent)
+                }
+                2.0 -> {
+                    if(rightconst!=null) {
+                        val value = NumericLiteralValue(leftDt, 2.0.pow(rightconst.number.toDouble()), expr.position)
+                        modifications += IAstModification.ReplaceNode(expr, value, parent)
+                    } else {
+                        val rightDt = expr.right.inferType(program).typeOrElse(DataType.STRUCT)
+                        if(leftDt in IntegerDatatypes && rightDt in IntegerDatatypes) {
+                            val targetDt =
+                                when (parent) {
+                                    is Assignment -> parent.target.inferType(program).typeOrElse(DataType.STRUCT)
+                                    is VarDecl -> parent.datatype
+                                    else -> leftDt
+                                }
+                            val one = NumericLiteralValue(targetDt, 1, expr.position)
+                            val shift = BinaryExpression(one, "<<", expr.right, expr.position)
+                            modifications += IAstModification.ReplaceNode(expr, shift, parent)
+                        }
+                    }
+                }
+            }
         }
-        if(subExpr!=null) {
-            val subleftconst = subExpr.left.constValue(program)
-            val subrightconst = subExpr.right.constValue(program)
-            if ((subleftconst != null && subrightconst == null) || (subleftconst==null && subrightconst!=null)) {
-                // try reordering.
-                val change = groupTwoConstsTogether(expr, subExpr,
+
+        if(expr.inferType(program).istype(DataType.FLOAT)) {
+            val subExpr: BinaryExpression? = when {
+                leftconst != null -> expr.right as? BinaryExpression
+                rightconst != null -> expr.left as? BinaryExpression
+                else -> null
+            }
+            if (subExpr != null) {
+                val subleftconst = subExpr.left.constValue(program)
+                val subrightconst = subExpr.right.constValue(program)
+                if ((subleftconst != null && subrightconst == null) || (subleftconst == null && subrightconst != null)) {
+                    // try reordering.
+                    val change = groupTwoFloatConstsTogether(
+                        expr, subExpr,
                         leftconst != null, rightconst != null,
-                        subleftconst != null, subrightconst != null)
-                return change?.let { listOf(it) } ?: noModifications
+                        subleftconst != null, subrightconst != null
+                    )
+                    if (change != null)
+                        modifications += change
+                }
             }
         }
 
@@ -119,10 +165,10 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         if(leftconst != null && rightconst != null) {
             val evaluator = ConstExprEvaluator()
             val result = evaluator.evaluate(leftconst, expr.operator, rightconst)
-            return listOf(IAstModification.ReplaceNode(expr, result, parent))
+            modifications += IAstModification.ReplaceNode(expr, result, parent)
         }
 
-        return noModifications
+        return modifications
     }
 
     override fun after(array: ArrayLiteralValue, parent: Node): Iterable<IAstModification> {
@@ -178,7 +224,7 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
                     range.step
                 }
 
-            return RangeExpr(fromCast.valueOrZero(), toCast.valueOrZero(), newStep, range.position)
+            return RangeExpr(fromCast.valueOrZero(), toCast.valueOrZero(), newStep, compTarget, range.position)
         }
 
         // adjust the datatype of a range expression in for loops to the loop variable.
@@ -187,7 +233,8 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         val rangeTo = iterableRange.to as? NumericLiteralValue
         if(rangeFrom==null || rangeTo==null) return noModifications
 
-        val loopvar = forLoop.loopVar.targetVarDecl(program.namespace)!!
+        val loopvar = forLoop.loopVar.targetVarDecl(program) ?: throw UndefinedSymbolError(forLoop.loopVar)
+
         val stepLiteral = iterableRange.step as? NumericLiteralValue
         when(loopvar.datatype) {
             DataType.UBYTE -> {
@@ -258,13 +305,15 @@ internal class ConstantFoldingOptimizer(private val program: Program) : AstWalke
         }
     }
 
-    private fun groupTwoConstsTogether(expr: BinaryExpression,
-                                       subExpr: BinaryExpression,
-                                       leftIsConst: Boolean,
-                                       rightIsConst: Boolean,
-                                       subleftIsConst: Boolean,
-                                       subrightIsConst: Boolean): IAstModification?
+    private fun groupTwoFloatConstsTogether(expr: BinaryExpression,
+       subExpr: BinaryExpression,
+       leftIsConst: Boolean,
+       rightIsConst: Boolean,
+       subleftIsConst: Boolean,
+       subrightIsConst: Boolean): IAstModification?
     {
+        // NOTE: THIS IS ONLY VALID ON FLOATING POINT CONSTANTS
+
         // todo: this implements only a small set of possible reorderings at this time
         if(expr.operator==subExpr.operator) {
             // both operators are the same.

compiler/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -4,27 +4,33 @@ import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
 import prog8.ast.statements.ArrayIndex
 import prog8.ast.statements.AssignTarget
 import prog8.ast.statements.ForLoop
 import prog8.ast.statements.VarDecl
-import prog8.compiler.target.CompilationTarget
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
 
 // Fix up the literal value's type to match that of the vardecl
-internal class VarConstantValueTypeAdjuster(private val program: Program) : AstWalker() {
+internal class VarConstantValueTypeAdjuster(private val program: Program, private val errors: IErrorReporter) : AstWalker() {
     private val noModifications = emptyList<IAstModification>()
 
     override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
-        val declConstValue = decl.value?.constValue(program)
-        if(declConstValue!=null && (decl.type==VarDeclType.VAR || decl.type==VarDeclType.CONST)
+        try {
+            val declConstValue = decl.value?.constValue(program)
+            if(declConstValue!=null && (decl.type==VarDeclType.VAR || decl.type==VarDeclType.CONST)
                 && !declConstValue.inferType(program).istype(decl.datatype)) {
-            // cast the numeric literal to the appropriate datatype of the variable
-            val cast = declConstValue.cast(decl.datatype)
-            if(cast.isValid)
-                return listOf(IAstModification.ReplaceNode(decl.value!!, cast.valueOrZero(), decl))
+                // cast the numeric literal to the appropriate datatype of the variable
+                val cast = declConstValue.cast(decl.datatype)
+                if(cast.isValid)
+                    return listOf(IAstModification.ReplaceNode(decl.value!!, cast.valueOrZero(), decl))
+            }
+        } catch (x: UndefinedSymbolError) {
+            errors.err(x.message, x.position)
         }
+
         return noModifications
     }
 }
@@ -33,7 +39,7 @@ internal class VarConstantValueTypeAdjuster(private val program: Program) : AstW
 // Replace all constant identifiers with their actual value,
 // and the array var initializer values and sizes.
 // This is needed because further constant optimizations depend on those.
-internal class ConstantIdentifierReplacer(private val program: Program, private val errors: ErrorReporter) : AstWalker() {
+internal class ConstantIdentifierReplacer(private val program: Program, private val errors: IErrorReporter, private val compTarget: ICompilationTarget) : AstWalker() {
     private val noModifications = emptyList<IAstModification>()
 
     override fun after(identifier: IdentifierReference, parent: Node): Iterable<IAstModification> {
@@ -47,18 +53,29 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
         if(forloop!=null && identifier===forloop.loopVar)
             return noModifications
 
