also optimize BRA+RTS into just BRA

release 12.0

.aiignore

@@ -0,0 +1,17 @@
+# An .aiignore file follows the same syntax as a .gitignore file.
+# .gitignore documentation: https://git-scm.com/docs/gitignore
+
+# you can ignore files
+.DS_Store
+*.log
+*.tmp
+*.bin
+*.prg
+
+# or folders
+dist/
+build/
+out/
+output/
+.gradle/
+.kotlin/

.github/workflows/all-ci.yml

@@ -20,10 +20,10 @@ jobs:
           make -j4
           sudo make install          
 
-      - name: Set up JDK 11
+      - name: Set up JDK 17
         uses: actions/setup-java@v4
         with:
-          java-version: 11
+          java-version: 17
           distribution: temurin
 
       - name: Build and test with Gradle

.idea/libraries/KotlinJavaRuntime.xml

@@ -1,23 +1,23 @@
 <component name="libraryTable">
   <library name="KotlinJavaRuntime" type="repository">
-    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:2.2.20" />
+    <properties maven-id="org.jetbrains.kotlin:kotlin-stdlib-jdk8:2.2.21" />
     <CLASSES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.20/kotlin-stdlib-jdk8-2.2.20.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.20/kotlin-stdlib-2.2.20.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.21/kotlin-stdlib-jdk8-2.2.21.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.21/kotlin-stdlib-2.2.21.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.20/kotlin-stdlib-jdk7-2.2.20.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.21/kotlin-stdlib-jdk7-2.2.21.jar!/" />
     </CLASSES>
     <JAVADOC>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.20/kotlin-stdlib-jdk8-2.2.20-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.20/kotlin-stdlib-2.2.20-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.21/kotlin-stdlib-jdk8-2.2.21-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.21/kotlin-stdlib-2.2.21-javadoc.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-javadoc.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.20/kotlin-stdlib-jdk7-2.2.20-javadoc.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.21/kotlin-stdlib-jdk7-2.2.21-javadoc.jar!/" />
     </JAVADOC>
     <SOURCES>
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.20/kotlin-stdlib-jdk8-2.2.20-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.20/kotlin-stdlib-2.2.20-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk8/2.2.21/kotlin-stdlib-jdk8-2.2.21-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib/2.2.21/kotlin-stdlib-2.2.21-sources.jar!/" />
       <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/annotations/13.0/annotations-13.0-sources.jar!/" />
-      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.20/kotlin-stdlib-jdk7-2.2.20-sources.jar!/" />
+      <root url="jar://$MAVEN_REPOSITORY$/org/jetbrains/kotlin/kotlin-stdlib-jdk7/2.2.21/kotlin-stdlib-jdk7-2.2.21-sources.jar!/" />
     </SOURCES>
   </library>
 </component>

.idea/modules.xml

@@ -15,6 +15,7 @@
       <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://$PROJECT_DIR$/intermediate/intermediate.iml" filepath="$PROJECT_DIR$/intermediate/intermediate.iml" />
+      <module fileurl="file://$PROJECT_DIR$/languageServer/languageServer.iml" filepath="$PROJECT_DIR$/languageServer/languageServer.iml" />
       <module fileurl="file://$PROJECT_DIR$/parser/parser.iml" filepath="$PROJECT_DIR$/parser/parser.iml" />
       <module fileurl="file://$PROJECT_DIR$/.idea/modules/prog8.iml" filepath="$PROJECT_DIR$/.idea/modules/prog8.iml" />
       <module fileurl="file://$PROJECT_DIR$/simpleAst/simpleAst.iml" filepath="$PROJECT_DIR$/simpleAst/simpleAst.iml" />

README.md

@@ -34,8 +34,11 @@ How to get it/build it
   - you can also compile it yourself from source. [Instructions here](https://prog8.readthedocs.io/en/latest/compiling.html).
     Note that if you are not using *gradle* to build it, you might have to perform some manual
     tasks once to make it compile fully. These are explained in the linked instructions. 
-- Alternatively, you can also install the compiler as a package on some linux distros:
+- Alternatively, you can also install the compiler as a package on some linux distros (which will take care of dependencies automatically):
   - Arch (via AUR): [`prog8`](https://aur.archlinux.org/packages/prog8)
+- Finally there's a Homebrew recipe for Mac OS (but also for Linux, and WSL2 on Windows, this also takes care of dependencies automatically):
+  ``brew install prog8``
+
 
 Community
 ---------
@@ -58,9 +61,9 @@ What does Prog8 provide?
 
 - all advantages of a higher level language over having to write assembly code manually
 - programs run very fast because it's compiled to native machine code
-- code often is smaller and faster than equivalent C code compiled with CC65 or even LLVM-MOS
+- compiled code is very small; much smaller than equivalent C code compiled with CC65, and usually runs faster as well
 - modularity, symbol scoping, subroutines. No need for forward declarations.
-- various data types other than just bytes (16-bit words, floats, strings)
+- various data types other than just bytes (16-bit words, long integers, floats, strings)
 - Structs and typed pointers
 - floating point math is supported on certain targets
 - access to most Kernal ROM routines as external subroutine definitions you can call normally
@@ -82,7 +85,7 @@ What does Prog8 provide?
 - supports the sixteen 'virtual' 16-bit registers R0 - R15 from the Commander X16 (also available on other targets)
 - encode strings and characters into petscii or screencodes or even other encodings
 - Automatic ROM/RAM bank switching on certain compiler targets when calling routines in other banks
-- 50 Kb of available program RAM size on the C64 by default; because Basic ROM is banked out altogether 
+- 50 Kb of available program RAM size on the C64 by default (41 Kb on the C128) because Basic ROM is banked out by default 
 
 *Rapid edit-compile-run-debug cycle:*
 
@@ -193,3 +196,12 @@ For instance here's a well known space ship animated in 3D with hidden line remo
 in the CommanderX16 emulator:
 
 ![cobra3d](docs/source/_static/cobra3d.png)
+
+
+Performance comparison with various C compilers
+-----------------------------------------------
+
+Here is a performance comparison with various C compilers for the 6502/C64 so you can get
+an idea of how Prog8 stacks up.
+
+[comparison](benchmark-c/benchmark.md)

benchmark-c/benchmark.md

@@ -0,0 +1,26 @@
+# C-Bench-64 comparison
+
+Several benchmarks of https://thred.github.io/c-bench-64/
+have been ported to equivalent Prog8 code and the benchmark results have been penciled in in the graphs below.
+
+Maybe one day I'll try to integrate the prog8 data properly but their benchmark site is comparing C compilers, which Prog8 clearly is not.
+
+However conclusions so far (note: these are micro benchmarks so interpret the results as you will!)
+
+* Prog8 program size is consistently the smallest by a fair bit.
+* Prog8 execution speed places it more or less in the middle of the stack.
+
+Measured with Prog8 V12.0 .
+
+
+![crc8](result-crc8.png)
+
+![crc16](result-crc16.png)
+
+![crc32](result-crc32.png)
+
+![pow](result-pow.png)
+
+![sieve](result-sieve.png)
+
+![sieve-bit](result-sieve-bit.png)

benchmark-c/crc16.p8

@@ -0,0 +1,71 @@
+%import textio
+%import floats
+
+; note: Prog8 has a complete CRC16 routine in its library: math.crc16
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+test {
+
+    const uword EXPECTED = $ffd0
+    uword crc_result
+
+    sub benchmark_name()
+    {
+        txt.print("crc16.p8\n")
+        txt.print("Calculates the CRC16 of the C64 Kernal\n")
+    }
+
+    sub benchmark()
+    {
+        crc_result = CRC16($e000, $2000)
+    }
+    
+    sub benchmark_check() -> bool
+    {
+        txt.print("CRC=")
+        txt.print_uwhex(crc_result, true)
+
+        if crc_result == EXPECTED
+        {
+            txt.print(" [OK]")
+            return false
+        }
+
+        txt.print(" [FAIL] - expected ")
+        txt.print_uwhex(EXPECTED, true)
+        return true
+    }
+    
+    sub CRC16(^^ubyte data, uword length) -> uword {
+        ; CRC-16/XMODEM
+        uword crc
+
+        repeat length
+        {
+            crc ^= mkword(@(data),0)     ; currently there's no easy way to affect only the MSB of the variable
+
+            repeat 8
+            {
+                crc <<=1
+                if_cs
+                    crc ^= $1021
+            }
+            data++
+        }
+        return crc
+    }
+
+}

benchmark-c/crc32.p8

@@ -0,0 +1,72 @@
+%import textio
+%import floats
+
+; note: Prog8 has a complete CRC32 routine in its library: math.crc32
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+test {
+
+    const long EXPECTED = $e1fa84c6
+    long crc_result
+
+    sub benchmark_name()
+    {
+        txt.print("crc32.p8\n")
+        txt.print("Calculates the CRC32 of the C64 Kernal\n")
+    }
+
+    sub benchmark()
+    {
+        crc_result = CRC32($e000, $2000)
+    }
+    
+    sub benchmark_check() -> bool
+    {
+        txt.print("CRC=")
+        txt.print_ulhex(crc_result, true)
+
+        if crc_result == EXPECTED
+        {
+            txt.print(" [OK]")
+            return false
+        }
+
+        txt.print(" [FAIL] - expected ")
+        txt.print_ulhex(EXPECTED, true)
+        return true
+    }
+    
+    sub CRC32(^^ubyte data, uword length) -> long {
+        ; CRC-32/CKSUM
+        long crc
+
+        repeat length
+        {
+            crc ^= (@(data) as long)<<24     ; currently there's no easy way to affect only the MSB of the variable
+
+            repeat 8
+            {
+                crc <<=1
+                if_cs
+                    crc ^= $04c11db7
+            }
+            data++
+        }
+        crc ^= $ffffffff
+        return crc
+    }
+
+}

benchmark-c/crc8.p8

@@ -0,0 +1,72 @@
+%import textio
+%import floats
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+
+test {
+
+    sub benchmark_name()
+    {
+        txt.print("crc8.p8\n")
+        txt.print("Calculates the CRC8 of the C64 Kernal\n")
+    }
+
+    sub benchmark()
+    {
+        crc_result = CRC8($e000, $2000)
+    }
+    
+    sub benchmark_check() -> bool
+    {
+        txt.print("CRC=")
+        txt.print_ubhex(crc_result, true)
+
+        if crc_result == EXPECTED
+        {
+            txt.print(" [OK]")
+            return false
+        }
+
+        txt.print(" [FAIL] - expected ")
+        txt.print_ubhex(EXPECTED, true)
+        return true
+    }
+    
+
+    const ubyte EXPECTED = $a2
+    ubyte crc_result
+
+    sub CRC8(^^ubyte data, uword length) -> ubyte
+    {
+        ; CRC-8/GSM-A
+        ubyte crc
+
+        repeat length
+        {
+            crc ^= @(data)
+
+            repeat 8
+            {
+                if (crc & $80) != 0
+                    crc = (crc << 1) ^ $1d
+                else
+                    crc <<= 1
+            }
+            data++
+        }
+        return crc
+    }
+}

benchmark-c/pow.p8

@@ -0,0 +1,93 @@
+%import textio
+%import floats
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+test {
+
+    const ubyte N_ITER = 10
+    const ubyte SIZE = 32
+    float[SIZE] array
+
+    const float expected = 3614007361536.000000
+    const float epsilon = 10000000
+    float res
+
+    sub benchmark_name()
+    {
+        txt.print("pow.p8\n")
+        txt.print("Calculates floating point exponential (")
+        txt.print_uw(N_ITER)
+        txt.print(" iterations)\n")
+    }
+
+    sub benchmark()
+    {
+        ubyte i,j
+        res = 0
+        
+        for j in 0 to SIZE-1 {
+            array[j]=0
+        }
+
+        for i in 0 to N_ITER-1 {
+            for j in 0 to SIZE-1 {
+                array[j] += testpow(2.5 / ((i + 1) as float), j)
+            }
+        }
+        
+        for j in 0 to SIZE-1 {
+            res += array[j]
+        }
+    }
+
+    sub testpow(float x, ubyte y) -> float
+    {
+        float tmp = x
+    
+        if y == 0
+            return 1
+    
+        repeat y-1
+            tmp *= x
+    
+        return tmp
+    }
+
+
+    sub benchmark_check() -> bool
+    {
+        txt.print("res      = ")
+        txt.print_f(res)
+        float diff = expected - res;
+        txt.print("\nexpected = ")
+        txt.print_f(expected)
+        txt.print("\nepsilon  = ")
+        txt.print_f(epsilon)
+        txt.print("\ndiff     = ")
+        txt.print_f(diff)
+
+        if (diff < epsilon and diff > -epsilon)
+        {
+            txt.print(" [OK]\n")
+            return false
+        }
+
+        txt.print("[FAIL]\n")
+        return true
+    }
+    
+
+}

benchmark-c/sieve.p8

@@ -0,0 +1,85 @@
+%import textio
+%import floats
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+
+test {
+    const ubyte N_ITER = 10
+    const uword SIZE = 8191
+    const uword EXPECTED = 1900
+    uword prime_count
+    ^^bool @zp flags = memory("flags", SIZE, 0)
+
+    sub benchmark_name()
+    {
+        txt.print("sieve.c\n")
+        txt.print("Calculates the primes from 1 to ")
+        txt.print_uw(SIZE * 2 + 2)
+        txt.print(" (")
+        txt.print_ub(N_ITER)
+        txt.print(" iterations)\n")
+    }
+
+    sub benchmark()
+    {
+        repeat N_ITER
+            prime_count = sieve(SIZE)
+    }
+    
+    sub benchmark_check() -> bool
+    {
+        txt.print("count=")
+        txt.print_uw(prime_count)
+
+        if prime_count == EXPECTED
+        {
+            txt.print(" [OK]")
+            return false
+        }
+
+        txt.print(" [FAIL] - expected ")
+        txt.print_uw(EXPECTED)
+        return true
+    }
+
+    
+    sub sieve(uword size) -> uword 
+    {
+        uword i, prime, k
+        uword count = 1
+
+        for i in 0 to size-1
+            flags[i] = true
+
+        for i in 0 to size-1 
+        {
+            if flags[i]
+            {
+                prime = i + i + 3
+                k = i + prime
+                while k < size
+                {
+                    flags[k] = false
+                    k += prime
+                }
+                count++
+            }
+        }
+    
+        return count
+    }
+
+}

benchmark-c/sieve_bit.p8

@@ -0,0 +1,107 @@
+%import textio
+%import floats
+
+
+main {
+    sub start() {
+        txt.lowercase()
+        test.benchmark_name()
+        cbm.SETTIM(0,0,0)
+        test.benchmark()
+        txt.print_f(floats.time() / 60)
+        txt.print(" seconds\n")
+        void test.benchmark_check()
+        repeat {}
+    }    
+}
+
+
+test {
+    const ubyte N_ITER = 4
+    const uword SIZE = 16000
+    const uword EXPECTED = 3432
+    uword prime_count
+    ^^ubyte @zp flags = memory("flags", SIZE/8+1, 0)
+
+    sub benchmark_name()
+    {
+        txt.print("sieve_bit.p8\n")
+        txt.print("Calculates the primes from 1 to ")
+        txt.print_uw(SIZE * 2 + 2)
+        txt.print(" (")
+        txt.print_ub(N_ITER)
+        txt.print(" iterations)\n")
+    }
+
+    sub benchmark()
+    {
+        repeat N_ITER
+            prime_count = sieve(SIZE)
+    }
+    
+    sub benchmark_check() -> bool
+    {
+        txt.print("count=")
+        txt.print_uw(prime_count)
+
+        if prime_count == EXPECTED
+        {
+            txt.print(" [OK]")
+            return false
+        }
+
+        txt.print(" [FAIL] - expected ")
+        txt.print_uw(EXPECTED)
+        return true
+    }
+
+    ubyte[] bitv = [
+        $01,    
+        $02,
+        $04,
+        $08,
+        $10,
+        $20,
+        $40,
+        $80
+    ]
+
+    sub check_flag(uword idx) -> bool
+    {
+        return flags[idx / 8] & bitv[lsb(idx) & 7] != 0
+    }
+
+    sub clear_flag(uword idx)
+    {
+        flags[idx / 8] &= ~(bitv[lsb(idx) & 7])
+    }
+
+    
+    sub sieve(uword size) -> uword 
+    {
+        uword i, prime, k
+        uword count=1
+    
+        for i in 0 to (size / 8)
+            flags[i] = $ff
+
+        for i in 0 to size-1
+        {
+            if check_flag(i)
+            {
+                prime = i + i + 3
+                k = i + prime;
+                while k < size
+                {
+                    clear_flag(k)
+                    k += prime
+                }
+                count++
+            }
+        }
+
+   
+        return count
+    }
+
+}