-        val cval = identifier.constValue(program) ?: return noModifications
-        return when (cval.type) {
-            in NumericDatatypes -> listOf(IAstModification.ReplaceNode(identifier, NumericLiteralValue(cval.type, cval.number, identifier.position), identifier.parent))
-            in PassByReferenceDatatypes -> throw FatalAstException("pass-by-reference type should not be considered a constant")
-            else -> noModifications
+        try {
+            val cval = identifier.constValue(program) ?: return noModifications
+            return when (cval.type) {
+                in NumericDatatypes -> listOf(
+                    IAstModification.ReplaceNode(
+                        identifier,
+                        NumericLiteralValue(cval.type, cval.number, identifier.position),
+                        identifier.parent
+                    )
+                )
+                in PassByReferenceDatatypes -> throw FatalAstException("pass-by-reference type should not be considered a constant")
+                else -> noModifications
+            }
+        } catch (x: UndefinedSymbolError) {
+            errors.err(x.message, x.position)
+            return noModifications
         }
     }
 
     override fun before(decl: VarDecl, parent: Node): Iterable<IAstModification> {
         // the initializer value can't refer to the variable itself (recursive definition)
         // TODO: use call graph for this?
-        if(decl.value?.referencesIdentifier(decl.name) == true || decl.arraysize?.indexVar?.referencesIdentifier(decl.name) == true) {
+        if(decl.value?.referencesIdentifier(decl.name) == true || decl.arraysize?.indexExpr?.referencesIdentifier(decl.name) == true) {
             errors.err("recursive var declaration", decl.position)
             return noModifications
         }
@@ -76,14 +93,6 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                                 decl
                         ))
                     }
-                } else if(arraysize.constIndex()==null) {
-                    // see if we can calculate the size from other fields
-                    val cval = arraysize.indexVar?.constValue(program) ?: arraysize.origExpression?.constValue(program)
-                    if(cval!=null) {
-                        arraysize.indexVar = null
-                        arraysize.origExpression = null
-                        arraysize.indexNum = cval
-                    }
                 }
             }
 
@@ -170,7 +179,7 @@ internal class ConstantIdentifierReplacer(private val program: Program, private
                     if(rangeExpr==null && litval!=null) {
                         // arraysize initializer is a single int, and we know the size.
                         val fillvalue = litval.number.toDouble()
-                        if (fillvalue < CompilationTarget.instance.machine.FLOAT_MAX_NEGATIVE || fillvalue > CompilationTarget.instance.machine.FLOAT_MAX_POSITIVE)
+                        if (fillvalue < compTarget.machine.FLOAT_MAX_NEGATIVE || fillvalue > compTarget.machine.FLOAT_MAX_POSITIVE)
                             errors.err("float value overflow", litval.position)
                         else {
                             // create the array itself, filled with the fillvalue.

compiler/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -2,10 +2,14 @@ package prog8.optimizer
 
 import prog8.ast.Node
 import prog8.ast.Program
-import prog8.ast.base.*
+import prog8.ast.base.DataType
+import prog8.ast.base.FatalAstException
+import prog8.ast.base.IntegerDatatypes
+import prog8.ast.base.NumericDatatypes
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
+import prog8.ast.statements.Assignment
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
 import kotlin.math.abs
 import kotlin.math.log2
 import kotlin.math.pow
@@ -98,6 +102,12 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (leftVal != null && expr.operator in associativeOperators && rightVal == null)
             return listOf(IAstModification.SwapOperands(expr))
 
+        // NonBinaryExpression  <associativeoperator>  BinaryExpression  -->  BinaryExpression  <associativeoperator>  NonBinaryExpression
+        if (expr.operator in associativeOperators && expr.left !is BinaryExpression && expr.right is BinaryExpression) {
+            if(parent !is Assignment || !(expr.left isSameAs parent.target))
+                return listOf(IAstModification.SwapOperands(expr))
+        }
+
         // X + (-A)  -->  X - A
         if (expr.operator == "+" && (expr.right as? PrefixExpression)?.operator == "-") {
             return listOf(IAstModification.ReplaceNode(
@@ -338,7 +348,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (leftVal == null && rightVal == null)
             return null
 
-        val (expr2, _, rightVal2) = reorderAssociative(expr, leftVal)
+        val (expr2, _, rightVal2) = reorderAssociativeWithConstant(expr, leftVal)
         if (rightVal2 != null) {
             // right value is a constant, see if we can optimize
             val rightConst: NumericLiteralValue = rightVal2
@@ -562,7 +572,7 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         if (leftVal == null && rightVal == null)
             return null
 
-        val (expr2, _, rightVal2) = reorderAssociative(expr, leftVal)
+        val (expr2, _, rightVal2) = reorderAssociativeWithConstant(expr, leftVal)
         if (rightVal2 != null) {
             // right value is a constant, see if we can optimize
             val leftValue: Expression = expr2.left
@@ -662,12 +672,15 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
             DataType.UWORD -> {
                 if (amount >= 16) {
                     return NumericLiteralValue.optimalInteger(0, expr.position)
-                } else if (amount >= 8) {
+                }
+                else if (amount >= 8) {
                     val msb = FunctionCall(IdentifierReference(listOf("msb"), expr.position), mutableListOf(expr.left), expr.position)
                     if (amount == 8) {
-                        return TypecastExpression(msb, DataType.UWORD, true, expr.position)
+                        // mkword(0, msb(v))
+                        val zero = NumericLiteralValue(DataType.UBYTE, 0, expr.position)
+                        return FunctionCall(IdentifierReference(listOf("mkword"), expr.position), mutableListOf(zero, msb), expr.position)
                     }
-                    return BinaryExpression(msb, ">>", NumericLiteralValue.optimalInteger(amount - 8, expr.position), expr.position)
+                    return TypecastExpression(BinaryExpression(msb, ">>", NumericLiteralValue.optimalInteger(amount - 8, expr.position), expr.position), DataType.UWORD, true, expr.position)
                 }
             }
             DataType.WORD -> {
@@ -682,17 +695,17 @@ internal class ExpressionSimplifier(private val program: Program) : AstWalker()
         return null
     }
 
-    private fun reorderAssociative(expr: BinaryExpression, leftVal: NumericLiteralValue?): ReorderedAssociativeBinaryExpr {
+    private fun reorderAssociativeWithConstant(expr: BinaryExpression, leftVal: NumericLiteralValue?): BinExprWithConstants {
         if (expr.operator in associativeOperators && leftVal != null) {
             // swap left and right so that right is always the constant
             val tmp = expr.left
             expr.left = expr.right
             expr.right = tmp
-            return ReorderedAssociativeBinaryExpr(expr, expr.right.constValue(program), leftVal)
+            return BinExprWithConstants(expr, expr.right.constValue(program), leftVal)
         }
-        return ReorderedAssociativeBinaryExpr(expr, leftVal, expr.right.constValue(program))
+        return BinExprWithConstants(expr, leftVal, expr.right.constValue(program))
     }
 
-    private data class ReorderedAssociativeBinaryExpr(val expr: BinaryExpression, val leftVal: NumericLiteralValue?, val rightVal: NumericLiteralValue?)
+    private data class BinExprWithConstants(val expr: BinaryExpression, val leftVal: NumericLiteralValue?, val rightVal: NumericLiteralValue?)
 