benchmark-program/b_btree.p8

@@ -0,0 +1,234 @@
+; Binary Search Tree.
+; It's a simple implementation for test/demonstration purposes of the pointer support;
+; no balancing is done and memory is not freed when elements are removed.
+
+%import textio
+
+btree {
+
+    sub benchmark(uword max_time) -> uword {
+        txt.nl()
+        cbm.SETTIM(0,0,0)
+        uword score
+        while cbm.RDTIM16() < max_time {
+            bench_operations()
+            txt.chrout('.')
+            score++
+        }
+        txt.nl()
+        return score
+    }
+
+    sub bench_operations() {
+        arena.freeall()
+        btree.root = 0
+
+        for cx16.r0 in [321, 719, 194, 550, 187, 203, 520, 562, 221, 676, 97, 852, 273, 326, 589, 606, 275, 794, 63, 716]
+            btree.add(cx16.r0)
+
+        cx16.r0L = btree.size()
+        btree.process_tree_inorder()
+        btree.process_tree_preorder()
+
+        void btree.contains(203)
+        void btree.contains(204)
+        void btree.contains(605)
+        void btree.contains(606)
+
+        btree.remove(9999)
+        btree.remove(97)
+        btree.remove(187)
+        btree.remove(203)
+        btree.remove(275)
+        btree.remove(321)
+        btree.remove(520)
+        btree.remove(562)
+        btree.remove(606)
+        btree.remove(719)
+        btree.remove(794)
+
+        cx16.r0L = btree.size()
+        btree.process_tree_inorder()
+        btree.process_tree_preorder()
+    }
+
+    struct Node {
+        ^^Node left
+        ^^Node right
+        uword value
+    }
+
+    ^^Node root = 0
+
+    sub add(uword value) {
+        ^^Node node = arena.alloc(sizeof(Node))
+        node.value = value
+        node.left = node.right = 0
+
+        if root==0
+            root=node
+        else {
+            ^^Node parent = root
+            repeat {
+                if parent.value >= value {
+                    if parent.left!=0
+                        parent = parent.left
+                    else {
+                        parent.left = node
+                        return
+                    }
+                } else {
+                    if parent.right!=0
+                        parent = parent.right
+                    else {
+                        parent.right = node
+                        return
+                    }
+                }
+            }
+        }
+    }
+
+    sub contains(uword value) -> bool {
+        ^^Node r = root
+        while r!=0 {
+            if r.value==value
+                return true
+            if r.value>value
+                r = r.left
+            else
+                r = r.right
+        }
+        return false
+    }
+
+    sub size() -> ubyte {
+        ubyte count
+
+        if root!=0
+            count_node(root)
+
+        return count
+
+        sub count_node(^^Node r) {
+            count++
+            if r.left!=0 {
+                sys.pushw(r)
+                count_node(r.left)
+                r = sys.popw()
+            }
+            if r.right!=0 {
+                sys.pushw(r)
+                count_node(r.right)
+                r = sys.popw()
+            }
+        }
+    }
+
+    sub remove(uword value) {
+        ; note: we don't deallocate the memory from the node, for simplicity sake
+        ^^Node n = root
+        ^^Node parent = 0
+        while n!=0 {
+            if n.value==value {
+                if n.left==0
+                    replacechild(parent, n, n.right)
+                else if n.right==0
+                    replacechild(parent, n, n.left)
+                else {
+                    ; Both left & right subtrees are present.
+                    ; N = node to delete.
+                    ;    Find N's successor S. (N's right subtree's minimum element)
+                    ;    Attach N's left subtree to S.left (S doesn't have a left child)
+                    ;    Attach N's right subtree to Parent in place of N.
+                    ^^Node successor = find_successor(n)
+                    successor.left = n.left
+                    replacechild(parent, n, n.right)
+                }
+                return
+            }
+            parent = n
+            if n.value>value
+                n = n.left
+            else
+                n = n.right
+        }
+
+        sub find_successor(^^Node p) -> ^^Node {
+            ^^Node succ = p
+            p = p.right
+            while p!=0 {
+                succ = p
+                p = p.left
+            }
+            return succ
+        }
+
+        sub replacechild(^^Node p, ^^Node child, ^^Node newchild) {
+            if p.left==child
+                p.left = newchild
+            else
+                p.right = newchild
+        }
+    }
+
+
+    sub process_tree_inorder() {
+        if root!=0
+            process_tree(root)
+
+        sub process_tree(^^Node r) {
+            if r.left!=0 {
+                sys.pushw(r)
+                process_tree(r.left)
+                r = sys.popw()
+            }
+            cx16.r0 = r.value
+            if r.right!=0 {
+                sys.pushw(r)
+                process_tree(r.right)
+                r = sys.popw()
+            }
+        }
+    }
+
+
+    sub process_tree_preorder() {
+        if root!=0
+            process_tree(root,0)
+
+        sub process_tree(^^Node r, ubyte depth) {
+            cx16.r0 = r.value
+            if r.left!=0 {
+                sys.pushw(r)
+                sys.push(depth)
+                process_tree(r.left, depth+1)
+                depth = sys.pop()
+                r = sys.popw()
+            }
+            if r.right!=0 {
+                sys.pushw(r)
+                sys.push(depth)
+                process_tree(r.right, depth+1)
+                depth = sys.pop()
+                r = sys.popw()
+            }
+        }
+    }
+}
+
+
+arena {
+    ; extremely trivial arena allocator (that never frees)
+    uword buffer = memory("arena", 2000, 0)
+    uword next = buffer
+
+    sub alloc(ubyte size) -> uword {
+        defer next += size
+        return next
+    }
+
+    sub freeall() {
+        next = buffer
+    }
+}

benchmark-program/b_textelite.p8

@@ -472,7 +472,7 @@ elite_galaxy {
             generate_next_planet()
             textelite.num_commands++
         }
-        elite_planet.name = make_current_planet_name()
+        void strings.copy(make_current_planet_name(), elite_planet.name)
         init_market_for_planet()
     }
 
@@ -494,7 +494,7 @@ elite_galaxy {
         ubyte pi
         for pi in 0 to 255 {
             generate_next_planet()
-            elite_planet.name = make_current_planet_name()
+            void strings.copy(make_current_planet_name(), elite_planet.name)
             if elite_util.prefix_matches(nameptr, elite_planet.name) {
                 ubyte distance = elite_planet.distance(x, y)
                 if distance < current_distance {
@@ -532,7 +532,7 @@ elite_galaxy {
 ;                else
 ;                    txt.chrout('-')
 ;                txt.spc()
-                elite_planet.name = make_current_planet_name()
+                void strings.copy(make_current_planet_name(), elite_planet.name)
                 elite_planet.display(true, distance)
             }
             pn++
@@ -548,7 +548,7 @@ elite_galaxy {
         str current_name = "        "       ; 8 max
         ubyte pn = 0
 
-        current_name = elite_planet.name
+        void strings.copy(elite_planet.name, current_name)
         init(number)
 ;        txt.clear_screen()
 ;        txt.print("Galaxy #")
@@ -569,7 +569,7 @@ elite_galaxy {
             generate_next_planet()
             ubyte distance = elite_planet.distance(px, py)
             if distance <= max_distance {
-                elite_planet.name = make_current_planet_name()
+                void strings.copy(make_current_planet_name(), elite_planet.name)
                 elite_planet.name[0] = strings.upperchar(elite_planet.name[0])
                 uword tx = elite_planet.x
                 uword ty = elite_planet.y

benchmark-program/benchmark.p8

@@ -16,6 +16,7 @@
 %import b_textelite
 %import b_maze
 %import b_sprites
+%import b_btree
 
 %zeropage basicsafe
 %option no_sysinit
@@ -34,7 +35,7 @@ main {
         txt.color2(1, 6)
         txt.clear_screen()
 
-        txt.print("\n\n\n  prog8 compiler benchmark tests.\n")
+        txt.print("\n\n\n  prog8 compiler benchmark tests.\n\n  benchmark version v12\n\n")
         sys.wait(60)
 
         benchmark_number = 0
@@ -64,13 +65,9 @@ main {
         benchmark_number++
 
         announce_benchmark("circles with gfx_lores")
-        benchmark_score[benchmark_number]  = circles.draw(false, 300)
+        benchmark_score[benchmark_number]  = circles.draw(false, 400)
         benchmark_number++
 
-;        announce_benchmark("circles with kernal")
-;        benchmark_score[benchmark_number]  = circles.draw(true, 300)
-;        benchmark_number++
-
         announce_benchmark("text-elite")
         benchmark_score[benchmark_number]  = textelite.bench(120)
         benchmark_number++
@@ -79,28 +76,32 @@ main {
         benchmark_score[benchmark_number]  = animsprites.benchmark(300)
         benchmark_number++
 
+        announce_benchmark("btree-struct-pointers")
+        benchmark_score[benchmark_number]  = btree.benchmark(200)
+        benchmark_number++
+
         benchmark_names[benchmark_number] = 0
         benchmark_score[benchmark_number] = 0
 
         cx16.set_screen_mode(3)
         txt.uppercase()
         txt.color2(1, 6)
-        uword final_score
+        uword total_score
         benchmark_number = 0
-        txt.print("\nscore benchmark\n\n")
+        txt.print("\nscore benchmark (v12)\n\n")
         do {
             txt.spc()
             txt.print_uw(benchmark_score[benchmark_number])
             txt.column(6)
             txt.print(benchmark_names[benchmark_number])
-            final_score += benchmark_score[benchmark_number]
+            total_score += benchmark_score[benchmark_number]
             txt.nl()
             benchmark_number++
         } until benchmark_names[benchmark_number]==0
 
-        txt.print("\n\nfinal score : ")
-        txt.print_uw(final_score)
-        txt.nl()
+        txt.print("\n\ntotal score : ")
+        txt.print_uw(total_score)
+        txt.print(" (higher=better)\n")
 
         sub announce_benchmark(str name) {
             benchmark_names[benchmark_number] = name

build.gradle.kts

@@ -5,7 +5,7 @@ import org.jetbrains.kotlin.gradle.dsl.KotlinVersion
 
 
 plugins {
-    kotlin("jvm") version "2.2.20"
+    kotlin("jvm") version "2.2.21"
 }
 
 allprojects {

codeCore/src/prog8/code/Globals.kt

@@ -5,9 +5,10 @@ import java.nio.file.Path
 import kotlin.io.path.absolute
 
 
-// the automatically generated module where all string literals are interned to:
 const val INTERNED_STRINGS_MODULENAME = "prog8_interned_strings"
 
+val PROG8_CONTAINER_MODULES = arrayOf(INTERNED_STRINGS_MODULENAME)      // option to add more if needed one day
+
 // all automatically generated labels everywhere need to have the same label name prefix:
 const val GENERATED_LABEL_PREFIX = "p8_label_gen_"
 

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

@@ -37,6 +37,7 @@ class FSignature(val pure: Boolean,      // does it have side effects?
         val returns: ReturnConvention = when (returnType) {
             BaseDataType.UBYTE, BaseDataType.BYTE -> ReturnConvention(returnType, RegisterOrPair.A)
             BaseDataType.UWORD, BaseDataType.WORD -> ReturnConvention(returnType, RegisterOrPair.AY)
+            BaseDataType.LONG -> ReturnConvention(returnType, RegisterOrPair.R14R15_32)
             BaseDataType.FLOAT -> ReturnConvention(returnType, RegisterOrPair.FAC1)
             in IterableDatatypes -> ReturnConvention(returnType!!, RegisterOrPair.AY)
             null -> ReturnConvention(null, null)
@@ -45,6 +46,7 @@ class FSignature(val pure: Boolean,      // does it have side effects?
                 when (val paramType = actualParamTypes.first()) {
                     BaseDataType.UBYTE, BaseDataType.BYTE -> ReturnConvention(paramType, RegisterOrPair.A)
                     BaseDataType.UWORD, BaseDataType.WORD -> ReturnConvention(paramType, RegisterOrPair.AY)
+                    BaseDataType.LONG -> ReturnConvention(returnType, RegisterOrPair.R14R15_32)
                     BaseDataType.FLOAT -> ReturnConvention(paramType, RegisterOrPair.FAC1)
                     in IterableDatatypes -> ReturnConvention(paramType, RegisterOrPair.AY)
                     else -> ReturnConvention(paramType, null)
@@ -61,6 +63,7 @@ class FSignature(val pure: Boolean,      // does it have side effects?
                 val paramConv = when(val paramType = actualParamTypes[0]) {
                     BaseDataType.UBYTE, BaseDataType.BYTE -> ParamConvention(paramType, RegisterOrPair.A, false)
                     BaseDataType.UWORD, BaseDataType.WORD -> ParamConvention(paramType, RegisterOrPair.AY, false)
+                    BaseDataType.LONG -> ParamConvention(paramType, RegisterOrPair.R14R15_32, false)
                     BaseDataType.FLOAT -> ParamConvention(paramType, RegisterOrPair.AY, false)      // NOTE: for builtin functions, floating point arguments are passed by reference (so you get a pointer in AY)
                     in IterableDatatypes -> ParamConvention(paramType, RegisterOrPair.AY, false)
                     else -> ParamConvention(paramType, null, false)
@@ -88,11 +91,11 @@ class FSignature(val pure: Boolean,      // does it have side effects?
 val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "setlsb"  to FSignature(false, null, FParam("variable", BaseDataType.WORD, BaseDataType.UWORD), FParam("value", BaseDataType.BYTE, BaseDataType.UBYTE)),
     "setmsb"  to FSignature(false, null, FParam("variable", BaseDataType.WORD, BaseDataType.UWORD), FParam("value", BaseDataType.BYTE, BaseDataType.UBYTE)),
-    "rol"     to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD)),
-    "ror"     to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD)),
-    "rol2"    to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD)),
-    "ror2"    to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD)),
-    "cmp"     to FSignature(false, null, FParam("value1", *IntegerDatatypes), FParam("value2", *NumericDatatypes)),  // cmp returns a status in the carry flag, but not a proper return value
+    "rol"     to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD, BaseDataType.LONG)),
+    "ror"     to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD, BaseDataType.LONG)),
+    "rol2"    to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD, BaseDataType.LONG)),
+    "ror2"    to FSignature(false, null, FParam("item", BaseDataType.UBYTE, BaseDataType.UWORD, BaseDataType.LONG)),
+    "cmp"     to FSignature(false, null, FParam("value1", *IntegerDatatypes), FParam("value2", *NumericDatatypes)),  // cmp returns result in the cpu status flags, but not asa proper return value
     "prog8_lib_stringcompare"     to FSignature(true, BaseDataType.BYTE, FParam("str1", BaseDataType.STR), FParam("str2", BaseDataType.STR)),
     "prog8_lib_square_byte"       to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.BYTE, BaseDataType.UBYTE)),
     "prog8_lib_square_word"       to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.WORD, BaseDataType.UWORD)),
@@ -100,46 +103,56 @@ val BuiltinFunctions: Map<String, FSignature> = mapOf(
     "abs"           to FSignature(true, null, FParam("value", *NumericDatatypes)),
     "abs__byte"     to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.BYTE)),
     "abs__word"     to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.WORD)),
+    "abs__long"     to FSignature(true, BaseDataType.LONG, FParam("value", BaseDataType.LONG)),
     "abs__float"    to FSignature(true, BaseDataType.FLOAT, FParam("value", BaseDataType.FLOAT)),
     "len"           to FSignature(true, BaseDataType.UWORD, FParam("values", *IterableDatatypes)),
     "sizeof"        to FSignature(true, BaseDataType.UBYTE, FParam("object", *(BaseDataType.entries - BaseDataType.STRUCT_INSTANCE).toTypedArray())),
+    "offsetof"      to FSignature(true, BaseDataType.UBYTE, FParam("field", BaseDataType.UBYTE)),
     "sgn"           to FSignature(true, BaseDataType.BYTE, FParam("value", *NumericDatatypes)),
     "sqrt"          to FSignature(true, null, FParam("value", *NumericDatatypes)),
     "sqrt__ubyte"   to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.UBYTE)),
     "sqrt__uword"   to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.UWORD)),
+    "sqrt__long"    to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.LONG)),
     "sqrt__float"   to FSignature(true, BaseDataType.FLOAT, FParam("value", BaseDataType.FLOAT)),
     "divmod"        to FSignature(false, null, FParam("dividend", BaseDataType.UBYTE, BaseDataType.UWORD), FParam("divisor", BaseDataType.UBYTE, BaseDataType.UWORD), FParam("quotient", BaseDataType.UBYTE, BaseDataType.UWORD), FParam("remainder", BaseDataType.UBYTE, BaseDataType.UWORD)),
     "divmod__ubyte" to FSignature(false, null, FParam("dividend", BaseDataType.UBYTE), FParam("divisor", BaseDataType.UBYTE), FParam("quotient", BaseDataType.UBYTE), FParam("remainder", BaseDataType.UBYTE)),
     "divmod__uword" to FSignature(false, null, FParam("dividend", BaseDataType.UWORD), FParam("divisor", BaseDataType.UWORD), FParam("quotient", BaseDataType.UWORD), FParam("remainder", BaseDataType.UWORD)),
     "lsb"           to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)),
-    "lsw"           to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)),
+    "lsb__long"     to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.LONG)),
     "msb"           to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)),
+    "msb__long"     to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.LONG)),
+    "lsw"           to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)),
     "msw"           to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)),
     "mkword"        to FSignature(true, BaseDataType.UWORD, FParam("msb", BaseDataType.UBYTE), FParam("lsb", BaseDataType.UBYTE)),
+    "mklong"        to FSignature(true, BaseDataType.LONG, FParam("msb", BaseDataType.UBYTE), FParam("b2", BaseDataType.UBYTE), FParam("b1", BaseDataType.UBYTE), FParam("lsb", BaseDataType.UBYTE)),
+    "mklong2"       to FSignature(true, BaseDataType.LONG, FParam("msw", BaseDataType.UWORD), FParam("lsw", BaseDataType.UWORD)),
     "clamp"         to FSignature(true, null, FParam("value", BaseDataType.BYTE), FParam("minimum", BaseDataType.BYTE), FParam("maximum", BaseDataType.BYTE)),
     "clamp__byte"   to FSignature(true, BaseDataType.BYTE, FParam("value", BaseDataType.BYTE), FParam("minimum", BaseDataType.BYTE), FParam("maximum", BaseDataType.BYTE)),
     "clamp__ubyte"  to FSignature(true, BaseDataType.UBYTE, FParam("value", BaseDataType.UBYTE), FParam("minimum", BaseDataType.UBYTE), FParam("maximum", BaseDataType.UBYTE)),
     "clamp__word"   to FSignature(true, BaseDataType.WORD, FParam("value", BaseDataType.WORD), FParam("minimum", BaseDataType.WORD), FParam("maximum", BaseDataType.WORD)),
     "clamp__uword"  to FSignature(true, BaseDataType.UWORD, FParam("value", BaseDataType.UWORD), FParam("minimum", BaseDataType.UWORD), FParam("maximum", BaseDataType.UWORD)),
+    "clamp__long"   to FSignature(true, BaseDataType.LONG, FParam("value", BaseDataType.LONG), FParam("minimum", BaseDataType.LONG), FParam("maximum", BaseDataType.LONG)),
     "min"           to FSignature(true, null, FParam("val1", BaseDataType.BYTE), FParam("val2", BaseDataType.BYTE)),
     "min__byte"     to FSignature(true, BaseDataType.BYTE, FParam("val1", BaseDataType.BYTE), FParam("val2", BaseDataType.BYTE)),
     "min__ubyte"    to FSignature(true, BaseDataType.UBYTE, FParam("val1", BaseDataType.UBYTE), FParam("val2", BaseDataType.UBYTE)),
     "min__word"     to FSignature(true, BaseDataType.WORD, FParam("val1", BaseDataType.WORD), FParam("val2", BaseDataType.WORD)),
     "min__uword"    to FSignature(true, BaseDataType.UWORD, FParam("val1", BaseDataType.UWORD), FParam("val2", BaseDataType.UWORD)),
+    "min__long"     to FSignature(true, BaseDataType.LONG, FParam("val1", BaseDataType.LONG), FParam("val2", BaseDataType.LONG)),
     "max"           to FSignature(true, null, FParam("val1", BaseDataType.BYTE), FParam("val2", BaseDataType.BYTE)),
     "max__byte"     to FSignature(true, BaseDataType.BYTE, FParam("val1", BaseDataType.BYTE), FParam("val2", BaseDataType.BYTE)),
     "max__ubyte"    to FSignature(true, BaseDataType.UBYTE, FParam("val1", BaseDataType.UBYTE), FParam("val2", BaseDataType.UBYTE)),
     "max__word"     to FSignature(true, BaseDataType.WORD, FParam("val1", BaseDataType.WORD), FParam("val2", BaseDataType.WORD)),
     "max__uword"    to FSignature(true, BaseDataType.UWORD, FParam("val1", BaseDataType.UWORD), FParam("val2", BaseDataType.UWORD)),
+    "max__long"     to FSignature(true, BaseDataType.LONG, FParam("val1", BaseDataType.LONG), FParam("val2", BaseDataType.LONG)),
     "peek"          to FSignature(true, BaseDataType.UBYTE, FParam("address", BaseDataType.UWORD)),
     "peekbool"      to FSignature(true, BaseDataType.BOOL, FParam("address", BaseDataType.UWORD)),
     "peekw"         to FSignature(true, BaseDataType.UWORD, FParam("address", BaseDataType.UWORD)),
     "peekl"         to FSignature(true, BaseDataType.LONG, FParam("address", BaseDataType.UWORD)),
     "peekf"         to FSignature(true, BaseDataType.FLOAT, FParam("address", BaseDataType.UWORD)),
-    "poke"          to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.UBYTE)),
+    "poke"          to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.UBYTE, BaseDataType.BYTE)),
     "pokebool"      to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.BOOL)),
     "pokebowl"      to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.BOOL)),
-    "pokew"         to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.UWORD)),
+    "pokew"         to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.UWORD, BaseDataType.WORD)),
     "pokel"         to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.LONG)),
     "pokef"         to FSignature(false, null, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.FLOAT)),
     "pokemon"       to FSignature(false, BaseDataType.UBYTE, FParam("address", BaseDataType.UWORD), FParam("value", BaseDataType.UBYTE)),

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

@@ -13,6 +13,7 @@ class CompilationOptions(val output: OutputType,
                          val noSysInit: Boolean,
                          val romable: Boolean,
                          val compTarget: ICompilationTarget,
+                         val compilerVersion: String,
                          // these are set later, based on command line arguments or options in the source code:
                          var loadAddress: UInt,
                          var memtopAddress: UInt,
@@ -28,7 +29,6 @@ class CompilationOptions(val output: OutputType,
                          var varsGolden: Boolean = false,
                          var slabsHighBank: Int? = null,
                          var slabsGolden: Boolean = false,
-                         var dontSplitWordArrays: Boolean = false,
                          var breakpointCpuInstruction: String? = null,
                          var ignoreFootguns: Boolean = false,
                          var outputDir: Path = Path(""),

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

@@ -13,7 +13,7 @@ fun Number.toHex(): String {
     //  256..65536 -> "$0100".."$ffff"
     //  larger -> "$12345678"
     // negative values are prefixed with '-'.
-    val integer = this.toInt()
+    val integer = this.toLong()
     if(integer<0)
         return '-' + abs(integer).toHex()
     return when (integer) {

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

@@ -50,8 +50,10 @@ enum class BaseDataType {
 val BaseDataType.isByte get() = this in arrayOf(BaseDataType.UBYTE, BaseDataType.BYTE)
 val BaseDataType.isByteOrBool get() = this in arrayOf(BaseDataType.UBYTE, BaseDataType.BYTE, BaseDataType.BOOL)
 val BaseDataType.isWord get() = this in arrayOf(BaseDataType.UWORD, BaseDataType.WORD)
+val BaseDataType.isLong get() = this == BaseDataType.LONG
 val BaseDataType.isInteger get() = this in arrayOf(BaseDataType.UBYTE, BaseDataType.BYTE, BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG)
 val BaseDataType.isIntegerOrBool get() = this in arrayOf(BaseDataType.UBYTE, BaseDataType.BYTE, BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.LONG, BaseDataType.BOOL)
+val BaseDataType.isWordOrByteOrBool get() = this in arrayOf(BaseDataType.UBYTE, BaseDataType.BYTE, BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.BOOL)
 val BaseDataType.isNumeric get() = this == BaseDataType.FLOAT || this.isInteger
 val BaseDataType.isNumericOrBool get() = this == BaseDataType.BOOL || this.isNumeric
 val BaseDataType.isSigned get() = this in arrayOf(BaseDataType.BYTE, BaseDataType.WORD, BaseDataType.LONG, BaseDataType.FLOAT)
@@ -143,7 +145,7 @@ class DataType private constructor(val base: BaseDataType, val sub: BaseDataType
             return if(splitwordarray && actualElementDt.isWord)
                 DataType(BaseDataType.ARRAY_SPLITW, actualElementDt, null)
             else {
-                if(actualElementDt.isNumericOrBool && actualElementDt != BaseDataType.LONG)
+                if(actualElementDt.isNumericOrBool)
                     DataType(BaseDataType.ARRAY, actualElementDt, null)
                 else
                     throw NoSuchElementException("invalid basic element dt $elementDt")
@@ -222,6 +224,7 @@ class DataType private constructor(val base: BaseDataType, val sub: BaseDataType
                 BaseDataType.WORD -> "word[]"
                 BaseDataType.UBYTE -> "ubyte[]"
                 BaseDataType.UWORD -> "uword[]"
+                BaseDataType.LONG -> "long[]"
                 else -> throw IllegalArgumentException("invalid sub type")
             }
         }
@@ -284,6 +287,7 @@ class DataType private constructor(val base: BaseDataType, val sub: BaseDataType
                 BaseDataType.BOOL -> "bool["
                 BaseDataType.BYTE -> "byte["
                 BaseDataType.WORD -> "@nosplit word["
+                BaseDataType.LONG -> "long["
                 BaseDataType.FLOAT -> "float["
                 else -> throw IllegalArgumentException("invalid sub type")
             }
@@ -344,6 +348,7 @@ class DataType private constructor(val base: BaseDataType, val sub: BaseDataType
     val isUnsignedWord =  base == BaseDataType.UWORD
     val isSignedWord =  base == BaseDataType.WORD
     val isInteger = base.isInteger
+    val isWordOrByteOrBool = base.isWordOrByteOrBool
     val isIntegerOrBool = base.isIntegerOrBool
     val isNumeric = base.isNumeric
     val isNumericOrBool = base.isNumericOrBool
@@ -362,6 +367,7 @@ class DataType private constructor(val base: BaseDataType, val sub: BaseDataType
     val isWordArray = base.isArray && !base.isPointerArray && (sub == BaseDataType.UWORD || sub == BaseDataType.WORD)
     val isUnsignedWordArray = base.isArray && !base.isPointerArray && sub == BaseDataType.UWORD
     val isSignedWordArray = base.isArray && !base.isPointerArray && sub == BaseDataType.WORD
+    val isLongArray = base.isArray && sub == BaseDataType.LONG
     val isFloatArray = base.isArray && !base.isPointerArray && sub == BaseDataType.FLOAT
     val isString = base == BaseDataType.STR
     val isBool = base == BaseDataType.BOOL
@@ -394,7 +400,9 @@ enum class RegisterOrPair {
     FAC2,
     // cx16 virtual registers:
     R0, R1, R2, R3, R4, R5, R6, R7,
-    R8, R9, R10, R11, R12, R13, R14, R15;
+    R8, R9, R10, R11, R12, R13, R14, R15,
+    // combined virtual registers to store 32 bits longs:
+    R0R1_32, R2R3_32, R4R5_32, R6R7_32, R8R9_32, R10R11_32, R12R13_32, R14R15_32;
 
     companion object {
         val names by lazy { entries.map { it.toString()} }
@@ -407,6 +415,18 @@ enum class RegisterOrPair {
         }
     }
 
+    fun startregname() = when(this) {
+        R0R1_32 -> "r0"
+        R2R3_32 -> "r2"
+        R4R5_32 -> "r4"
+        R6R7_32 -> "r6"
+        R8R9_32 -> "r8"
+        R10R11_32 -> "r10"
+        R12R13_32 -> "r12"
+        R14R15_32 -> "r14"
+        else -> throw IllegalArgumentException("must be a combined virtual register $this")
+    }
+
     fun asCpuRegister(): CpuRegister = when(this) {
         A -> CpuRegister.A
         X -> CpuRegister.X
@@ -420,13 +440,14 @@ enum class RegisterOrPair {
             BaseDataType.UBYTE, BaseDataType.BOOL -> "L"
             BaseDataType.BYTE -> "sL"
             BaseDataType.WORD -> "s"
-            BaseDataType.UWORD, null -> ""
-            else -> throw IllegalArgumentException("invalid register param type")
+            BaseDataType.UWORD, BaseDataType.POINTER, null -> ""
+            else -> throw IllegalArgumentException("invalid register param type for cx16 virtual reg")
         }
         return listOf("cx16", name.lowercase()+suffix)
     }
 
     fun isWord() = this==AX || this == AY || this==XY || this in Cx16VirtualRegisters
+    fun isLong() = this in combinedLongRegisters
 
 }       // only used in parameter and return value specs in asm subroutines
 
@@ -463,6 +484,17 @@ val Cx16VirtualRegisters = arrayOf(
     RegisterOrPair.R12, RegisterOrPair.R13, RegisterOrPair.R14, RegisterOrPair.R15
 )
 
+val combinedLongRegisters = arrayOf(
+    RegisterOrPair.R0R1_32,
+    RegisterOrPair.R2R3_32,
+    RegisterOrPair.R4R5_32,
+    RegisterOrPair.R6R7_32,
+    RegisterOrPair.R8R9_32,
+    RegisterOrPair.R10R11_32,
+    RegisterOrPair.R12R13_32,
+    RegisterOrPair.R14R15_32
+)
+
 val CpuRegisters = arrayOf(
     RegisterOrPair.A, RegisterOrPair.X, RegisterOrPair.Y,
     RegisterOrPair.AX, RegisterOrPair.AY, RegisterOrPair.XY
@@ -498,6 +530,5 @@ enum class ZeropageWish {
 
 enum class SplitWish {
     DONTCARE,
-    SPLIT,
     NOSPLIT
 }

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

@@ -6,7 +6,15 @@ import java.nio.file.InvalidPathException
 import kotlin.io.path.Path
 
 data class Position(val file: String, val line: Int, val startCol: Int, val endCol: Int) {
-    override fun toString(): String = "[$file: line $line col ${startCol+1}-${endCol+1}]"
+    override fun toString(): String = "[$file: line $line col ${startCol}-${endCol}]"
+
+    init {
+        if(!file.startsWith('~') ||!file.endsWith('~'))
+            require(line>0 && startCol>=0 && endCol>=startCol) {
+                "Invalid position: $this"
+            }
+    }
+
     fun toClickableStr(): String {
         if(this===DUMMY)
             return ""

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

@@ -14,6 +14,7 @@ internal class NormalMemSizer(val floatsize: Int): IMemSizer {
             return when(dt.sub) {
                 BaseDataType.BOOL, BaseDataType.UBYTE, BaseDataType.BYTE -> numElements
                 BaseDataType.UWORD, BaseDataType.WORD, BaseDataType.STR -> numElements * 2
+                BaseDataType.LONG -> numElements * 4
                 BaseDataType.FLOAT-> numElements * floatsize
                 BaseDataType.UNDEFINED -> throw IllegalArgumentException("undefined has no memory size")
                 else -> throw IllegalArgumentException("invalid sub type")

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

@@ -252,6 +252,14 @@ private fun PtVariable.prefix(parent: PtNode, st: SymbolTable): PtVariable {
                         newValue.add(newAddr)
                     }
                 }
+                is PtBuiltinFunctionCall -> {
+                    // could be a struct instance or memory slab "allocation"
+                    if (elt.name != "prog8_lib_structalloc" && elt.name != "memory")
+                        throw AssemblyError("weird array value element $elt")
+                    else {
+                        newValue.add(elt)
+                    }
+                }
                 else -> throw AssemblyError("weird array value element $elt")
             }
         }
@@ -443,7 +451,7 @@ class AsmGen6502Internal (
             if(line.length>=4 && invalid.matchesAt(line, 0)) {
                 errors.err(
                     "invalid assembly instruction used (not compatible with the 65816 CPU): ${line.trim()}",
-                    Position("<output-assemblycode>", index, 1, 1)
+                    Position("~output-assemblycode~", index+1, 1, 1)
                 )
             }
         }
@@ -550,7 +558,7 @@ class AsmGen6502Internal (
                 } else {
                     return if (allocator.isZpVar((target as PtNamedNode).scopedName)) {
                         // pointervar is already in the zero page, no need to copy
-                        loadAFromZpPointerVar(sourceName, true)
+                        loadAFromZpPointerVar(sourceName)
                         sourceName
                     } else {
                         out("""
@@ -609,7 +617,7 @@ class AsmGen6502Internal (
         }
     }
 