 }

compiler/src/prog8/optimizer/Extensions.kt

@@ -1,16 +1,19 @@
 package prog8.optimizer
 
+import prog8.ast.IBuiltinFunctions
 import prog8.ast.Program
-import prog8.ast.base.ErrorReporter
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
+import java.nio.file.Path
 
 
-internal fun Program.constantFold(errors: ErrorReporter) {
-    val valuetypefixer = VarConstantValueTypeAdjuster(this)
+internal fun Program.constantFold(errors: IErrorReporter, compTarget: ICompilationTarget) {
+    val valuetypefixer = VarConstantValueTypeAdjuster(this, errors)
     valuetypefixer.visit(this)
     if(errors.isEmpty()) {
         valuetypefixer.applyModifications()
 
-        val replacer = ConstantIdentifierReplacer(this, errors)
+        val replacer = ConstantIdentifierReplacer(this, errors, compTarget)
         replacer.visit(this)
         if (errors.isEmpty()) {
             replacer.applyModifications()
@@ -19,7 +22,7 @@ internal fun Program.constantFold(errors: ErrorReporter) {
             if(errors.isEmpty()) {
                 valuetypefixer.applyModifications()
 
-                val optimizer = ConstantFoldingOptimizer(this)
+                val optimizer = ConstantFoldingOptimizer(this, compTarget)
                 optimizer.visit(this)
                 while (errors.isEmpty() && optimizer.applyModifications() > 0) {
                     optimizer.visit(this)
@@ -38,8 +41,11 @@ internal fun Program.constantFold(errors: ErrorReporter) {
 }
 
 
-internal fun Program.optimizeStatements(errors: ErrorReporter): Int {
-    val optimizer = StatementOptimizer(this, errors)
+internal fun Program.optimizeStatements(errors: IErrorReporter,
+                                        functions: IBuiltinFunctions,
+                                        compTarget: ICompilationTarget,
+                                        asmFileLoader: (filename: String, source: Path)->String): Int {
+    val optimizer = StatementOptimizer(this, errors, functions, compTarget, asmFileLoader)
     optimizer.visit(this)
     val optimizationCount = optimizer.applyModifications()
 
@@ -54,8 +60,8 @@ internal fun Program.simplifyExpressions() : Int {
     return opti.applyModifications()
 }
 
-internal fun Program.splitBinaryExpressions() : Int {
-    val opti = BinExprSplitter(this)
+internal fun Program.splitBinaryExpressions(compTarget: ICompilationTarget) : Int {
+    val opti = BinExprSplitter(this, compTarget)
     opti.visit(this)
     return opti.applyModifications()
 }

compiler/src/prog8/optimizer/StatementOptimizer.kt

@@ -1,30 +1,36 @@
 package prog8.optimizer
 
+import prog8.ast.IBuiltinFunctions
 import prog8.ast.INameScope
 import prog8.ast.Node
 import prog8.ast.Program
 import prog8.ast.base.*
 import prog8.ast.expressions.*
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
-import prog8.ast.processing.IAstVisitor
 import prog8.ast.statements.*
-import prog8.compiler.target.CompilationTarget
-import prog8.functions.BuiltinFunctions
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.ast.walk.IAstVisitor
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
+import java.nio.file.Path
 import kotlin.math.floor
 
 
 internal class StatementOptimizer(private val program: Program,
-                                  private val errors: ErrorReporter) : AstWalker() {
+                                  private val errors: IErrorReporter,
+                                  private val functions: IBuiltinFunctions,
+                                  private val compTarget: ICompilationTarget,
+                                  asmFileLoader: (filename: String, source: Path)->String
+) : AstWalker() {
 
     private val noModifications = emptyList<IAstModification>()
-    private val callgraph = CallGraph(program)
-    private val pureBuiltinFunctions = BuiltinFunctions.filter { it.value.pure }
+    private val callgraph = CallGraph(program, asmFileLoader)
 
     override fun after(block: Block, parent: Node): Iterable<IAstModification> {
         if("force_output" !in block.options()) {
             if (block.containsNoCodeNorVars()) {
-                errors.warn("removing empty block '${block.name}'", block.position)
+                if(block.name != program.internedStringsModuleName)
+                    errors.warn("removing empty block '${block.name}'", block.position)
                 return listOf(IAstModification.Remove(block, parent as INameScope))
             }
 
@@ -39,7 +45,7 @@ internal class StatementOptimizer(private val program: Program,
     override fun after(subroutine: Subroutine, parent: Node): Iterable<IAstModification> {
         val forceOutput = "force_output" in subroutine.definingBlock().options()
         if(subroutine.asmAddress==null && !forceOutput) {
-            if(subroutine.containsNoCodeNorVars()) {
+            if(subroutine.containsNoCodeNorVars() && !subroutine.inline) {
                 errors.warn("removing empty subroutine '${subroutine.name}'", subroutine.position)
                 val removals = callgraph.calledBy.getValue(subroutine).map {
                     IAstModification.Remove(it, it.definingScope())
@@ -50,8 +56,10 @@ internal class StatementOptimizer(private val program: Program,
         }
 
         if(subroutine !in callgraph.usedSymbols && !forceOutput) {
-            errors.warn("removing unused subroutine '${subroutine.name}'", subroutine.position)
-            return listOf(IAstModification.Remove(subroutine, subroutine.definingScope()))
+            if(!subroutine.isAsmSubroutine) {
+                errors.warn("removing unused subroutine '${subroutine.name}'", subroutine.position)
+                return listOf(IAstModification.Remove(subroutine, subroutine.definingScope()))
+            }
         }
 
         return noModifications
@@ -70,60 +78,58 @@ internal class StatementOptimizer(private val program: Program,
     }
 
     override fun after(functionCallStatement: FunctionCallStatement, parent: Node): Iterable<IAstModification> {
-        if(functionCallStatement.target.nameInSource.size==1 && functionCallStatement.target.nameInSource[0] in BuiltinFunctions) {
+        if(functionCallStatement.target.nameInSource.size==1 && functionCallStatement.target.nameInSource[0] in functions.names) {
             val functionName = functionCallStatement.target.nameInSource[0]
-            if (functionName in pureBuiltinFunctions) {
+            if (functionName in functions.purefunctionNames) {
                 errors.warn("statement has no effect (function return value is discarded)", functionCallStatement.position)
                 return listOf(IAstModification.Remove(functionCallStatement, functionCallStatement.definingScope()))
             }
         }
 