-    internal fun loadAFromZpPointerVar(zpPointerVar: String, keepY: Boolean) {
+    internal fun loadAFromZpPointerVar(zpPointerVar: String, keepY: Boolean=false) {
         if (isTargetCpu(CpuType.CPU65C02))
             out("  lda  ($zpPointerVar)")
         else {
@@ -750,6 +758,15 @@ class AsmGen6502Internal (
                     CpuRegister.Y -> out(" tay")
                 }
             }
+            expr.type.isLong -> {
+                assignExpressionToRegister(expr.index, RegisterOrPair.A)
+                out("  asl  a |  asl  a")
+                when (register) {
+                    CpuRegister.A -> {}
+                    CpuRegister.X -> out(" tax")
+                    CpuRegister.Y -> out(" tay")
+                }
+            }
             expr.type.isFloat -> {
                 if(options.compTarget.FLOAT_MEM_SIZE != 5)
                     TODO("support float size other than 5 ${expr.position}")
@@ -798,6 +815,7 @@ class AsmGen6502Internal (
             RegisterOrPair.AY,
             RegisterOrPair.XY -> assignmentAsmGen.assignRegisterpairWord(target, reg)
             in Cx16VirtualRegisters -> assignmentAsmGen.assignVirtualRegister(target, reg)
+            in combinedLongRegisters -> assignmentAsmGen.assignVirtualRegister(target, reg)
             RegisterOrPair.FAC1 -> assignmentAsmGen.assignFAC1float(target)
             RegisterOrPair.FAC2 -> assignmentAsmGen.assignFAC2float(target)
             else -> throw AssemblyError("invalid register")
@@ -805,6 +823,9 @@ class AsmGen6502Internal (
     }
 
     internal fun assignExpressionTo(value: PtExpression, target: AsmAssignTarget) {
+
+        // this is basically the fallback assignment routine, it is RARELY called
+
         when {
             target.datatype.isByteOrBool -> {
                 if (value.asConstInteger()==0) {
@@ -832,7 +853,7 @@ class AsmGen6502Internal (
                             return
                         }
                         TargetStorageKind.POINTER -> {
-                            TODO("assign to pointer ${target.position}")
+                            pointerGen.assignByte(PtrTarget(target), 0)
                             return
                         }
                         else -> { }
@@ -842,7 +863,10 @@ class AsmGen6502Internal (
                 assignExpressionToRegister(value, RegisterOrPair.A)
                 assignmentAsmGen.assignRegisterByte(target, CpuRegister.A, target.datatype.isSigned, false)
             }
-            target.datatype.isPointer -> TODO("assign expression to pointer ${target.position}")
+            target.datatype.isPointer -> {
+                assignExpressionToRegister(value, RegisterOrPair.AX)
+                pointerGen.assignWordReg(PtrTarget(target), RegisterOrPair.AX)
+            }
             target.datatype.isWord || target.datatype.isPassByRef -> {
                 assignExpressionToRegister(value, RegisterOrPair.AY)
                 translateNormalAssignment(
@@ -852,6 +876,102 @@ class AsmGen6502Internal (
                     ), value.definingISub()
                 )
             }
+            target.datatype.isLong -> {
+                if(value is PtNumber) {
+                    val hex = value.asConstInteger()!!.toLongHex()
+                    when(target.kind) {
+                        TargetStorageKind.VARIABLE -> {
+                            out("""
+                                lda  #$${hex.substring(6,8)}
+                                sta  ${target.asmVarname}
+                                lda  #$${hex.substring(4, 6)}
+                                sta  ${target.asmVarname}+1
+                                lda  #$${hex.substring(2, 4)}
+                                sta  ${target.asmVarname}+2
+                                lda  #$${hex.take(2)}
+                                sta  ${target.asmVarname}+3""")
+                        }
+                        TargetStorageKind.ARRAY -> TODO("assign long to array  ${target.position}")
+                        TargetStorageKind.MEMORY -> throw AssemblyError("memory is bytes not long ${target.position}")
+                        TargetStorageKind.REGISTER -> assignExpressionToRegister(value, target.register!!, true)
+                        TargetStorageKind.POINTER -> pointerGen.assignLong(target.pointer!!, value.number.toInt())
+                        TargetStorageKind.VOID -> { /* do nothing */ }
+                    }
+                } else if(value is PtIdentifier && value.type.isLong) {
+                    when (target.kind) {
+                        TargetStorageKind.VARIABLE -> {
+                            val valuesym = asmSymbolName(value)
+                            out(
+                                """
+                                lda  $valuesym
+                                sta  ${target.asmVarname}
+                                lda  $valuesym+1
+                                sta  ${target.asmVarname}+1
+                                lda  $valuesym+2
+                                sta  ${target.asmVarname}+2
+                                lda  $valuesym+3
+                                sta  ${target.asmVarname}+3"""
+                            )
+                        }
+
+                        TargetStorageKind.ARRAY -> TODO("assign long to array  ${target.position}")
+                        TargetStorageKind.MEMORY -> throw AssemblyError("memory is bytes not long ${target.position}")
+                        TargetStorageKind.REGISTER -> assignExpressionToRegister(value, target.register!!, true)
+                        TargetStorageKind.POINTER -> pointerGen.assignLongVar(target.pointer!!, asmSymbolName(value))
+                        TargetStorageKind.VOID -> { /* do nothing */ }
+                    }
+                } else if(value is PtTypeCast && value.type.isLong) {
+                    if(value.value is PtIdentifier) {
+                        val valuesym = asmSymbolName((value.value as PtIdentifier).name)
+                        if(value.value.type.isByte) {
+                            when (target.kind) {
+                                TargetStorageKind.VARIABLE -> {
+                                    out("  lda  $valuesym |  sta  ${target.asmVarname}")
+                                    signExtendLongVariable(target.asmVarname, value.value.type.base)
+                                }
+                                TargetStorageKind.ARRAY -> TODO("assign typecasted long to array  ${target.position}")
+                                TargetStorageKind.MEMORY -> throw AssemblyError("memory is bytes not long ${target.position}")
+                                TargetStorageKind.REGISTER -> assignExpressionToRegister(value, target.register!!, true)
+                                TargetStorageKind.POINTER -> TODO("assign typecasted long into pointer  ${target.position}")
+                                TargetStorageKind.VOID -> { /* do nothing */ }
+                            }
+                        } else if(value.value.type.isWord) {
+                            when (target.kind) {
+                                TargetStorageKind.VARIABLE -> {
+                                    out("""
+                                        lda  $valuesym
+                                        sta  ${target.asmVarname}
+                                        lda  $valuesym+1
+                                        sta  ${target.asmVarname}+1""")
+                                    signExtendLongVariable(target.asmVarname, value.value.type.base)
+                                }
+                                TargetStorageKind.ARRAY -> TODO("assign typecasted long to array  ${target.position}")
+                                TargetStorageKind.MEMORY -> throw AssemblyError("memory is bytes not long ${target.position}")
+                                TargetStorageKind.REGISTER -> {
+                                    require(target.register in combinedLongRegisters)
+                                    val startreg = target.register!!.startregname()
+                                    out("""
+                                        lda  $valuesym
+                                        sta  cx16.$startreg
+                                        lda  $valuesym+1
+                                        sta  cx16.$startreg+1""")
+                                    signExtendLongVariable("cx16.$startreg", value.value.type.base)
+                                }
+                                TargetStorageKind.POINTER -> TODO("assign typecasted long into pointer  ${target.position}")
+                                TargetStorageKind.VOID -> { /* do nothing */ }
+                            }
+                        } else throw AssemblyError("weird casted type")
+                    } else {
+                        TODO("assign typecasted expression $value to a long target ${target.kind} at  ${target.position}  - use simple expressions and temporary variables for now")
+                    }
+                } else if(target.kind == TargetStorageKind.REGISTER) {
+                    assignExpressionToRegister(value, target.register!!, true)
+                } else if(target.kind == TargetStorageKind.VARIABLE) {
+                    assignExpressionToVariable(value, target.asmVarname, target.datatype)
+                } else {
+                    TODO("assign long expression $value to a target ${target.kind} at  ${target.position} - use simple expressions and temporary variables for now")
+                }
+            }
             target.datatype.isFloat -> {
                 assignExpressionToRegister(value, RegisterOrPair.FAC1)
                 assignRegister(RegisterOrPair.FAC1, target)
@@ -1157,12 +1277,12 @@ $repeatLabel""")
                             // print a message when more optimal code is possible for 65C02 cpu
                             val variable = symbolTable.lookup(arrayVariable.name)!!
                             if(variable is StStaticVariable && variable.length!!<=128u)
-                                errors.info("the jump address array is @split, but @nosplit would create more efficient code here", jump.position)
+                                errors.info("the jump address array is split, but @nosplit would create more efficient code here", jump.position)
                         }
                     } else {
                         // print a message when more optimal code is possible for 6502 cpu
                         if(!arrayIdx.splitWords)
-                            errors.info("the jump address array is @nosplit, but @split would create more efficient code here", jump.position)
+                            errors.info("the jump address array is @nosplit, but split would create more efficient code here", jump.position)
                     }
                 }
                 // we can do the address evaluation right now and just use a temporary pointer variable
@@ -1182,7 +1302,7 @@ $repeatLabel""")
         if(returnvalue!=null) {
             val returnDt = sub.signature.returns.single()
             if (returnDt.isNumericOrBool || returnDt.isPointer) {
-                assignExpressionToRegister(returnvalue, returnRegs.single().first.registerOrPair!!)
+                assignExpressionToRegister(returnvalue, returnRegs.single().first.registerOrPair!!, returnDt.isSigned)
             }
             else {
                 // all else take its address and assign that also to AY register pair
@@ -1203,7 +1323,7 @@ $repeatLabel""")
                 assignExpressionTo(it.first as PtExpression, tgt)
             }
             assigns.first().also {
-                assignExpressionToRegister(it.first as PtExpression, it.second.first.registerOrPair!!)
+                assignExpressionToRegister(it.first as PtExpression, it.second.first.registerOrPair!!, (it.first as PtExpression).type.isSigned)
             }
         }
         out("  rts")
@@ -1289,6 +1409,53 @@ $repeatLabel""")
         }
     }
 
+    internal fun signExtendLongVariable(asmvar: String, valueDt: BaseDataType) {
+        // sign extend signed word in a var to a full long in that variable
+        when(valueDt) {
+            BaseDataType.UBYTE -> {
+                if(isTargetCpu(CpuType.CPU65C02)) {
+                    out("""
+                        stz  $asmvar+1
+                        stz  $asmvar+2
+                        stz  $asmvar+3""")
+                }
+                else {
+                    out("""
+                        lda  #0
+                        sta  $asmvar+1
+                        sta  $asmvar+2
+                        sta  $asmvar+3""")
+                }
+            }
+            BaseDataType.BYTE -> {
+                out("""
+                    lda  $asmvar
+                    ora  #$7f
+                    bmi  +
+                    lda  #0
++                   sta  $asmvar+1
+                    sta  $asmvar+2
+                    sta  $asmvar+3""")
+            }
+            BaseDataType.UWORD -> {
+                if(isTargetCpu(CpuType.CPU65C02))
+                    out("  stz  $asmvar+2 |  stz  $asmvar+3")
+                else
+                    out("  lda  #0 |  sta  $asmvar+2 |  sta  $asmvar+3")
+            }
+            BaseDataType.WORD -> {
+                out("""
+                    lda  $asmvar+1
+                    ora  #$7f
+                    bmi  +
+                    lda  #0
++                   sta  $asmvar+2
+                    sta  $asmvar+3""")
+            }
+            else -> throw AssemblyError("need byte or word type")
+        }
+    }
+
     internal fun isZpVar(variable: PtIdentifier): Boolean = allocator.isZpVar(variable.name)
 
     internal fun jmp(asmLabel: String, indirect: Boolean=false, indexedX: Boolean=false) {
@@ -1364,7 +1531,7 @@ $repeatLabel""")
                 val constOffset = (ptrAndIndex.second as? PtNumber)?.number?.toInt()
                 if(addrOf!=null && constOffset!=null) {
                     if(addrOf.isFromArrayElement) {
-                        TODO("address-of array element $addrOf")
+                        TODO("address-of array element ${addrOf.position}")
                     } else if(addrOf.dereference!=null) {
                         throw AssemblyError("write &dereference, makes no sense at ${addrOf.position}")
                     } else {
@@ -1406,9 +1573,9 @@ $repeatLabel""")
             val constOffset = (ptrAndIndex.second as? PtNumber)?.number?.toInt()
             if(addrOf!=null && constOffset!=null) {
                 if(addrOf.isFromArrayElement) {
-                    TODO("address-of array element $addrOf")
+                    TODO("address-of array element ${addrOf.position}")
                 } else if(addrOf.dereference!=null) {
-                    TODO("read &dereference")
+                    TODO("read &dereference ${addrOf.position}")
                 } else {
                     out("  lda  ${asmSymbolName(addrOf.identifier!!)}+${constOffset}")
                     return true
@@ -1456,7 +1623,7 @@ $repeatLabel""")
                 val constOffset = (ptrAndIndex.second as? PtNumber)?.number?.toInt()
                 if(addrOf!=null && constOffset!=null) {
                     if(addrOf.isFromArrayElement) {
-                        TODO("address-of array element $addrOf")
+                        TODO("address-of array element ${addrOf.position}")
                     } else if(addrOf.dereference!=null) {
                         throw AssemblyError("write &dereference, makes no sense at ${addrOf.position}")
                     } else {
@@ -1489,9 +1656,9 @@ $repeatLabel""")
                 val constOffset = (ptrAndIndex.second as? PtNumber)?.number?.toInt()
                 if(addrOf!=null && constOffset!=null) {
                     if(addrOf.isFromArrayElement) {
-                        TODO("address-of array element $addrOf")
+                        TODO("address-of array element ${addrOf.position}")
                     } else if(addrOf.dereference!=null) {
-                        TODO("read &dereference")
+                        TODO("read &dereference ${addrOf.position}")
                     } else {
                         out("  lda  ${asmSymbolName(addrOf.identifier!!)}-${constOffset}")
                         return true
@@ -1836,9 +2003,9 @@ $repeatLabel""")
                 lda  #<$leftName
                 ldy  #>$leftName""")
             if(lessOrEquals)
-                out("jsr  floats.vars_lesseq_f")
+                out("  jsr  floats.vars_lesseq_f")
             else
-                out("jsr  floats.vars_less_f")
+                out("  jsr  floats.vars_less_f")
         }
         fun lessf(expr: PtExpression, rightName: String) {
             assignExpressionToRegister(expr, RegisterOrPair.FAC1, true)
@@ -1957,6 +2124,22 @@ $repeatLabel""")
         }
     }
 
+    internal fun loadIndirectLongIntoR14R15(zpPtrVar: String, offset: UByte) {
+        out("""
+            ldy  #$offset
+            lda  ($zpPtrVar),y
+            sta  cx16.r14
+            iny
+            lda  ($zpPtrVar),y
+            sta  cx16.r14+1
+            iny
+            lda  ($zpPtrVar),y
+            sta  cx16.r14+2
+            iny
+            lda  ($zpPtrVar),y
+            sta  cx16.r14+3""")
+    }
+
     internal fun storeIndirectByte(byte: Int, zpPtrVar: String, offset: UByte) {
         if (offset > 0u) {
             out("  lda  #$byte |  ldy  #$offset |  sta  ($zpPtrVar),y")
@@ -2093,7 +2276,7 @@ $repeatLabel""")
                 }
 
                 in Cx16VirtualRegisters -> {
-                    val regname = regs.asScopedNameVirtualReg(DataType.UWORD)
+                    val regname = regs.asScopedNameVirtualReg(DataType.UWORD).joinToString(".")
                     out("""
                         lda  $regname
                         ldy  #$offset
@@ -2160,7 +2343,7 @@ $repeatLabel""")
                 }
 
                 in Cx16VirtualRegisters -> {
-                    val regname = regs.asScopedNameVirtualReg(DataType.UWORD)
+                    val regname = regs.asScopedNameVirtualReg(DataType.UWORD).joinToString(".")
                     out("""
                         lda  $regname
                         ldy  #0
@@ -2269,6 +2452,69 @@ $repeatLabel""")
             }
         }
     }
+
+    internal fun pushLongRegisters(startreg: RegisterOrPair, numberOfLongs: Int) {
+        if(numberOfLongs==0) return
+
+        val startregName = startreg.startregname()
+        if(numberOfLongs==1) {
+            out("""
+                sta  P8ZP_SCRATCH_REG
+                lda  cx16.$startregName
+                pha
+                lda  cx16.$startregName+1
+                pha
+                lda  cx16.$startregName+2
+                pha
+                lda  cx16.$startregName+3
+                pha
+                lda  P8ZP_SCRATCH_REG""")
+            return
+        }
+        val numbytes = numberOfLongs * options.compTarget.memorySize(BaseDataType.LONG)
+        out("""
+            sta  P8ZP_SCRATCH_REG
+            sty  P8ZP_SCRATCH_B1
+            ldy  #0
+-           lda  cx16.$startregName,y
+            pha
+            iny
+            cpy  #$numbytes
+            bne  -
+            ldy  P8ZP_SCRATCH_B1
+            lda  P8ZP_SCRATCH_REG""")
+    }
+
+    internal fun popLongRegisters(startreg: RegisterOrPair, numberOfLongs: Int) {
+        if(numberOfLongs==0) return
+
+        val startregName = startreg.startregname()
+        if(numberOfLongs==1) {
+            out("""
+                sta  P8ZP_SCRATCH_REG
+                pla
+                sta  cx16.$startregName+3
+                pla
+                sta  cx16.$startregName+2
+                pla
+                sta  cx16.$startregName+1
+                pla
+                sta  cx16.$startregName
+                lda  P8ZP_SCRATCH_REG""")
+            return
+        }
+        val numbytes = numberOfLongs * options.compTarget.memorySize(BaseDataType.LONG)
+        out("""
+            sta  P8ZP_SCRATCH_REG
+            sty  P8ZP_SCRATCH_B1
+            ldy  #$numbytes-1
+-           pla
+            sta  cx16.$startregName,y
+            dey
+            bpl  -
+            ldy  P8ZP_SCRATCH_B1
+            lda  P8ZP_SCRATCH_REG""")
+    }
 }
 