         // printing a literal string of just 2 or 1 characters is replaced by directly outputting those characters
-        // this is a C-64 specific optimization
-        if(functionCallStatement.target.nameInSource==listOf("c64scr", "print")) {
+        if(functionCallStatement.target.nameInSource==listOf("txt", "print")) {
             val arg = functionCallStatement.args.single()
-            val stringVar: IdentifierReference?
-            stringVar = if(arg is AddressOf) {
+            val stringVar: IdentifierReference? = if(arg is AddressOf) {
                 arg.identifier
             } else {
                 arg as? IdentifierReference
             }
             if(stringVar!=null) {
-                val vardecl = stringVar.targetVarDecl(program.namespace)!!
+                val vardecl = stringVar.targetVarDecl(program)!!
                 val string = vardecl.value as? StringLiteralValue
                 if(string!=null) {
                     val pos = functionCallStatement.position
                     if (string.value.length == 1) {
-                        val firstCharEncoded = CompilationTarget.instance.encodeString(string.value, string.altEncoding)[0]
+                        val firstCharEncoded = compTarget.encodeString(string.value, string.altEncoding)[0]
                         val chrout = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
+                                IdentifierReference(listOf("txt", "chrout"), pos),
                                 mutableListOf(NumericLiteralValue(DataType.UBYTE, firstCharEncoded.toInt(), pos)),
                                 functionCallStatement.void, pos
                         )
                         return listOf(IAstModification.ReplaceNode(functionCallStatement, chrout, parent))
                     } else if (string.value.length == 2) {
-                        val firstTwoCharsEncoded = CompilationTarget.instance.encodeString(string.value.take(2), string.altEncoding)
+                        val firstTwoCharsEncoded = compTarget.encodeString(string.value.take(2), string.altEncoding)
                         val chrout1 = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
+                                IdentifierReference(listOf("txt", "chrout"), pos),
                                 mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[0].toInt(), pos)),
                                 functionCallStatement.void, pos
                         )
                         val chrout2 = FunctionCallStatement(
-                                IdentifierReference(listOf("c64", "CHROUT"), pos),
+                                IdentifierReference(listOf("txt", "chrout"), pos),
                                 mutableListOf(NumericLiteralValue(DataType.UBYTE, firstTwoCharsEncoded[1].toInt(), pos)),
                                 functionCallStatement.void, pos
                         )
-                        val anonscope = AnonymousScope(mutableListOf(), pos)
-                        anonscope.statements.add(chrout1)
-                        anonscope.statements.add(chrout2)
-                        return listOf(IAstModification.ReplaceNode(functionCallStatement, anonscope, parent))
+                        return listOf(
+                            IAstModification.InsertBefore(functionCallStatement, chrout1, parent as INameScope),
+                            IAstModification.ReplaceNode(functionCallStatement, chrout2, parent)
+                        )
                     }
                 }
             }
         }
 
         // if the first instruction in the called subroutine is a return statement, remove the jump altogeter
-        val subroutine = functionCallStatement.target.targetSubroutine(program.namespace)
+        val subroutine = functionCallStatement.target.targetSubroutine(program)
         if(subroutine!=null) {
             val first = subroutine.statements.asSequence().filterNot { it is VarDecl || it is Directive }.firstOrNull()
             if(first is Return)
@@ -135,7 +141,7 @@ internal class StatementOptimizer(private val program: Program,
 
     override fun before(functionCall: FunctionCall, parent: Node): Iterable<IAstModification> {
         // if the first instruction in the called subroutine is a return statement with constant value, replace with the constant value
-        val subroutine = functionCall.target.targetSubroutine(program.namespace)
+        val subroutine = functionCall.target.targetSubroutine(program)
         if(subroutine!=null) {
             val first = subroutine.statements.asSequence().filterNot { it is VarDecl || it is Directive }.firstOrNull()
             if(first is Return && first.value!=null) {
@@ -203,14 +209,14 @@ internal class StatementOptimizer(private val program: Program,
                 return listOf(IAstModification.ReplaceNode(forLoop, scope, parent))
             }
         }
-        val iterable = (forLoop.iterable as? IdentifierReference)?.targetVarDecl(program.namespace)
+        val iterable = (forLoop.iterable as? IdentifierReference)?.targetVarDecl(program)
         if(iterable!=null) {
             if(iterable.datatype==DataType.STR) {
                 val sv = iterable.value as StringLiteralValue
                 val size = sv.value.length
                 if(size==1) {
                     // loop over string of length 1 -> just assign the single character
-                    val character = CompilationTarget.instance.encodeString(sv.value, sv.altEncoding)[0]
+                    val character = compTarget.encodeString(sv.value, sv.altEncoding)[0]
                     val byte = NumericLiteralValue(DataType.UBYTE, character, iterable.position)
                     val scope = AnonymousScope(mutableListOf(), forLoop.position)
                     scope.statements.add(Assignment(AssignTarget(forLoop.loopVar, null, null, forLoop.position), byte, forLoop.position))
@@ -291,22 +297,10 @@ internal class StatementOptimizer(private val program: Program,
         return noModifications
     }
 
-    override fun after(whenStatement: WhenStatement, parent: Node): Iterable<IAstModification> {
-        // remove empty choices
-        class ChoiceRemover(val choice: WhenChoice) : IAstModification {
-            override fun perform() {
-                whenStatement.choices.remove(choice)
-            }
-        }
-        return whenStatement.choices
-                .filter { !it.statements.containsCodeOrVars() }
-                .map { ChoiceRemover(it) }
-    }
-
     override fun after(jump: Jump, parent: Node): Iterable<IAstModification> {
         // if the jump is to the next statement, remove the jump
         val scope = jump.definingScope()
-        val label = jump.identifier?.targetStatement(scope)
+        val label = jump.identifier?.targetStatement(program)
         if(label!=null && scope.statements.indexOf(label) == scope.statements.indexOf(jump)+1)
             return listOf(IAstModification.Remove(jump, jump.definingScope()))
 
@@ -323,7 +317,7 @@ internal class StatementOptimizer(private val program: Program,
                     val op1 = binExpr.operator
                     val op2 = rExpr.operator
 