 /**
@@ -2282,4 +2528,8 @@ internal class SubroutineExtraAsmInfo {
     var usedFloatEvalResultVar2 = false
 
     val extraVars = mutableListOf<Triple<BaseDataType, String, UInt?>>()
-}
+}
+
+
+internal fun Double.toLongHex(): String = this.toUInt().toString(16).padStart(8, '0')
+internal fun Int.toLongHex(): String = this.toUInt().toString(16).padStart(8, '0')

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

@@ -463,15 +463,19 @@ private fun optimizeStoreLoadSame(
                 mods.add(Modification(lines[2].index, true, null))
         }
 
-        //  all 3 registers:  lda VALUE + sta SOMEWHERE + lda VALUE  -> last load can be eliminated
+        //  all 3 registers:  lda VALUE + sta SOMEWHERE + lda VALUE  -> last load can be eliminated IF NOT IO ADDRESS
         if (first.startsWith("lda ") && second.startsWith("sta ") && third.startsWith("lda ") ||
             first.startsWith("ldx ") && second.startsWith("stx ") && third.startsWith("ldx ") ||
             first.startsWith("ldy ") && second.startsWith("sty ") && third.startsWith("ldy ")
         ) {
             val firstVal = first.substring(4).trimStart()
             val thirdVal = third.substring(4).trimStart()
-            if (firstVal == thirdVal)
-                mods.add(Modification(lines[3].index, true, null))
+            if (firstVal == thirdVal) {
+                val address = getAddressArg(third, symbolTable)
+                if (address != null && !machine.isIOAddress(address)) {
+                    mods.add(Modification(lines[3].index, true, null))
+                }
+            }
         }
     }
     return mods
@@ -540,7 +544,7 @@ private fun optimizeJsrRtsAndOtherCombinations(linesByFour: Sequence<List<Indexe
         val third = lines[2].value
 
         if(!haslabel(second)) {
-            if ((" jmp" in first || "\tjmp" in first ) && (" rts" in second || "\trts" in second)) {
+            if ((" jmp" in first || "\tjmp" in first || " bra" in first || "\tbra" in first ) && (" rts" in second || "\trts" in second)) {
                 mods += Modification(lines[1].index, true, null)
             }
             else if ((" jsr" in first || "\tjsr" in first ) && (" rts" in second || "\trts" in second)) {

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

@@ -1,5 +1,7 @@
 package prog8.codegen.cpu6502
 
+import prog8.code.StMemorySlabBlockName
+import prog8.code.StStructInstanceBlockName
 import prog8.code.SymbolTable
 import prog8.code.ast.*
 import prog8.code.core.*
@@ -25,17 +27,20 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         val sscope = fcall.definingISub()
 
         when (fcall.name) {
-            "lsw" -> throw AssemblyError("lsw() should have been removed or replaced by a const value")
-            "msw" -> throw AssemblyError("msw() should have been removed or replaced by a const value")
+            "msw" -> funcMsw(fcall, resultRegister)
+            "lsw" -> funcLsw(fcall, resultRegister)
             "msb" -> funcMsb(fcall, resultRegister)
-            "lsb" -> funcLsb(fcall, resultRegister)
+            "msb__long" -> funcMsbLong(fcall, resultRegister)
+            "lsb" -> funcLsb(fcall, resultRegister, false)
+            "lsb__long" -> funcLsb(fcall, resultRegister, true)
             "mkword" -> funcMkword(fcall, resultRegister)
-            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(fcall, resultRegister)
-            "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(fcall, resultRegister)
-            "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(fcall, resultRegister)
-            "abs__byte", "abs__word", "abs__float" -> funcAbs(fcall, resultRegister, sscope)
+            "mklong", "mklong2" -> funcMklong(fcall)  // result is in R14:R15
+            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword", "clamp__long" -> funcClamp(fcall, resultRegister)
+            "min__byte", "min__ubyte", "min__word", "min__uword", "min__long" -> funcMin(fcall, resultRegister)
+            "max__byte", "max__ubyte", "max__word", "max__uword", "max__long" -> funcMax(fcall, resultRegister)
+            "abs__byte", "abs__word", "abs__long", "abs__float" -> funcAbs(fcall, resultRegister, sscope)
             "sgn" -> funcSgn(fcall, resultRegister, sscope)
-            "sqrt__ubyte", "sqrt__uword", "sqrt__float" -> funcSqrt(fcall, resultRegister, sscope)
+            "sqrt__ubyte", "sqrt__uword", "sqrt__long", "sqrt__float" -> funcSqrt(fcall, resultRegister, sscope)
             "divmod__ubyte" -> funcDivmod(fcall)
             "divmod__uword" -> funcDivmodW(fcall)
             "rol" -> funcRol(fcall)
@@ -46,10 +51,12 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             "setmsb" -> funcSetLsbMsb(fcall, true)
             "memory" -> funcMemory(fcall, discardResult, resultRegister)
             "peekw" -> funcPeekW(fcall, resultRegister)
+            "peekl" -> funcPeekL(fcall, resultRegister)
             "peekf" -> funcPeekF(fcall, resultRegister)
             "peekbool" -> funcPeekBool(fcall, resultRegister)
             "peek" -> throw AssemblyError("peek() should have been replaced by @()")
             "pokew" -> funcPokeW(fcall)
+            "pokel" -> funcPokeL(fcall)
             "pokef" -> funcPokeF(fcall)
             "pokemon" -> {
                 val memread = PtMemoryByte(fcall.position)
@@ -344,38 +351,112 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 }
             } else
                 throw AssemblyError("args for cmp() should have same dt")
-        } else {
-            // arg1 is a word
+        } else if(arg1.type.isWord) {
             if(arg2.type.isWord) {
-                when (arg2) {
-                    is PtIdentifier -> {
-                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
-                        asmgen.out("""
-                            cpy  ${asmgen.asmVariableName(arg2)}+1
-                            bne  +
-                            cmp  ${asmgen.asmVariableName(arg2)}
-+""")
+                if(arg1.type.isSigned) {
+                    when (arg2) {
+                        is PtIdentifier -> {
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                            asmgen.out("""
+                                sec
+                                sbc  ${asmgen.asmVariableName(arg2)}
+                                tya
+                                sbc  ${asmgen.asmVariableName(arg2)}+1""")
+                        }
+                        is PtBool -> TODO("word compare against bool  ${arg2.position}")
+                        is PtNumber -> {
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                            asmgen.out("""
+                                sec
+                                sbc  #<${arg2.number.toInt()}
+                                tya
+                                sbc  #>${arg2.number.toInt()}""")
+                        }
+                        else -> {
+                            asmgen.assignWordOperandsToAYAndVar(arg1, arg2, "P8ZP_SCRATCH_W1")
+                            asmgen.out("""
+                                sec
+                                sbc  P8ZP_SCRATCH_W1
+                                tya
+                                sbc  P8ZP_SCRATCH_W1+1""")
+                        }
                     }
-                    is PtBool -> TODO("word compare against bool  ${arg2.position}")
-                    is PtNumber -> {
-                        asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
-                        asmgen.out("""
-                            cpy  #>${arg2.number.toInt()}
-                            bne  +
-                            cmp  #<${arg2.number.toInt()}
+                } else {
+                    when (arg2) {
+                        is PtIdentifier -> {
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                            asmgen.out("""
+                                cpy  ${asmgen.asmVariableName(arg2)}+1
+                                bne  +
+                                cmp  ${asmgen.asmVariableName(arg2)}
 +""")
-                    }
-                    else -> {
-                        asmgen.assignWordOperandsToAYAndVar(arg1, arg2, "P8ZP_SCRATCH_W1")
-                        asmgen.out("""
-                            cpy  P8ZP_SCRATCH_W1+1
-                            bne  +
-                            cmp  P8ZP_SCRATCH_W1
+                        }
+                        is PtBool -> TODO("word compare against bool  ${arg2.position}")
+                        is PtNumber -> {
+                            asmgen.assignExpressionToRegister(arg1, RegisterOrPair.AY)
+                            asmgen.out("""
+                                cpy  #>${arg2.number.toInt()}
+                                bne  +
+                                cmp  #<${arg2.number.toInt()}
 +""")
+                        }
+                        else -> {
+                            asmgen.assignWordOperandsToAYAndVar(arg1, arg2, "P8ZP_SCRATCH_W1")
+                            asmgen.out("""
+                                cpy  P8ZP_SCRATCH_W1+1
+                                bne  +
+                                cmp  P8ZP_SCRATCH_W1
++""")
+                        }
                     }
                 }
             } else
                 throw AssemblyError("args for cmp() should have same dt")
+        } else if(arg1.type.isLong) {
+            if(arg2.type.isLong) {
+                if(arg1 is PtIdentifier && arg2 is PtNumber) {
+                    val var1 = asmgen.asmVariableName(arg1)
+                    val hex = arg2.number.toLongHex()
+                    asmgen.out("""
+                        sec
+                        lda  $var1
+                        sbc  #$${hex.substring(6, 8)}
+                        lda  $var1+1
+                        sbc  #$${hex.substring(4, 6)}
+                        lda  $var1+2
+                        sbc  #$${hex.substring(2, 4)}
+                        lda  $var1+3
+                        sbc  #$${hex.take(2)}""")
+                } else if(arg1 is PtIdentifier && arg2 is PtIdentifier) {
+                    val var1 = asmgen.asmVariableName(arg1)
+                    val var2 = asmgen.asmVariableName(arg2)
+                    asmgen.out("""
+                        sec
+                        lda  $var1
+                        sbc  $var2
+                        lda  $var1+1
+                        sbc  $var2+1
+                        lda  $var1+2
+                        sbc  $var2+2
+                        lda  $var1+3
+                        sbc  $var2+3""")
+                } else {
+                    // cmp() doesn't return a value and as such can't be used in an expression, so no need to save the temp registers' original values
+                    assignAsmGen.assignExpressionToRegister(arg2, RegisterOrPair.R14R15_32, true)
+                    assignAsmGen.assignExpressionToRegister(arg1, RegisterOrPair.R12R13_32, true)
+                    asmgen.out("""
+                        sec
+                        lda  cx16.r12
+                        sbc  cx16.r14
+                        lda  cx16.r12+1
+                        sbc  cx16.r14+1
+                        lda  cx16.r12+2
+                        sbc  cx16.r14+2
+                        lda  cx16.r12+3
+                        sbc  cx16.r14+3""")
+                }
+            } else
+                throw AssemblyError("args for cmp() should have same dt")
         }
     }
 
@@ -385,7 +466,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         val name = (fcall.args[0] as PtString).value
         require(name.all { it.isLetterOrDigit() || it=='_' }) {"memory name should be a valid symbol name ${fcall.position}"}
 
-        val slabname = PtIdentifier("prog8_slabs.prog8_memoryslab_$name", DataType.UWORD, fcall.position)
+        val slabname = PtIdentifier("$StMemorySlabBlockName.memory_$name", DataType.UWORD, fcall.position)
         val addressOf = PtAddressOf(DataType.pointer(BaseDataType.UBYTE), false, fcall.position)
         addressOf.add(slabname)
         addressOf.parent = fcall
@@ -399,9 +480,10 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         if(discardResult)
             throw AssemblyError("should not discard result of struct allocation at $fcall")
         // ... don't need to pay attention to args here because struct instance is put together elsewhere we just have to get a pointer to it
-        val slabname = SymbolTable.labelnameForStructInstance(fcall)
+        val prefix = if(fcall.args.isEmpty()) "${StStructInstanceBlockName}_bss" else StStructInstanceBlockName
+        val labelname = PtIdentifier("$prefix.${SymbolTable.labelnameForStructInstance(fcall)}", fcall.type, fcall.position)
         val addressOf = PtAddressOf(fcall.type, true, fcall.position)
-        addressOf.add(PtIdentifier(slabname, fcall.type, fcall.position))
+        addressOf.add(labelname)
         addressOf.parent = fcall
         val src = AsmAssignSource(SourceStorageKind.EXPRESSION, program, asmgen, fcall.type, expression = addressOf)
         val target = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, null, asmgen)
@@ -421,6 +503,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 asmgen.out("  jsr  prog8_lib.func_sqrt16_into_A")
                 assignAsmGen.assignRegisterByte(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, false, fcall.position, scope, asmgen), CpuRegister.A, false, false)
             }
+            BaseDataType.LONG -> TODO("sqrt LONG ${fcall.position}")
             BaseDataType.FLOAT -> {
                 asmgen.out("  jsr  floats.func_sqrt_into_FAC1")
                 assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
@@ -455,7 +538,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  lda  $variable |  lsr  a |  bcc  + |  ora  #$80 |+  |  sta  $variable")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
                 }
             }
             BaseDataType.UWORD -> {
@@ -474,7 +557,27 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  lsr  $variable+1 |  ror  $variable |  bcc  + |  lda  $variable+1 |  ora  #$80 |  sta  $variable+1 |+  ")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
+                }
+            }
+            BaseDataType.LONG -> {
+                // a bit strange, rotating a signed type, but it has to do for now
+                when(what) {
+                    is PtArrayIndexer -> TODO("ror2 long array ${what.position}")
+                    is PtIdentifier -> {
+                        val variable = asmgen.asmVariableName(what)
+                        asmgen.out("""
+                            lsr  $variable+3
+                            ror  $variable+2
+                            ror  $variable+1
+                            ror  $variable
+                            bcc  +
+                            lda  $variable+3
+                            ora  #$80
+                            sta  $variable+3
++""")
+                    }
+                    else -> throw AssemblyError("weird node")
                 }
             }
             else -> throw AssemblyError("weird type")
@@ -517,7 +620,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  ror  $variable")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
                 }
             }
             BaseDataType.UWORD -> {
@@ -538,7 +641,18 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  ror  $variable+1 |  ror  $variable")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
+                }
+            }
+            BaseDataType.LONG -> {
+                // a bit strange, rotating a signed type, but it has to do for now
+                when(what) {
+                    is PtArrayIndexer -> TODO("ror long array ${what.position}")
+                    is PtIdentifier -> {
+                        val variable = asmgen.asmVariableName(what)
+                        asmgen.out("  ror  $variable+3 |  ror  $variable+2 |  ror  $variable+1 |  ror  $variable")
+                    }
+                    else -> throw AssemblyError("weird node")
                 }
             }
             else -> throw AssemblyError("weird type")
@@ -570,7 +684,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  lda  $variable |  cmp  #$80 |  rol  a |  sta  $variable")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
                 }
             }
             BaseDataType.UWORD -> {
@@ -589,7 +703,25 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  asl  $variable |  rol  $variable+1 |  bcc  + |  inc  $variable |+  ")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
+                }
+            }
+            BaseDataType.LONG -> {
+                // a bit strange, rotating a signed type, but it has to do for now
+                when(what) {
+                    is PtArrayIndexer -> TODO("rol2 long array ${what.position}")
+                    is PtIdentifier -> {
+                        val variable = asmgen.asmVariableName(what)
+                        asmgen.out("""
+                            asl  $variable
+                            rol  $variable+1
+                            rol  $variable+2
+                            rol  $variable+3
+                            bcc  +
+                            inc  $variable
++""")
+                    }
+                    else -> throw AssemblyError("weird node")
                 }
             }
             else -> throw AssemblyError("weird type")
@@ -631,7 +763,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  rol  $variable")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
                 }
             }
             BaseDataType.UWORD -> {
@@ -652,7 +784,18 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         val variable = asmgen.asmVariableName(what)
                         asmgen.out("  rol  $variable |  rol  $variable+1")
                     }
-                    else -> throw AssemblyError("weird type")
+                    else -> throw AssemblyError("weird node")
+                }
+            }
+            BaseDataType.LONG -> {
+                // a bit strange, rotating a signed type, but it has to do for now
+                when(what) {
+                    is PtArrayIndexer -> TODO("rol long array ${what.position}")
+                    is PtIdentifier -> {
+                        val variable = asmgen.asmVariableName(what)
+                        asmgen.out("  rol  $variable |  rol  $variable+1 |  rol  $variable+2 |  rol  $variable+3")
+                    }
+                    else -> throw AssemblyError("weird node")
                 }
             }
             else -> throw AssemblyError("weird type")
@@ -726,6 +869,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
             BaseDataType.BYTE -> asmgen.out("  jsr  prog8_lib.func_sign_b_into_A")
             BaseDataType.UWORD -> asmgen.out("  jsr  prog8_lib.func_sign_uw_into_A")
             BaseDataType.WORD -> asmgen.out("  jsr  prog8_lib.func_sign_w_into_A")
+            BaseDataType.LONG -> asmgen.out("  jsr  prog8_lib.func_sign_l_r14r15_into_A")          // note: long arg is stored in R14:R15
             BaseDataType.FLOAT -> asmgen.out("  jsr  floats.func_sign_f_into_A")
             else -> throw AssemblyError("weird type $dt")
         }
@@ -744,6 +888,10 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 asmgen.out("  jsr  prog8_lib.abs_w_into_AY")
                 assignAsmGen.assignRegisterpairWord(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, false, fcall.position, scope, asmgen), RegisterOrPair.AY)
             }
+            BaseDataType.LONG -> {
+                asmgen.out("  jsr  prog8_lib.abs_l_into_R14R15")
+                assignAsmGen.assignRegisterLong(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.R14R15_32, true, fcall.position, scope, asmgen), RegisterOrPair.R14R15_32)
+            }
             BaseDataType.FLOAT -> {
                 asmgen.out("  jsr  floats.func_abs_f_into_FAC1")
                 assignAsmGen.assignFAC1float(AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.FAC1, true, fcall.position, scope, asmgen))
@@ -821,6 +969,19 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.restoreRegisterStack(CpuRegister.Y, true)
                     asmgen.out("  sta  ($varname),y")
                     return
+                } else if(addrExpr.operator=="+" && addrExpr.left is PtIdentifier) {
+                    asmgen.assignExpressionToRegister(addrExpr.right, RegisterOrPair.AY, false)
+                    val ptrName = asmgen.asmVariableName(addrExpr.left as PtIdentifier)
+                    asmgen.out("""
+                        clc
+                        adc  $ptrName
+                        sta  P8ZP_SCRATCH_W2
+                        tya
+                        adc  $ptrName+1
+                        sta  P8ZP_SCRATCH_W2+1""")
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)      // TODO *could* overwerite SCRATCH_W2 if it's a compliated expression...
+                    asmgen.storeIndirectByteReg(CpuRegister.A, "P8ZP_SCRATCH_W2", 0u, false, false)
+                    return
                 }
             }
             else -> { /* fall through */ }
@@ -871,6 +1032,20 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                             iny
                             sta  ($varname),y""")
                     return
+                } else if(addrExpr.operator=="+" && addrExpr.left is PtIdentifier) {
+                    asmgen.assignExpressionToRegister(addrExpr.right, RegisterOrPair.AY, false)
+                    val ptrName = asmgen.asmVariableName(addrExpr.left as PtIdentifier)
+                    asmgen.out("""
+                        clc
+                        adc  $ptrName
+                        sta  P8ZP_SCRATCH_W2
+                        tya
+                        adc  $ptrName+1
+                        sta  P8ZP_SCRATCH_W2+1""")
+                    asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.AY)      // TODO *could* overwerite SCRATCH_W2 if it's a compliated expression...
+                    asmgen.out("  jsr  prog8_lib.func_pokew_scratchW2")
+                    //asmgen.storeIndirectWordReg(RegisterOrPair.AY, "P8ZP_SCRATCH_W2", 0u)
+                    return
                 }
             }
             else -> { /* fall through */ }
@@ -881,6 +1056,18 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         asmgen.out("  jsr  prog8_lib.func_pokew")
     }
 