-                    if(rExpr.left is NumericLiteralValue && op2 in setOf("+", "*", "&", "|")) {
+                    if(rExpr.left is NumericLiteralValue && op2 in associativeOperators) {
                         // associative operator, make sure the constant numeric value is second (right)
                         return listOf(IAstModification.SwapOperands(rExpr))
                     }
@@ -390,7 +384,7 @@ internal class StatementOptimizer(private val program: Program,
                 // assignments of the form:  X = X <operator> <expr>
                 // remove assignments that have no effect (such as X=X+0)
                 // optimize/rewrite some other expressions
-                val vardeclDt = (assignment.target.identifier?.targetVarDecl(program.namespace))?.type
+                val vardeclDt = (assignment.target.identifier?.targetVarDecl(program))?.type
                 when (bexpr.operator) {
                     "+" -> {
                         if (rightCv == 0.0) {
@@ -441,6 +435,47 @@ internal class StatementOptimizer(private val program: Program,
         return noModifications
     }
 
+    override fun after(returnStmt: Return, parent: Node): Iterable<IAstModification> {
+        fun returnViaIntermediary(value: Expression): Iterable<IAstModification>? {
+            val returnDt = returnStmt.definingSubroutine()!!.returntypes.single()
+            if (returnDt in IntegerDatatypes) {
+                // first assign to intermediary, then return that register
+                val returnValueIntermediary =
+                    when(returnDt) {
+                        DataType.UBYTE -> IdentifierReference(listOf("prog8_lib", "retval_interm_ub"), returnStmt.position)
+                        DataType.BYTE -> IdentifierReference(listOf("prog8_lib", "retval_interm_b"), returnStmt.position)
+                        DataType.UWORD -> IdentifierReference(listOf("prog8_lib", "retval_interm_uw"), returnStmt.position)
+                        DataType.WORD -> IdentifierReference(listOf("prog8_lib", "retval_interm_w"), returnStmt.position)
+                        else -> throw FatalAstException("weird return dt")
+                    }
+                val tgt = AssignTarget(returnValueIntermediary, null, null, returnStmt.position)
+                val assign = Assignment(tgt, value, returnStmt.position)
+                val returnReplacement = Return(returnValueIntermediary, returnStmt.position)
+                return listOf(
+                    IAstModification.InsertBefore(returnStmt, assign, parent as INameScope),
+                    IAstModification.ReplaceNode(returnStmt, returnReplacement, parent)
+                )
+            }
+            return null
+        }
+
+        when(returnStmt.value) {
+            is PrefixExpression -> {
+                val mod = returnViaIntermediary(returnStmt.value!!)
+                if(mod!=null)
+                    return mod
+            }
+            is BinaryExpression -> {
+                val mod = returnViaIntermediary(returnStmt.value!!)
+                if(mod!=null)
+                    return mod
+            }
+            else -> {}
+        }
+
+        return super.after(returnStmt, parent)
+    }
+
     private fun hasBreak(scope: INameScope): Boolean {
 
         class Searcher: IAstVisitor

compiler/src/prog8/optimizer/UnusedCodeRemover.kt

@@ -3,23 +3,34 @@ package prog8.optimizer
 import prog8.ast.INameScope
 import prog8.ast.Node
 import prog8.ast.Program
-import prog8.ast.base.ErrorReporter
-import prog8.ast.processing.AstWalker
-import prog8.ast.processing.IAstModification
+import prog8.ast.expressions.BinaryExpression
+import prog8.ast.expressions.FunctionCall
+import prog8.ast.expressions.PrefixExpression
+import prog8.ast.expressions.TypecastExpression
 import prog8.ast.statements.*
+import prog8.ast.walk.AstWalker
+import prog8.ast.walk.IAstModification
+import prog8.compiler.IErrorReporter
+import prog8.compiler.target.ICompilationTarget
+import java.nio.file.Path
 
 
-internal class UnusedCodeRemover(private val program: Program, private val errors: ErrorReporter): AstWalker() {
+internal class UnusedCodeRemover(private val program: Program,
+                                 private val errors: IErrorReporter,
+                                 private val compTarget: ICompilationTarget,
+                                 private val asmFileLoader: (filename: String, source: Path)->String): AstWalker() {
 
     override fun before(program: Program, parent: Node): Iterable<IAstModification> {
-        val callgraph = CallGraph(program)
+        val callgraph = CallGraph(program, asmFileLoader)
         val removals = mutableListOf<IAstModification>()
 
         // remove all subroutines that aren't called, or are empty
+        // NOTE: part of this is also done already in the StatementOptimizer
         val entrypoint = program.entrypoint()
         program.modules.forEach {
             callgraph.forAllSubroutines(it) { sub ->
-                if (sub !== entrypoint && !sub.isAsmSubroutine && (callgraph.calledBy[sub].isNullOrEmpty() || sub.containsNoCodeNorVars())) {
+                val forceOutput = "force_output" in sub.definingBlock().options()
+                if (sub !== entrypoint && !forceOutput && !sub.isAsmSubroutine && (callgraph.calledBy[sub].isNullOrEmpty() || sub.containsNoCodeNorVars())) {
                     removals.add(IAstModification.Remove(sub, sub.definingScope()))
                 }
             }
@@ -91,9 +102,16 @@ internal class UnusedCodeRemover(private val program: Program, private val error
             val assign1 = stmtPairs[0] as? Assignment
             val assign2 = stmtPairs[1] as? Assignment
             if (assign1 != null && assign2 != null && !assign2.isAugmentable) {
-                if (assign1.target.isSameAs(assign2.target, program) && assign1.target.isInRegularRAM(program.namespace))  {
+                if (assign1.target.isSameAs(assign2.target, program) && compTarget.isInRegularRAM(assign1.target, program))  {
                     if(assign2.target.identifier==null || !assign2.value.referencesIdentifier(*(assign2.target.identifier!!.nameInSource.toTypedArray())))
-                        linesToRemove.add(assign1)
+                        // only remove the second assignment if its value is a simple expression!
+                        when(assign2.value) {
+                            is PrefixExpression,
+                            is BinaryExpression,
+                            is TypecastExpression,
+                            is FunctionCall -> { /* don't remove */ }
+                            else -> linesToRemove.add(assign1)
+                        }
                 }
             }
         }

compiler/src/prog8/server/dbus/IrmenDbusTest.kt

@@ -1,22 +0,0 @@
-package prog8.server.dbus
-
-//import org.freedesktop.dbus.interfaces.DBusInterface
-//
-//
-//interface IrmenDbusTest: DBusInterface
-//{
-//    fun Status(address: String): Map<Int, String>
-//}
-//
-//
-//internal class TestService: IrmenDbusTest {
-//    override fun Status(address: String): Map<Int, String> {
-//        return mapOf(
-//            5 to "hello",
-//            42 to address
-//        )
-//    }
-//
-//    override fun isRemote() = true
-//    override fun getObjectPath() = "/razorvine/TestService"
-//}

compiler/src/prog8/server/dbus/clientmain.kt

@@ -1,17 +0,0 @@
-package prog8.server.dbus
-
-
-//import org.freedesktop.dbus.connections.impl.DBusConnection
-//
-//
-//fun main() {
-//    DBusConnection.getConnection(DBusConnection.DBusBusType.SESSION).use {
-//        println(it.names.toList())
-//        println(it.uniqueName)
-//        println(it.address)
-//        println(it.machineId)
-//        val obj = it.getRemoteObject("local.net.razorvine.dbus.test", "/razorvine/TestService", IrmenDbusTest::class.java)
-//        println(obj.Status("irmen"))
-//    }
-//}
-//

compiler/src/prog8/server/dbus/testdbus.kt

@@ -1,18 +0,0 @@
-package prog8.server.dbus
-
-//import org.freedesktop.dbus.connections.impl.DBusConnection
-//
-//
-//fun main() {
-//    DBusConnection.getConnection(DBusConnection.DBusBusType.SESSION).use {
-//        it.requestBusName("local.net.razorvine.dbus.test")
-//        println(it.names.toList())
-//        println(it.uniqueName)
-//        println(it.address)
-//        println(it.machineId)
-//        val service = TestService()
-//        it.exportObject(service.objectPath, service)
-//
-//        Thread.sleep(100000)
-//    }
-//}

compiler/test/UnitTests.kt

@@ -5,19 +5,24 @@ import org.hamcrest.Matchers.closeTo
 import org.hamcrest.Matchers.equalTo
 import org.junit.jupiter.api.Test
 import org.junit.jupiter.api.TestInstance
-import prog8.ast.base.*
+import prog8.ast.*
+import prog8.ast.base.DataType
+import prog8.ast.base.ParentSentinel
+import prog8.ast.base.Position
+import prog8.ast.base.VarDeclType
 import prog8.ast.expressions.*
 import prog8.ast.statements.*
 import prog8.compiler.*
 import prog8.compiler.target.C64Target
-import prog8.compiler.target.CompilationTarget
+import prog8.compiler.target.Cx16Target
 import prog8.compiler.target.c64.C64MachineDefinition.C64Zeropage
 import prog8.compiler.target.c64.C64MachineDefinition.FLOAT_MAX_NEGATIVE
 import prog8.compiler.target.c64.C64MachineDefinition.FLOAT_MAX_POSITIVE
 import prog8.compiler.target.c64.C64MachineDefinition.Mflpt5
 import prog8.compiler.target.c64.Petscii
-import prog8.compiler.target.cx16.CX16MachineDefinition
+import prog8.compiler.target.cx16.CX16MachineDefinition.CX16Zeropage
 import java.io.CharConversionException
+import java.nio.file.Path
 import kotlin.test.*
 
 @TestInstance(TestInstance.Lifecycle.PER_CLASS)
@@ -50,8 +55,8 @@ class TestCompiler {
         assertEquals("-\$c382", (-50050).toHex())
         assertEquals("-\$ffff", (-65535).toHex())
         assertEquals("-\$ffff", (-65535L).toHex())
-        assertFailsWith<CompilerException> { 65536.toHex()  }
-        assertFailsWith<CompilerException> { 65536L.toHex()  }
+        assertFailsWith<IllegalArgumentException> { 65536.toHex()  }
+        assertFailsWith<IllegalArgumentException> { 65536L.toHex()  }
     }
 
     @Test
@@ -130,7 +135,7 @@ class TestC64Zeropage {
 
     @Test
     fun testNames() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false, C64Target))
 
         zp.allocate("", DataType.UBYTE, null, errors)
         zp.allocate("", DataType.UBYTE, null, errors)
@@ -143,37 +148,37 @@ class TestC64Zeropage {
 
     @Test
     fun testZpFloatEnable() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, C64Target))
         assertFailsWith<CompilerException> {
             zp.allocate("", DataType.FLOAT, null, errors)
         }
-        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.DONTUSE, emptyList(), true, false))
+        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.DONTUSE, emptyList(), true, false, C64Target))
         assertFailsWith<CompilerException> {
             zp2.allocate("", DataType.FLOAT, null, errors)
         }
-        val zp3 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), true, false))
+        val zp3 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), true, false, C64Target))
         zp3.allocate("", DataType.FLOAT, null, errors)
     }
 
     @Test
     fun testZpModesWithFloats() {
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false))
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), false, false))
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false))
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), false, false))
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false))
-        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), true, false))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, C64Target))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), false, false, C64Target))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false, C64Target))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), false, false, C64Target))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false, C64Target))
+        C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), true, false, C64Target))
         assertFailsWith<CompilerException> {
-            C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), true, false))
+            C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), true, false, C64Target))
         }
         assertFailsWith<CompilerException> {
-            C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), true, false))
+            C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), true, false, C64Target))
         }
     }
 
     @Test
     fun testZpDontuse() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.DONTUSE, emptyList(), false, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.DONTUSE, emptyList(), false, false, C64Target))
         println(zp.free)
         assertEquals(0, zp.available())
         assertFailsWith<CompilerException> {
@@ -183,19 +188,19 @@ class TestC64Zeropage {
 
     @Test
     fun testFreeSpaces() {
-        val zp1 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false))
+        val zp1 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false, C64Target))
         assertEquals(18, zp1.available())
-        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), false, false))
-        assertEquals(89, zp2.available())
-        val zp3 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), false, false))
+        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FLOATSAFE, emptyList(), false, false, C64Target))
+        assertEquals(85, zp2.available())
+        val zp3 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), false, false, C64Target))
         assertEquals(125, zp3.available())
-        val zp4 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false))
+        val zp4 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, C64Target))
         assertEquals(238, zp4.available())
     }
 
     @Test
     fun testReservedSpace() {
-        val zp1 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false))
+        val zp1 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, C64Target))
         assertEquals(238, zp1.available())
         assertTrue(50 in zp1.free)
         assertTrue(100 in zp1.free)
@@ -204,7 +209,7 @@ class TestC64Zeropage {
         assertTrue(200 in zp1.free)
         assertTrue(255 in zp1.free)
         assertTrue(199 in zp1.free)
-        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, listOf(50 .. 100, 200..255), false, false))
+        val zp2 = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, listOf(50 .. 100, 200..255), false, false, C64Target))
         assertEquals(139, zp2.available())
         assertFalse(50 in zp2.free)
         assertFalse(100 in zp2.free)
@@ -217,7 +222,7 @@ class TestC64Zeropage {
 
     @Test
     fun testBasicsafeAllocation() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false, C64Target))
         assertEquals(18, zp.available())
 
         assertFailsWith<ZeropageDepletedError> {
@@ -240,7 +245,7 @@ class TestC64Zeropage {
 
     @Test
     fun testFullAllocation() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, C64Target))
         assertEquals(238, zp.available())
         val loc = zp.allocate("", DataType.UWORD, null, errors)
         assertTrue(loc > 3)
@@ -270,7 +275,7 @@ class TestC64Zeropage {
 