+    private fun funcPokeL(fcall: PtBuiltinFunctionCall) {
+        // TODO optimize for the simple cases
+        asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+        asmgen.saveRegisterStack(CpuRegister.A, false)
+        asmgen.saveRegisterStack(CpuRegister.Y, false)
+        // it's a statement so no need to preserve R14:R15
+        asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.R14R15_32, true)
+        asmgen.restoreRegisterStack(CpuRegister.Y, false)
+        asmgen.restoreRegisterStack(CpuRegister.A, false)
+        asmgen.out("  jsr  prog8_lib.func_pokel")
+    }
+
     private fun funcPeekF(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
         asmgen.out("  jsr  floats.MOVFM")
@@ -915,6 +1102,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     val varname = asmgen.asmVariableName(pointer)
                     asmgen.assignExpressionToRegister(result.second, RegisterOrPair.Y)
                     asmgen.out("  lda  ($varname),y")
+                } else if(addrExpr.operator=="+" && addrExpr.left is PtIdentifier) {
+                    readValueFromPointerPlusOffset(addrExpr.left as PtIdentifier, addrExpr.right, BaseDataType.BOOL)
                 } else fallback()
             }
             else -> fallback()
@@ -960,6 +1149,8 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                         lda  ($varname),y
                         tay
                         txa""")
+                } else if(addrExpr.operator=="+" && addrExpr.left is PtIdentifier) {
+                    readValueFromPointerPlusOffset(addrExpr.left as PtIdentifier, addrExpr.right, BaseDataType.UWORD)
                 } else fallback()
             }
             else -> fallback()
@@ -977,6 +1168,36 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         }
     }
 
+    private fun readValueFromPointerPlusOffset(ptr: PtIdentifier, offset: PtExpression, dt: BaseDataType) {
+        val varname = asmgen.asmVariableName(ptr)
+        asmgen.assignExpressionToRegister(offset, RegisterOrPair.AY)
+        asmgen.out("""
+            clc
+            adc  $varname
+            sta  P8ZP_SCRATCH_W1
+            tya
+            adc  $varname+1
+            sta  P8ZP_SCRATCH_W1+1""")
+        if (dt.isByteOrBool) {
+            if(asmgen.isTargetCpu(CpuType.CPU65C02)) {
+                asmgen.out("  lda  (P8ZP_SCRATCH_W1)")
+            } else {
+                asmgen.out("  ldy  #0 |  lda  (P8ZP_SCRATCH_W1),y")
+            }
+        } else if(dt.isWord) {
+            asmgen.out("  jsr  prog8_lib.func_peekw.from_scratchW1")
+        } else throw AssemblyError("unsupported type for peek $dt")
+    }
+
+    private fun funcPeekL(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        // TODO optimize for the simple cases
+        asmgen.assignExpressionToRegister(fcall.args[0], RegisterOrPair.AY)
+        asmgen.out("  jsr  prog8_lib.func_peekl")
+        val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.R14R15_32, true, fcall.position, fcall.definingISub(), asmgen)
+        assignAsmGen.assignRegisterLong(targetReg, RegisterOrPair.R14R15_32)
+    }
+
+
     private fun funcClamp(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type.isSigned
         when {
@@ -996,6 +1217,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
                 assignAsmGen.assignRegisterpairWord(targetReg, RegisterOrPair.AY)
             }
+            fcall.type.isLong -> {
+                TODO("clamp long  ${fcall.position}")
+            }
             else -> throw AssemblyError("invalid dt")
         }
     }
@@ -1052,6 +1276,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
                 asmgen.assignRegister(RegisterOrPair.AY, targetReg)
             }
+            fcall.type.isLong -> {
+                TODO("min long  ${fcall.position}")
+            }
             else -> {
                 throw AssemblyError("min float not supported")
             }
@@ -1060,55 +1287,109 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
 
     private fun funcMax(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val signed = fcall.type.isSigned
-        if(fcall.type.isByte) {
-            asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_B1", fcall.type)     // left
-            asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)          // right
-            asmgen.out("  cmp  P8ZP_SCRATCH_B1")
-            if(signed) asmgen.out("  bpl  +") else asmgen.out("  bcs  +")
-            asmgen.out("""
-                lda  P8ZP_SCRATCH_B1
-+""")
-            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.A, signed, fcall.position, fcall.definingISub(), asmgen)
-            asmgen.assignRegister(RegisterOrPair.A, targetReg)
-        } else if(fcall.type.isWord) {
-            asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
-            asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
-            if(signed) {
+        when {
+            fcall.type.isByte -> {
+                asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_B1", fcall.type)     // left
+                asmgen.assignExpressionToRegister(fcall.args[1], RegisterOrPair.A)          // right
+                asmgen.out("  cmp  P8ZP_SCRATCH_B1")
+                if (signed) asmgen.out("  bpl  +") else asmgen.out("  bcs  +")
                 asmgen.out("""
-                    lda  P8ZP_SCRATCH_W1
-                    ldy  P8ZP_SCRATCH_W1+1
-                    cmp  P8ZP_SCRATCH_W2
-                    tya
-                    sbc  P8ZP_SCRATCH_W2+1
-                    bvc  +
-                    eor  #$80
-+                   bmi  +
-                    lda  P8ZP_SCRATCH_W1                   
-                    ldy  P8ZP_SCRATCH_W1+1
-                    jmp  ++
-+                   lda  P8ZP_SCRATCH_W2
-                    ldy  P8ZP_SCRATCH_W2+1
-+""")
-            } else {
-                asmgen.out("""
-                    lda  P8ZP_SCRATCH_W1+1
-                    cmp  P8ZP_SCRATCH_W2+1
-                    bcc  ++
-                    bne  +
-                    lda  P8ZP_SCRATCH_W1
-                    cmp  P8ZP_SCRATCH_W2
-                    bcc  ++
-+                   lda  P8ZP_SCRATCH_W1
-                    ldy  P8ZP_SCRATCH_W1+1
-                    jmp  ++
-+                   lda  P8ZP_SCRATCH_W2
-                    ldy  P8ZP_SCRATCH_W2+1
-+""")
+                    lda  P8ZP_SCRATCH_B1
++"""
+                )
+                val targetReg = AsmAssignTarget.fromRegisters(
+                    resultRegister ?: RegisterOrPair.A,
+                    signed,
+                    fcall.position,
+                    fcall.definingISub(),
+                    asmgen
+                )
+                asmgen.assignRegister(RegisterOrPair.A, targetReg)
+            }
+            fcall.type.isWord -> {
+                asmgen.assignExpressionToVariable(fcall.args[0], "P8ZP_SCRATCH_W1", fcall.type)     // left
+                asmgen.assignExpressionToVariable(fcall.args[1], "P8ZP_SCRATCH_W2", fcall.type)     // right
+                if (signed) {
+                    asmgen.out("""
+                        lda  P8ZP_SCRATCH_W1
+                        ldy  P8ZP_SCRATCH_W1+1
+                        cmp  P8ZP_SCRATCH_W2
+                        tya
+                        sbc  P8ZP_SCRATCH_W2+1
+                        bvc  +
+                        eor  #$80
++                       bmi  +
+                        lda  P8ZP_SCRATCH_W1                   
+                        ldy  P8ZP_SCRATCH_W1+1
+                        jmp  ++
++                       lda  P8ZP_SCRATCH_W2
+                        ldy  P8ZP_SCRATCH_W2+1
++""")
+                } else {
+                    asmgen.out("""
+                        lda  P8ZP_SCRATCH_W1+1
+                        cmp  P8ZP_SCRATCH_W2+1
+                        bcc  ++
+                        bne  +
+                        lda  P8ZP_SCRATCH_W1
+                        cmp  P8ZP_SCRATCH_W2
+                        bcc  ++
++                       lda  P8ZP_SCRATCH_W1
+                        ldy  P8ZP_SCRATCH_W1+1
+                        jmp  ++
++                       lda  P8ZP_SCRATCH_W2
+                        ldy  P8ZP_SCRATCH_W2+1
++""")
+                }
+                val targetReg = AsmAssignTarget.fromRegisters(
+                    resultRegister ?: RegisterOrPair.AY,
+                    signed,
+                    fcall.position,
+                    fcall.definingISub(),
+                    asmgen
+                )
+                asmgen.assignRegister(RegisterOrPair.AY, targetReg)
+            }
+            fcall.type.isLong -> {
+                TODO("max long  ${fcall.position}")
+            }
+            else -> {
+                throw AssemblyError("max float not supported")
+            }
+        }
+    }
+
+    private fun funcMklong(fcall: PtBuiltinFunctionCall) {
+        // result long in R14:R15   (r14=lsw, r15=msw)
+
+        fun isArgRegister(expression: PtExpression, reg: RegisterOrPair): Boolean {
+            if(expression !is PtIdentifier)
+                return false
+            return expression.name.startsWith("cx16.${reg.name.lowercase()}")
+        }
+
+        if(fcall.args.size==2) {
+            // mklong2(msw, lsw)
+            if(isArgRegister(fcall.args[0], RegisterOrPair.R14) || isArgRegister(fcall.args[0], RegisterOrPair.R15) ||
+                isArgRegister(fcall.args[1], RegisterOrPair.R14) || isArgRegister(fcall.args[1], RegisterOrPair.R15)) {
+                error("cannot use R14 and/or R15 as arguments for mklong2 because the result should go into R0:R1 ${fcall.position}")
+            } else {
+                assignAsmGen.assignExpressionToVariable(fcall.args[0], "cx16.r15", DataType.UWORD)
+                assignAsmGen.assignExpressionToVariable(fcall.args[1], "cx16.r14", DataType.UWORD)
             }
-            val targetReg = AsmAssignTarget.fromRegisters(resultRegister ?: RegisterOrPair.AY, signed, fcall.position, fcall.definingISub(), asmgen)
-            asmgen.assignRegister(RegisterOrPair.AY, targetReg)
         } else {
-            throw AssemblyError("max float not supported")
+            // mklong(msb, b2, b1, lsb)
+            if(isArgRegister(fcall.args[0], RegisterOrPair.R14) || isArgRegister(fcall.args[0], RegisterOrPair.R15) ||
+                isArgRegister(fcall.args[1], RegisterOrPair.R14) || isArgRegister(fcall.args[1], RegisterOrPair.R15) ||
+                isArgRegister(fcall.args[2], RegisterOrPair.R14) || isArgRegister(fcall.args[2], RegisterOrPair.R15) ||
+                isArgRegister(fcall.args[3], RegisterOrPair.R14) || isArgRegister(fcall.args[3], RegisterOrPair.R15)) {
+                error("cannot use R14 and/or R15 as arguments for mklong because the result should go into R14:R15 ${fcall.position}")
+            } else {
+                assignAsmGen.assignExpressionToVariable(fcall.args[0], "cx16.r15H", DataType.UBYTE)
+                assignAsmGen.assignExpressionToVariable(fcall.args[1], "cx16.r15L", DataType.UBYTE)
+                assignAsmGen.assignExpressionToVariable(fcall.args[2], "cx16.r14H", DataType.UBYTE)
+                assignAsmGen.assignExpressionToVariable(fcall.args[3], "cx16.r14L", DataType.UBYTE)
+            }
         }
     }
 
@@ -1179,10 +1460,43 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
         }
     }
 
+    private fun funcMsbLong(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        val arg = fcall.args.single()
+        if (!arg.type.isLong)
+            throw AssemblyError("msb__long requires long argument")
+        if (arg is PtNumber)
+            throw AssemblyError("msb(const) should have been const-folded away")
+
+        if (arg is PtIdentifier) {
+            val sourceName = asmgen.asmVariableName(arg)
+            when(resultRegister) {
+                null, RegisterOrPair.A -> asmgen.out("  lda  $sourceName+3")
+                RegisterOrPair.X -> asmgen.out("  ldx  $sourceName+3")
+                RegisterOrPair.Y -> asmgen.out("  ldy  $sourceName+3")
+                RegisterOrPair.AX -> asmgen.out("  lda  $sourceName+3 |  ldx  #0")
+                RegisterOrPair.AY -> asmgen.out("  lda  $sourceName+3 |  ldy  #0")
+                RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName+3 |  ldy  #0")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    if(asmgen.isTargetCpu(CpuType.CPU65C02))
+                        asmgen.out("  lda  $sourceName+3 |  sta  cx16.$regname |  stz  cx16.$regname+1")
+                    else
+                        asmgen.out("  lda  $sourceName+3 |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
+                }
+                in combinedLongRegisters -> {
+                    TODO("msb__long into long register ${fcall.position}")
+                }
+                else -> throw AssemblyError("invalid reg")
+            }
+        } else {
+            TODO("msb__long from $arg  ${fcall.position}")
+        }
+    }
+
     private fun funcMsb(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
         val arg = fcall.args.single()
         if (!arg.type.isWord)
-            throw AssemblyError("msb required word argument")
+            throw AssemblyError("msb requires word argument")
         if (arg is PtNumber)
             throw AssemblyError("msb(const) should have been const-folded away")
         if (arg is PtIdentifier) {
@@ -1201,6 +1515,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     else
                         asmgen.out("  lda  $sourceName+1 |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
                 }
+                in combinedLongRegisters -> {
+                    TODO("msb into long register ${fcall.position}")
+                }
                 else -> throw AssemblyError("invalid reg")
             }
         } else {
@@ -1277,15 +1594,21 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  sty  ${reg}L |  lda  #0 |  sta  ${reg}H |  lda  ${reg}L")
                 }
+                in combinedLongRegisters -> {
+                    TODO("msb into long register ${fcall.position}")
+                }
                 else -> throw AssemblyError("invalid reg")
             }
         }
     }
 
-    private fun funcLsb(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+    private fun funcLsb(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?, fromLong: Boolean) {
         val arg = fcall.args.single()
-        if (!arg.type.isWord)
-            throw AssemblyError("lsb required word argument")
+        if(fromLong) {
+            if (!arg.type.isLong) throw AssemblyError("lsb__long requires long")
+        } else {
+            if (!arg.type.isWord) throw AssemblyError("lsb requires word")
+        }
         if (arg is PtNumber)
             throw AssemblyError("lsb(const) should have been const-folded away")
 
@@ -1305,6 +1628,9 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     else
                         asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  lda  #0 |  sta  cx16.$regname+1")
                 }
+                in combinedLongRegisters -> {
+                    TODO("lsb into long register ${fcall.position}")
+                }
                 else -> throw AssemblyError("invalid reg")
             }
         } else {
@@ -1363,11 +1689,87 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     asmgen.assignExpressionToRegister(arg, RegisterOrPair.AY)
                     asmgen.out("  sta  ${reg}L |  ldy  #0 |  sty  ${reg}H |  cmp  #0")
                 }
+                in combinedLongRegisters -> {
+                    TODO("lsb into long register ${fcall.position}")
+                }
                 else -> throw AssemblyError("invalid reg")
             }
         }
     }
 
+    private fun funcMsw(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        val arg = fcall.args.single()
+        if (!arg.type.isLong)
+            throw AssemblyError("msw requires long argument")
+        if (arg is PtNumber)
+            throw AssemblyError("msw(const) should have been const-folded away")
+        if (arg is PtIdentifier) {
+            val sourceName = asmgen.asmVariableName(arg)
+            when(resultRegister) {
+                RegisterOrPair.AX -> asmgen.out("  lda  $sourceName+2 |  ldx  $sourceName+3")
+                null, RegisterOrPair.AY -> asmgen.out("  lda  $sourceName+2 |  ldy  $sourceName+3")
+                RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName+2 |  ldy  $sourceName+3")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    asmgen.out("  lda  $sourceName+2 |  sta  cx16.$regname |  lda  $sourceName+3 |  sta  cx16.$regname+1")
+                }
+                else -> throw AssemblyError("invalid reg")
+            }
+        } else {
+            // TODO can't efficiently preserve R14:R15 on the stack because the result of the routine is likely to be in CPU registers,
+            //      which in turn would be clobbered by popping stuff from the stack...
+            asmgen.assignExpressionToRegister(arg, RegisterOrPair.R14R15_32, true)
+            when(resultRegister) {
+                RegisterOrPair.AX -> asmgen.out("  lda  cx16.r15 |  ldx  cx16.r15+1")
+                null, RegisterOrPair.AY -> asmgen.out("  lda  cx16.r15 |  ldy  cx16.r15+1")
+                RegisterOrPair.XY -> asmgen.out("  ldx  cx16.r15 |  ldy  cx16.r15+1")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    asmgen.out("  lda  cx16.r15 |  sta  cx16.$regname |  lda  cx16.r15+1 |  sta  cx16.$regname+1")
+                }
+                else -> throw AssemblyError("invalid reg")
+            }
+        }
+    }
+
+    private fun funcLsw(fcall: PtBuiltinFunctionCall, resultRegister: RegisterOrPair?) {
+        val arg = fcall.args.single()
+        if (!arg.type.isLong)
+            throw AssemblyError("lsw requires long argument")
+        if (arg is PtNumber)
+            throw AssemblyError("lsw(const) should have been const-folded away")
+        if (arg is PtIdentifier) {
+            val sourceName = asmgen.asmVariableName(arg)
+            when(resultRegister) {
+                RegisterOrPair.AX -> asmgen.out("  lda  $sourceName |  ldx  $sourceName+1")
+                null, RegisterOrPair.AY -> asmgen.out("  lda  $sourceName |  ldy  $sourceName+1")
+                RegisterOrPair.XY -> asmgen.out("  ldx  $sourceName |  ldy  $sourceName+1")
+                in Cx16VirtualRegisters -> {
+                    val regname = resultRegister.name.lowercase()
+                    asmgen.out("  lda  $sourceName |  sta  cx16.$regname |  lda  $sourceName+1 |  sta  cx16.$regname+1")
+                }
+                else -> throw AssemblyError("invalid reg")
+            }
+        } else {
+            // TODO can't efficiently preserve R14:R15 on the stack because the result of the routine is likely to be in CPU registers,
+            //      which in turn would be clobbered by popping stuff from the stack...
+            asmgen.assignExpressionToRegister(arg, RegisterOrPair.R14R15_32, true)
+            when(resultRegister) {
+                RegisterOrPair.AX -> asmgen.out("  lda  cx16.r14 |  ldx  cx16.r14+1")
+                null, RegisterOrPair.AY -> asmgen.out("  lda  cx16.r14 |  ldy  cx16.r14+1")
+                RegisterOrPair.XY -> asmgen.out("  ldx  cx16.r14 |  ldy  cx16.r14+1")
+                in Cx16VirtualRegisters -> {
+                    if(resultRegister!=RegisterOrPair.R14) {
+                        val regname = resultRegister.name.lowercase()
+                        asmgen.out("  lda  cx16.r14 |  sta  cx16.$regname |  lda  cx16.r14+1 |  sta  cx16.$regname+1")
+                    }
+                }
+                else -> throw AssemblyError("invalid reg")
+            }
+        }
+
+    }
+
     private fun translateArguments(call: PtBuiltinFunctionCall, scope: IPtSubroutine?) {
         val signature = BuiltinFunctions.getValue(call.name)
         val callConv = signature.callConvention(call.args.map {
@@ -1410,6 +1812,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                     val varname = "prog8_lib.func_${call.name}._arg_${paramName}"
                     val src = when  {
                         conv.dt==BaseDataType.FLOAT -> getSourceForFloat(value)
+                        conv.dt==BaseDataType.LONG -> TODO("LONG argument for builtin func ${call.position}")
                         conv.dt.isPassByRef -> {
                             // put the address of the argument in AY
                             val addr = PtAddressOf(DataType.forDt(conv.dt).typeForAddressOf(false), false, value.position)
@@ -1428,6 +1831,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                 conv.reg != null -> {
                     val src = when {
                         conv.dt==BaseDataType.FLOAT -> getSourceForFloat(value)
+                        conv.dt==BaseDataType.LONG -> AsmAssignSource.fromAstSource(value, program, asmgen)
                         conv.dt.isPassByRef -> {
                             // put the address of the argument in AY
                             val addr = PtAddressOf(DataType.forDt(conv.dt).typeForAddressOf(false), false,value.position)
@@ -1439,7 +1843,7 @@ internal class BuiltinFunctionsAsmGen(private val program: PtProgram,
                             AsmAssignSource.fromAstSource(value, program, asmgen)
                         }
                     }
-                    val tgt = AsmAssignTarget.fromRegisters(conv.reg!!, false, value.position, null, asmgen)
+                    val tgt = AsmAssignTarget.fromRegisters(conv.reg!!, conv.dt.isSigned, value.position, null, asmgen)
                     val assign = AsmAssignment(src, listOf(tgt), program.memsizer, value.position)
                     asmgen.translateNormalAssignment(assign, scope)
                 }

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

@@ -505,137 +505,158 @@ $endLabel""")
             is StMemVar -> symbol.length!!
             else -> 0u
         }
-        when {
-            iterableDt.isString -> {
-                if(asmgen.options.romable) {
-                    val indexVar = asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
-                    asmgen.out("""
+
+        fun iterateStrings() {
+            if(asmgen.options.romable) {
+                val indexVar = asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
+                asmgen.out("""
                         ldy  #0
                         sty  $indexVar
 $loopLabel              lda  $iterableName,y
                         beq  $endLabel
                         sta  ${asmgen.asmVariableName(stmt.variable)}""")
-                    asmgen.translate(stmt.statements)
-                    asmgen.out("""
+                asmgen.translate(stmt.statements)
+                asmgen.out("""
                         inc  $indexVar
                         ldy  $indexVar
                         bne  $loopLabel
 $endLabel""")
-                } else {
-                    val indexVar = asmgen.makeLabel("for_index")
-                    asmgen.out("""
+            } else {
+                val indexVar = asmgen.makeLabel("for_index")
+                asmgen.out("""
                         ldy  #0
                         sty  $indexVar
 $loopLabel              lda  $iterableName,y
                         beq  $endLabel
                         sta  ${asmgen.asmVariableName(stmt.variable)}""")
-                    asmgen.translate(stmt.statements)
-                    asmgen.out("""
+                asmgen.translate(stmt.statements)
+                asmgen.out("""
                         inc  $indexVar
                         ldy  $indexVar
                         bne  $loopLabel
 $indexVar   .byte  0                        
 $endLabel""")
-                }
             }
-            iterableDt.isByteArray || iterableDt.isBoolArray -> {
-                val indexVar = if(asmgen.options.romable)
-                    asmgen.createTempVarReused(iterableDt.elementType().base, false, stmt)
-                else
-                    asmgen.makeLabel("for_index")
-                asmgen.out("""
+        }
+
+        fun iterateBytes() {
+            val indexVar = if(asmgen.options.romable)
+                asmgen.createTempVarReused(iterableDt.elementType().base, false, stmt)
+            else
+                asmgen.makeLabel("for_index")
+            asmgen.out("""
                     ldy  #0
 $loopLabel          sty  $indexVar
                     lda  $iterableName,y
                     sta  ${asmgen.asmVariableName(stmt.variable)}""")
-                asmgen.translate(stmt.statements)
-                if(numElements<=255u) {
-                    asmgen.out("""
+            asmgen.translate(stmt.statements)
+            if(numElements<=255u) {
+                asmgen.out("""
                         ldy  $indexVar
                         iny
                         cpy  #$numElements
                         beq  $endLabel
                         bne  $loopLabel""")
-                } else {
-                    // length is 256
-                    asmgen.out("""
+            } else {
+                // length is 256
+                asmgen.out("""
                         ldy  $indexVar
                         iny
                         bne  $loopLabel
                         beq  $endLabel""")
-                }
-                if(!asmgen.options.romable) {
-                    if(numElements>=16u) {
-                        // allocate index var on ZP if possible, otherwise inline
-                        val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
-                        result.fold(
-                            success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
-                            failure = { asmgen.out("$indexVar    .byte  0") }
-                        )
-                    } else {
-                        asmgen.out("$indexVar    .byte  0")
-                    }
-                }
-                asmgen.out(endLabel)
             }
-            iterableDt.isSplitWordArray -> {
-                val indexVar = if(asmgen.options.romable)
-                    asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
-                else
-                    asmgen.makeLabel("for_index")
-                val loopvarName = asmgen.asmVariableName(stmt.variable)
-                asmgen.out("""
+            if(!asmgen.options.romable) {
+                if(numElements>=16u) {
+                    // allocate index var on ZP if possible, otherwise inline
+                    val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
+                    result.fold(
+                        success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
+                        failure = { asmgen.out("$indexVar    .byte  0") }
+                    )
+                } else {
+                    asmgen.out("$indexVar    .byte  0")
+                }
+            }
+            asmgen.out(endLabel)
+        }
+
+        fun iterateSplitWords() {
+            val indexVar = if(asmgen.options.romable)
+                asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
+            else
+                asmgen.makeLabel("for_index")
+            val loopvarName = asmgen.asmVariableName(stmt.variable)
+            asmgen.out("""
                     ldy  #0
 $loopLabel          sty  $indexVar
                     lda  ${iterableName}_lsb,y
                     sta  $loopvarName
                     lda  ${iterableName}_msb,y
                     sta  $loopvarName+1""")
-                asmgen.translate(stmt.statements)
-                if(numElements<=255u) {
-                    asmgen.out("""
+            asmgen.translate(stmt.statements)
+            if(numElements<=255u) {
+                asmgen.out("""
                         ldy  $indexVar
                         iny
                         cpy  #$numElements
                         beq  $endLabel
                         bne  $loopLabel""")
-                } else {
-                    // length is 256
-                    asmgen.out("""
+            } else {
+                // length is 256
+                asmgen.out("""
                         ldy  $indexVar
                         iny
                         bne  $loopLabel
                         beq  $endLabel""")
-                }
-                if(!asmgen.options.romable) {
-                    if(numElements>=16u) {
-                        // allocate index var on ZP if possible, otherwise inline
-                        val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
-                        result.fold(
-                            success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
-                            failure = { asmgen.out("$indexVar    .byte  0") }
-                        )
-                    } else {
-                        asmgen.out("$indexVar    .byte  0")
-                    }
-                }
-                asmgen.out(endLabel)
             }
-            iterableDt.isWordArray -> {
-                val indexVar = if(asmgen.options.romable)
-                    asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
-                else
-                    asmgen.makeLabel("for_index")
-                val loopvarName = asmgen.asmVariableName(stmt.variable)
-                asmgen.out("""
+            if(!asmgen.options.romable) {
+                if(numElements>=16u) {
+                    // allocate index var on ZP if possible, otherwise inline
+                    val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
+                    result.fold(
+                        success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
+                        failure = { asmgen.out("$indexVar    .byte  0") }
+                    )
+                } else {
+                    asmgen.out("$indexVar    .byte  0")
+                }
+            }
+            asmgen.out(endLabel)
+        }
+
+        fun iterateWords(actuallyLongs: Boolean = false) {
+            val indexVar = if(asmgen.options.romable)
+                asmgen.createTempVarReused(BaseDataType.UBYTE, false, stmt)
+            else
+                asmgen.makeLabel("for_index")
+            val loopvarName = asmgen.asmVariableName(stmt.variable)
+            asmgen.out("""
                     ldy  #0
 $loopLabel          sty  $indexVar
                     lda  $iterableName,y
                     sta  $loopvarName
                     lda  $iterableName+1,y
                     sta  $loopvarName+1""")
-                asmgen.translate(stmt.statements)
-                if(numElements<=127u) {
+            if(actuallyLongs) {
+                asmgen.out("""
+                    lda  $iterableName+2,y
+                    sta  $loopvarName+2
+                    lda  $iterableName+3,y
+                    sta  $loopvarName+3""")
+            }
+            asmgen.translate(stmt.statements)
+            if(numElements<=127u) {
+                if(actuallyLongs) {
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        iny
+                        iny
+                        iny
+                        cpy  #${numElements*4u}
+                        beq  $endLabel
+                        bne  $loopLabel""")
+                } else {
                     asmgen.out("""
                         ldy  $indexVar
                         iny
@@ -643,6 +664,19 @@ $loopLabel          sty  $indexVar
                         cpy  #${numElements*2u}
                         beq  $endLabel
                         bne  $loopLabel""")
+                }
+            } else {
+                if(actuallyLongs) {
+                    // array size is 64 longs, 256 bytes
+                    asmgen.out("""
+                        ldy  $indexVar
+                        iny
+                        iny
+                        iny
+                        iny
+                        bne  $loopLabel
+                        beq  $endLabel""")
+
                 } else {
                     // array size is 128 words, 256 bytes
                     asmgen.out("""
@@ -652,23 +686,29 @@ $loopLabel          sty  $indexVar
                         bne  $loopLabel
                         beq  $endLabel""")
                 }
-                if(!asmgen.options.romable) {
-                    if(numElements>=16u) {
-                        // allocate index var on ZP if possible, otherwise inline
-                        val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
-                        result.fold(
-                            success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
-                            failure = { asmgen.out("$indexVar    .byte  0") }
-                        )
-                    } else {
-                        asmgen.out("$indexVar    .byte  0")
-                    }
+            }
+            if(!asmgen.options.romable) {
+                if(numElements>=16u) {
+                    // allocate index var on ZP if possible, otherwise inline
+                    val result = zeropage.allocate(indexVar, DataType.UBYTE, null, stmt.position, asmgen.errors)
+                    result.fold(
+                        success = { (address, _, _)-> asmgen.out("""$indexVar = $address  ; auto zp UBYTE""") },
+                        failure = { asmgen.out("$indexVar    .byte  0") }
+                    )
+                } else {
+                    asmgen.out("$indexVar    .byte  0")
                 }
-                asmgen.out(endLabel)
-            }
-            iterableDt.isFloatArray -> {
-                throw AssemblyError("for loop with floating point variables is not supported")
             }
+            asmgen.out(endLabel)
+        }
+
+        when {
+            iterableDt.isString -> iterateStrings()
+            iterableDt.isByteArray || iterableDt.isBoolArray -> iterateBytes()
+            iterableDt.isSplitWordArray -> iterateSplitWords()
+            iterableDt.isWordArray -> iterateWords()
+            iterableDt.isLongArray -> iterateWords(true)
+            iterableDt.isFloatArray -> throw AssemblyError("for loop with floating point variables is not supported")
             else -> throw AssemblyError("can't iterate over $iterableDt")
         }
         asmgen.loopEndLabels.removeLast()

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

@@ -21,7 +21,7 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
         // we consider them NOT to be optimized into (possibly different) CPU registers.
         // Just load them in whatever the register spec says.
         return when (params.size) {
-            1 -> params[0].register == null && (params[0].type.isIntegerOrBool || params[0].type.isPointer)
+            1 -> params[0].register == null && (params[0].type.isWordOrByteOrBool || params[0].type.isPointer)
             2 -> params[0].type.isByteOrBool && params[1].type.isByteOrBool && params[0].register == null && params[1].register == null
             else -> false
         }
@@ -206,8 +206,11 @@ internal class FunctionCallAsmGen(private val program: PtProgram, private val as
             is PtBuiltinFunctionCall -> {
                 if (arg.name in arrayOf("lsb", "msb", "lsw", "msw"))
                     return usesOtherRegistersWhileEvaluating(arg.args[0])
-                if (arg.name == "mkword")
+                if (arg.name == "mkword" || arg.name == "mklong2")
                     return usesOtherRegistersWhileEvaluating(arg.args[0]) || usesOtherRegistersWhileEvaluating(arg.args[1])
+                if (arg.name == "mklong")
+                    return usesOtherRegistersWhileEvaluating(arg.args[0]) || usesOtherRegistersWhileEvaluating(arg.args[1]) ||
+                           usesOtherRegistersWhileEvaluating(arg.args[2]) || usesOtherRegistersWhileEvaluating(arg.args[3])
                 return !arg.isSimple()
             }
             is PtAddressOf -> false