     @Test
     fun testEfficientAllocation() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(),  true, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(),  true, false, C64Target))
         assertEquals(18, zp.available())
         assertEquals(0x04, zp.allocate("", DataType.WORD, null, errors))
         assertEquals(0x06, zp.allocate("", DataType.UBYTE, null, errors))
@@ -289,7 +294,7 @@ class TestC64Zeropage {
 
     @Test
     fun testReservedLocations() {
-        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false))
+        val zp = C64Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false, C64Target))
         assertEquals(zp.SCRATCH_REG, zp.SCRATCH_B1+1, "zp _B1 and _REG must be next to each other to create a word")
     }
 }
@@ -299,9 +304,30 @@ class TestC64Zeropage {
 class TestCx16Zeropage {
     @Test
     fun testReservedLocations() {
-        val zp = CX16MachineDefinition.CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false))
+        val zp = CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), false, false, Cx16Target))
         assertEquals(zp.SCRATCH_REG, zp.SCRATCH_B1+1, "zp _B1 and _REG must be next to each other to create a word")
     }
+
+    @Test
+    fun testFreeSpaces() {
+        val zp1 = CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.BASICSAFE, emptyList(), true, false, Cx16Target))
+        assertEquals(88, zp1.available())
+        val zp3 = CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.KERNALSAFE, emptyList(), false, false, Cx16Target))
+        assertEquals(175, zp3.available())
+        val zp4 = CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, Cx16Target))
+        assertEquals(216, zp4.available())
+    }
+
+    @Test
+    fun testReservedSpace() {
+        val zp1 = CX16Zeropage(CompilationOptions(OutputType.RAW, LauncherType.NONE, ZeropageType.FULL, emptyList(), false, false, Cx16Target))
+        assertEquals(216, zp1.available())
+        assertTrue(0x22 in zp1.free)
+        assertTrue(0x80 in zp1.free)
+        assertTrue(0xff in zp1.free)
+        assertFalse(0x02 in zp1.free)
+        assertFalse(0x21 in zp1.free)
+    }
 }
 
 
@@ -401,74 +427,77 @@ class TestPetscii {
 
 
 class TestMemory {
+    private class DummyFunctions: IBuiltinFunctions {
+        override val names: Set<String> = emptySet()
+        override val purefunctionNames: Set<String> = emptySet()
+        override fun constValue(name: String, args: List<Expression>, position: Position, memsizer: IMemSizer): NumericLiteralValue? = null
+        override fun returnType(name: String, args: MutableList<Expression>) = InferredTypes.InferredType.unknown()
+    }
+
+    private class DummyMemsizer: IMemSizer {
+        override fun memorySize(dt: DataType): Int = 0
+    }
+
     @Test
     fun testInValidRamC64_memory_addresses() {
-        CompilationTarget.instance = C64Target
 
         var memexpr = NumericLiteralValue.optimalInteger(0x0000, Position.DUMMY)
         var target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        var scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertTrue(target.isInRegularRAM(scope))
+        val program = Program("test", mutableListOf(), DummyFunctions(), DummyMemsizer())
+        assertTrue(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0x1000, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertTrue(target.isInRegularRAM(scope))
+        assertTrue(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0x9fff, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertTrue(target.isInRegularRAM(scope))
+        assertTrue(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0xc000, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertTrue(target.isInRegularRAM(scope))
+        assertTrue(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0xcfff, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertTrue(target.isInRegularRAM(scope))
+        assertTrue(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testNotInValidRamC64_memory_addresses() {
-        CompilationTarget.instance = C64Target
 
         var memexpr = NumericLiteralValue.optimalInteger(0xa000, Position.DUMMY)
         var target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        var scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertFalse(target.isInRegularRAM(scope))
+        val program = Program("test", mutableListOf(), DummyFunctions(), DummyMemsizer())
+        assertFalse(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0xafff, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertFalse(target.isInRegularRAM(scope))
+        assertFalse(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0xd000, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertFalse(target.isInRegularRAM(scope))
+        assertFalse(C64Target.isInRegularRAM(target, program))
 
         memexpr = NumericLiteralValue.optimalInteger(0xffff, Position.DUMMY)
         target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertFalse(target.isInRegularRAM(scope))
+        assertFalse(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testInValidRamC64_memory_identifiers() {
-        CompilationTarget.instance = C64Target
         var target = createTestProgramForMemoryRefViaVar(0x1000, VarDeclType.VAR)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        val program = Program("test", mutableListOf(), DummyFunctions(), DummyMemsizer())
+
+        assertTrue(C64Target.isInRegularRAM(target, program))
         target = createTestProgramForMemoryRefViaVar(0xd020, VarDeclType.VAR)
-        assertFalse(target.isInRegularRAM(target.definingScope()))
+        assertFalse(C64Target.isInRegularRAM(target, program))
         target = createTestProgramForMemoryRefViaVar(0x1000, VarDeclType.CONST)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        assertTrue(C64Target.isInRegularRAM(target, program))
         target = createTestProgramForMemoryRefViaVar(0xd020, VarDeclType.CONST)
-        assertFalse(target.isInRegularRAM(target.definingScope()))
+        assertFalse(C64Target.isInRegularRAM(target, program))
         target = createTestProgramForMemoryRefViaVar(0x1000, VarDeclType.MEMORY)
-        assertFalse(target.isInRegularRAM(target.definingScope()))
+        assertFalse(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
@@ -477,90 +506,96 @@ class TestMemory {
         val memexpr = IdentifierReference(listOf("address"), Position.DUMMY)
         val target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        module.linkParents(ParentSentinel)
         return target
     }
 
     @Test
     fun testInValidRamC64_memory_expression() {
-        CompilationTarget.instance = C64Target
         val memexpr = PrefixExpression("+", NumericLiteralValue.optimalInteger(0x1000, Position.DUMMY), Position.DUMMY)
         val target = AssignTarget(null, null, DirectMemoryWrite(memexpr, Position.DUMMY), Position.DUMMY)
-        val scope = AnonymousScope(mutableListOf(), Position.DUMMY)
-        assertFalse(target.isInRegularRAM(scope))
+        val program = Program("test", mutableListOf(), DummyFunctions(), DummyMemsizer())
+        assertFalse(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testInValidRamC64_variable() {
-        CompilationTarget.instance = C64Target
         val decl = VarDecl(VarDeclType.VAR, DataType.BYTE, ZeropageWish.DONTCARE, null, "address", null, null, false, false, Position.DUMMY)
         val target = AssignTarget(IdentifierReference(listOf("address"), Position.DUMMY), null, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertTrue(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testInValidRamC64_memmap_variable() {
-        CompilationTarget.instance = C64Target
         val address = 0x1000
         val decl = VarDecl(VarDeclType.MEMORY, DataType.UBYTE, ZeropageWish.DONTCARE, null, "address", null, NumericLiteralValue.optimalInteger(address, Position.DUMMY), false, false, Position.DUMMY)
         val target = AssignTarget(IdentifierReference(listOf("address"), Position.DUMMY), null, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertTrue(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testNotInValidRamC64_memmap_variable() {
-        CompilationTarget.instance = C64Target
         val address = 0xd020
         val decl = VarDecl(VarDeclType.MEMORY, DataType.UBYTE, ZeropageWish.DONTCARE, null, "address", null, NumericLiteralValue.optimalInteger(address, Position.DUMMY), false, false, Position.DUMMY)
         val target = AssignTarget(IdentifierReference(listOf("address"), Position.DUMMY), null, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertFalse(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertFalse(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testInValidRamC64_array() {
-        CompilationTarget.instance = C64Target
         val decl = VarDecl(VarDeclType.VAR, DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, "address", null, null, false, false, Position.DUMMY)
         val arrayindexed = ArrayIndexedExpression(IdentifierReference(listOf("address"), Position.DUMMY), ArrayIndex(NumericLiteralValue.optimalInteger(1, Position.DUMMY), Position.DUMMY), Position.DUMMY)
         val target = AssignTarget(null, arrayindexed, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertTrue(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testInValidRamC64_array_memmapped() {
-        CompilationTarget.instance = C64Target
         val address = 0x1000
         val decl = VarDecl(VarDeclType.MEMORY, DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, "address", null, NumericLiteralValue.optimalInteger(address, Position.DUMMY), false, false, Position.DUMMY)
         val arrayindexed = ArrayIndexedExpression(IdentifierReference(listOf("address"), Position.DUMMY), ArrayIndex(NumericLiteralValue.optimalInteger(1, Position.DUMMY), Position.DUMMY), Position.DUMMY)
         val target = AssignTarget(null, arrayindexed, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertTrue(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertTrue(C64Target.isInRegularRAM(target, program))
     }
 
     @Test
     fun testNotValidRamC64_array_memmapped() {
-        CompilationTarget.instance = C64Target
         val address = 0xe000
         val decl = VarDecl(VarDeclType.MEMORY, DataType.ARRAY_UB, ZeropageWish.DONTCARE, null, "address", null, NumericLiteralValue.optimalInteger(address, Position.DUMMY), false, false, Position.DUMMY)
         val arrayindexed = ArrayIndexedExpression(IdentifierReference(listOf("address"), Position.DUMMY), ArrayIndex(NumericLiteralValue.optimalInteger(1, Position.DUMMY), Position.DUMMY), Position.DUMMY)
         val target = AssignTarget(null, arrayindexed, null, Position.DUMMY)
         val assignment = Assignment(target, NumericLiteralValue.optimalInteger(0, Position.DUMMY), Position.DUMMY)
-        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, mutableListOf(decl, assignment), Position.DUMMY)
-        subroutine.linkParents(ParentSentinel)
-        assertFalse(target.isInRegularRAM(target.definingScope()))
+        val subroutine = Subroutine("test", emptyList(), emptyList(), emptyList(), emptyList(), emptySet(), null, false, false, mutableListOf(decl, assignment), Position.DUMMY)
+        val module = Module("test", mutableListOf(subroutine), Position.DUMMY, false, Path.of(""))
+        val program = Program("test", mutableListOf(module), DummyFunctions(), DummyMemsizer())
+        module.linkParents(ParentSentinel)
+        assertFalse(C64Target.isInRegularRAM(target, program))
     }
 }

compiler/test/arithmetic/Makefile

@@ -0,0 +1,13 @@
+.PHONY: all clean test
+
+all: test
+
+clean:
+	rm -f *.prg *.asm *.vice-*
+
+test: clean
+	p8compile -target cx16 *.p8 >/dev/null
+	for program in *.prg; do \
+		echo "RUNNING:" $$program ; \
+		x16emu -run -prg $$program >/dev/null ; \
+	done

compiler/test/arithmetic/aggregates.p8

@@ -1,5 +1,6 @@
 %import floats
 %import textio
+%import string
 %zeropage basicsafe
 
 main {
@@ -22,10 +23,10 @@ main {
         if length!=5 txt.print("error len1\n")
         length = len(uwarr)
         if length!=5 txt.print("error len2\n")
-        length=strlen(name)
+        length=string.length(name)
         if length!=5 txt.print("error strlen1\n")
         name[3] = 0
-        length=strlen(name)
+        length=string.length(name)
         if length!=3 txt.print("error strlen2\n")
 
         ; MAX

compiler/test/arithmetic/bitshift.p8

@@ -5,72 +5,72 @@
 main {
 
     sub start() {
-        ubyte A
+        ubyte a
 
         txt.print("ubyte shift left\n")
-        A = shiftlb0()
-        txt.print_ubbin(A, true)
+        a = shiftlb0()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb1()
-        txt.print_ubbin(A, true)
+        a = shiftlb1()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb2()
-        txt.print_ubbin(A, true)
+        a = shiftlb2()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb3()
-        txt.print_ubbin(A, true)
+        a = shiftlb3()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb4()
-        txt.print_ubbin(A, true)
+        a = shiftlb4()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb5()
-        txt.print_ubbin(A, true)
+        a = shiftlb5()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb6()
-        txt.print_ubbin(A, true)
+        a = shiftlb6()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb7()
-        txt.print_ubbin(A, true)
+        a = shiftlb7()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb8()
-        txt.print_ubbin(A, true)
+        a = shiftlb8()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftlb9()
-        txt.print_ubbin(A, true)
+        a = shiftlb9()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
         txt.print("enter to continue:\n")
         void c64.CHRIN()
 
         txt.print("ubyte shift right\n")
-        A = shiftrb0()
-        txt.print_ubbin(A, true)
+        a = shiftrb0()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb1()
-        txt.print_ubbin(A, true)
+        a = shiftrb1()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb2()
-        txt.print_ubbin(A, true)
+        a = shiftrb2()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb3()
-        txt.print_ubbin(A, true)
+        a = shiftrb3()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb4()
-        txt.print_ubbin(A, true)
+        a = shiftrb4()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb5()
-        txt.print_ubbin(A, true)
+        a = shiftrb5()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb6()
-        txt.print_ubbin(A, true)
+        a = shiftrb6()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb7()
-        txt.print_ubbin(A, true)
+        a = shiftrb7()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb8()
-        txt.print_ubbin(A, true)
+        a = shiftrb8()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
-        A = shiftrb9()
-        txt.print_ubbin(A, true)
+        a = shiftrb9()
+        txt.print_ubbin(a, true)
         c64.CHROUT('\n')
         txt.print("enter to continue:\n")
         void c64.CHRIN()
@@ -474,10 +474,6 @@ main {
         return (q >> 17)
     }
 
-
-
-
-
     sub shiftrsw0() -> word {
         word q = -12345
         return q >> 0