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

@@ -46,6 +46,7 @@ internal class IfElseAsmGen(private val program: PtProgram,
             return when {
                 rightDt.isByteOrBool -> translateIfByte(stmt, jumpAfterIf)
                 rightDt.isWord || rightDt.isPointer -> translateIfWord(stmt, compareCond, jumpAfterIf)
+                rightDt.isLong -> translateIfLong(stmt, compareCond, jumpAfterIf)
                 rightDt.isFloat -> translateIfFloat(stmt, compareCond, jumpAfterIf)
                 else -> throw AssemblyError("weird dt")
             }
@@ -575,6 +576,29 @@ internal class IfElseAsmGen(private val program: PtProgram,
         }
     }
 
+    private fun translateIfLong(stmt: PtIfElse, condition: PtBinaryExpression, jumpAfterIf: PtJump?) {
+        val constValue = condition.right.asConstInteger()
+        if(constValue==0) {
+            // optimized comparisons with zero
+            return when (condition.operator) {
+                "==" -> longEqualsZero(condition.left, false, jumpAfterIf, stmt)
+                "!=" -> longEqualsZero(condition.left, true, jumpAfterIf, stmt)
+                "<" -> longLessZero(condition.left, false, jumpAfterIf, stmt)
+                "<=" -> longLessZero(condition.left, true, jumpAfterIf, stmt)
+                ">" -> longGreaterZero(condition.left, false, jumpAfterIf, stmt)
+                ">=" -> longGreaterZero(condition.left, true, jumpAfterIf, stmt)
+                else -> throw AssemblyError("expected comparison operator")
+            }
+        }
+
+        return when (condition.operator) {
+            "==" -> longEqualsValue(condition.left, condition.right, false, jumpAfterIf, stmt)
+            "!=" -> longEqualsValue(condition.left, condition.right, true, jumpAfterIf, stmt)
+            "<", "<=", ">", ">=" -> compareLongValues(condition.left, condition.operator, condition.right, jumpAfterIf, stmt)
+            else -> throw AssemblyError("expected comparison operator")
+        }
+    }
+
     private fun translateIfWord(stmt: PtIfElse, condition: PtBinaryExpression, jumpAfterIf: PtJump?) {
         val signed = condition.left.type.isSigned
         val constValue = condition.right.asConstInteger()
@@ -606,7 +630,7 @@ internal class IfElseAsmGen(private val program: PtProgram,
     private fun wordGreaterEqualsZero(value: PtExpression, signed: Boolean, jump: PtJump?, stmt: PtIfElse) {
         // special case for word>=0
         if(signed) {
-            loadAndCmp0MSB(value)
+            loadAndCmp0MSB(value, false)
             if (jump != null)
                 translateJumpElseBodies("bpl", "bmi", jump, stmt.elseScope)
             else
@@ -619,7 +643,7 @@ internal class IfElseAsmGen(private val program: PtProgram,
     private fun wordLessZero(value: PtExpression, signed: Boolean, jump: PtJump?, stmt: PtIfElse) {
         // special case for word<0
         if(signed) {
-            loadAndCmp0MSB(value)
+            loadAndCmp0MSB(value, false)
             if (jump != null)
                 translateJumpElseBodies("bmi", "bpl", jump, stmt.elseScope)
             else
@@ -894,23 +918,31 @@ _jump                       jmp  (${target.asmLabel})
     }
 
 
-    private fun loadAndCmp0MSB(value: PtExpression) {
+    private fun loadAndCmp0MSB(value: PtExpression, long: Boolean) {
         when(value) {
             is PtArrayIndexer -> {
                 if(value.variable==null)
                     TODO("support for ptr indexing ${value.position}")
                 val varname = asmgen.asmVariableName(value.variable!!)
                 asmgen.loadScaledArrayIndexIntoRegister(value, CpuRegister.Y)
-                if(value.splitWords)
+                if(value.splitWords) {
+                    require(!long)
                     asmgen.out("  lda  ${varname}_msb,y")
+                }
+                else if(long)
+                    asmgen.out("  lda  $varname+3,y")
                 else
                     asmgen.out("  lda  $varname+1,y")
             }
             is PtIdentifier -> {
                 val varname = asmgen.asmVariableName(value)
-                asmgen.out("  lda  $varname+1")
+                if(long)
+                    asmgen.out("  lda  $varname+3")
+                else
+                    asmgen.out("  lda  $varname+1")
             }
             is PtAddressOf -> {
+                require(!long)
                 if(value.isFromArrayElement) {
                     asmgen.assignExpressionToRegister(value, RegisterOrPair.AY, true)
                     asmgen.out("  cpy  #0")
@@ -923,6 +955,7 @@ _jump                       jmp  (${target.asmLabel})
                 }
             }
             else -> {
+                require(!long)
                 asmgen.assignExpressionToRegister(value, RegisterOrPair.AY, true)
                 asmgen.out("  cpy  #0")
             }
@@ -1226,6 +1259,63 @@ _jump                       jmp  (${target.asmLabel})
         }
     }
 
+    private fun longEqualsZero(value: PtExpression, notEquals: Boolean, jump: PtJump?, stmt: PtIfElse) {
+        if(value is PtIdentifier) {
+            val varname = asmgen.asmVariableName(value)
+            asmgen.out("""
+                lda  $varname
+                ora  $varname+1
+                ora  $varname+2
+                ora  $varname+3""")
+        } else {
+            asmgen.pushLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.assignExpressionToRegister(value, RegisterOrPair.R14R15_32, value.type.isSigned)
+            asmgen.out("""
+                lda  cx16.r14
+                ora  cx16.r14+1
+                ora  cx16.r14+2
+                ora  cx16.r14+3
+                sta  P8ZP_SCRATCH_REG""")
+            asmgen.popLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.out("  lda  P8ZP_SCRATCH_REG  ; restore flags")
+        }
+        if(notEquals) {
+            if (jump != null)
+                translateJumpElseBodies("bne", "beq", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("beq", stmt)
+        } else {
+            if (jump != null)
+                translateJumpElseBodies("beq", "bne", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("bne", stmt)
+        }
+    }
+
+    private fun longLessZero(value: PtExpression, lessEquals: Boolean, jump: PtJump?, stmt: PtIfElse) {
+        if(lessEquals) {
+            TODO("long <= 0 ${value.position}")
+        } else {
+            loadAndCmp0MSB(value, true)
+            if (jump != null)
+                translateJumpElseBodies("bmi", "bpl", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("bpl", stmt)
+        }
+    }
+
+    private fun longGreaterZero(value: PtExpression, lessEquals: Boolean, jump: PtJump?, stmt: PtIfElse) {
+        if(lessEquals) {
+            TODO("long >= 0 ${value.position}")
+        } else {
+            loadAndCmp0MSB(value, true)
+            if (jump != null)
+                translateJumpElseBodies("bpl", "bmi", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("bmi", stmt)
+        }
+    }
+
     private fun wordEqualsZero(value: PtExpression, notEquals: Boolean, signed: Boolean, jump: PtJump?, stmt: PtIfElse) {
 
         // special case for (u)word == 0
@@ -1303,6 +1393,145 @@ _jump                       jmp  (${target.asmLabel})
         }
     }
 
+    private fun longEqualsValue(
+        left: PtExpression,
+        right: PtExpression,
+        notEquals: Boolean,
+        jump: PtJump?,
+        stmt: PtIfElse
+    ) {
+        val constRight = right.asConstInteger()
+        val variableRight = (right as? PtIdentifier)?.name
+
+        if(left is PtIdentifier) {
+            val leftvar = asmgen.asmVariableName(left)
+            if(constRight!=null) {
+                val hex = constRight.toLongHex()
+                asmgen.out("""
+                    lda  $leftvar
+                    cmp  #$${hex.substring(6,8)}
+                    bne  +
+                    lda  $leftvar+1
+                    cmp  #$${hex.substring(4, 6)}
+                    bne  +
+                    lda  $leftvar+2
+                    cmp  #$${hex.substring(2, 4)}
+                    bne  +
+                    lda  $leftvar+3
+                    cmp  #$${hex.take(2)}
++""")
+            } else if(variableRight!=null) {
+                require(right.type.isLong)
+                asmgen.out("""
+                    lda  #<$leftvar
+                    ldy  #>$leftvar
+                    sta  P8ZP_SCRATCH_W1
+                    sty  P8ZP_SCRATCH_W1+1
+                    lda  #<$variableRight
+                    ldy  #>$variableRight
+                    jsr  prog8_lib.long_not_equals""")
+            }
+        }
+        else
+        {
+            // TODO cannot easily preserve R14:R15 on stack because we need the status flags of the comparison in between...
+            asmgen.assignExpressionToRegister(left, RegisterOrPair.R14R15_32, left.type.isSigned)
+            if(constRight!=null) {
+                val hex = constRight.toLongHex()
+                asmgen.out("""
+                    lda  cx16.r14
+                    cmp  #$${hex.substring(6,8)}
+                    bne  +
+                    lda  cx16.r14+1
+                    cmp  #$${hex.substring(4, 6)}
+                    bne  +
+                    lda  cx16.r14+2
+                    cmp  #$${hex.substring(2, 4)}
+                    bne  +
+                    lda  cx16.r14+3
+                    cmp  #$${hex.take(2)}
++""")
+            } else if(variableRight!=null) {
+                require(right.type.isLong)
+                asmgen.out("""
+                    lda  cx16.r14
+                    cmp  $variableRight
+                    bne  +
+                    lda  cx16.r14+1
+                    cmp  $variableRight+1
+                    bne  +
+                    lda  cx16.r14+2
+                    cmp  $variableRight+2
+                    bne  +
+                    lda  cx16.r14+3
+                    cmp  $variableRight+3
++""")
+            } else {
+                TODO("long == value expression ${right.position}")
+            }
+        }
+
+        if(notEquals) {
+            if (jump != null)
+                translateJumpElseBodies("bne", "beq", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("beq", stmt)
+        } else {
+            if (jump != null)
+                translateJumpElseBodies("beq", "bne", jump, stmt.elseScope)
+            else
+                translateIfElseBodies("bne", stmt)
+        }
+    }
+
+    private fun compareLongValues(
+        left: PtExpression,
+        operator: String,
+        right: PtExpression,
+        jump: PtJump?,
+        stmt: PtIfElse
+    ) {
+        // this comparison is not part of an expression but part of an if statement, there's no need to save the previous values of the temp registers
+
+        if(left !is PtNumber && left !is PtIdentifier || right !is PtNumber && right !is PtIdentifier) {
+            TODO("long comparison $operator with expressions - use temporary long variable instead to simplify for now   ${left.position}")
+        }
+
+        if(operator=="<" || operator ==">=") {
+            assignmentAsmGen.assignExpressionToRegister(right, RegisterOrPair.R14R15_32, left.type.isSigned)
+            assignmentAsmGen.assignExpressionToRegister(left, RegisterOrPair.R12R13_32, left.type.isSigned)
+        } else {
+            // flip operands
+            assignmentAsmGen.assignExpressionToRegister(left, RegisterOrPair.R14R15_32, left.type.isSigned)
+            assignmentAsmGen.assignExpressionToRegister(right, RegisterOrPair.R12R13_32, left.type.isSigned)
+        }
+        asmgen.out("""
+            sec
+            lda  cx16.r12
+            sbc  cx16.r14
+            lda  cx16.r12+1
+            sbc  cx16.r14+1
+            lda  cx16.r12+2
+            sbc  cx16.r14+2
+            lda  cx16.r12+3
+            sbc  cx16.r14+3""")
+        when(operator) {
+            "<", ">" -> {
+                if (jump != null)
+                    translateJumpElseBodies("bmi", "bpl", jump, stmt.elseScope)
+                else
+                    translateIfElseBodies("bpl", stmt)
+            }
+            ">=", "<=" -> {
+                if (jump != null)
+                    translateJumpElseBodies("bpl", "bmi", jump, stmt.elseScope)
+                else
+                    translateIfElseBodies("bmi", stmt)
+            }
+            else -> throw RuntimeException("invalid operator $operator")
+        }
+    }
+
     private fun wordEqualsValue(
         left: PtExpression,
         right: PtExpression,
@@ -1709,7 +1938,7 @@ _jump                       jmp  (${target.asmLabel})
                     if(notEquals)
                         translateAYNotEquals("P8ZP_SCRATCH_W1", "P8ZP_SCRATCH_W1+1")
                     else
-                        translateAYNotEquals("P8ZP_SCRATCH_W1", "P8ZP_SCRATCH_W1+1")
+                        translateAYEquals("P8ZP_SCRATCH_W1", "P8ZP_SCRATCH_W1+1")
                 }
             }
         }
@@ -1889,4 +2118,5 @@ _jump                       jmp  (${target.asmLabel})
             else -> throw AssemblyError("expected comparison operator")
         }
     }
-}
+}
+

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

@@ -15,32 +15,27 @@ internal class IfExpressionAsmGen(private val asmgen: AsmGen6502Internal, privat
         val falseLabel = asmgen.makeLabel("ifexpr_false")
         val endLabel = asmgen.makeLabel("ifexpr_end")
         evalIfExpressionConditonAndBranchWhenFalse(expr.condition, falseLabel)
-        when {
-            expr.type.isByteOrBool -> {
-                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.A)
-                asmgen.jmp(endLabel)
-                asmgen.out(falseLabel)
-                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.A)
-                asmgen.out(endLabel)
-                assignmentAsmGen.assignRegisterByte(target, CpuRegister.A, false, false)
-            }
-            expr.type.isWord || expr.type.isString -> {
-                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.AY)
-                asmgen.jmp(endLabel)
-                asmgen.out(falseLabel)
-                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.AY)
-                asmgen.out(endLabel)
-                assignmentAsmGen.assignRegisterpairWord(target, RegisterOrPair.AY)
-            }
-            expr.type.isFloat -> {
-                asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.FAC1, true)
-                asmgen.jmp(endLabel)
-                asmgen.out(falseLabel)
-                asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.FAC1, true)
-                asmgen.out(endLabel)
-                asmgen.assignRegister(RegisterOrPair.FAC1, target)
-            }
-            else -> throw AssemblyError("weird dt")
+
+        if(expr.type.isByteOrBool) {
+            asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.A)
+            asmgen.jmp(endLabel)
+            asmgen.out(falseLabel)
+            asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.A)
+            asmgen.out(endLabel)
+            assignmentAsmGen.assignRegisterByte(target, CpuRegister.A, false, false)
+        } else if(expr.type.isFloat) {
+            asmgen.assignExpressionToRegister(expr.truevalue, RegisterOrPair.FAC1, true)
+            asmgen.jmp(endLabel)
+            asmgen.out(falseLabel)
+            asmgen.assignExpressionToRegister(expr.falsevalue, RegisterOrPair.FAC1, true)
+            asmgen.out(endLabel)
+            asmgen.assignRegister(RegisterOrPair.FAC1, target)
+        } else {
+            asmgen.assignExpressionTo(expr.truevalue, target)
+            asmgen.jmp(endLabel)
+            asmgen.out(falseLabel)
+            asmgen.assignExpressionTo(expr.falsevalue, target)
+            asmgen.out(endLabel)
         }
     }
 
@@ -114,6 +109,7 @@ internal class IfExpressionAsmGen(private val asmgen: AsmGen6502Internal, privat
                 return when {
                     rightDt.isByteOrBool -> translateIfExpressionByteConditionBranch(condition, falseLabel)
                     rightDt.isWord -> translateIfExpressionWordConditionBranch(condition, falseLabel)
+                    rightDt.isLong -> translateIfExpressionLongConditionBranch(condition, falseLabel)
                     rightDt.isFloat -> translateIfExpressionFloatConditionBranch(condition, falseLabel)
                     else -> throw AssemblyError("weird dt")
                 }
@@ -200,6 +196,36 @@ internal class IfExpressionAsmGen(private val asmgen: AsmGen6502Internal, privat
         asmgen.out("  beq  $falseLabel")
     }
 
+    private fun translateIfExpressionLongConditionBranch(condition: PtBinaryExpression, falseLabel: String) {
+        // TODO can we reuse this whole thing from IfElse ?
+        val constValue = condition.right.asConstInteger()
+        if(constValue!=null) {
+            if (constValue == 0) {
+                when (condition.operator) {
+                    "==" -> return translateLongExprIsZero(condition.left, falseLabel)
+                    "!=" -> return translateLongExprIsNotZero(condition.left, falseLabel)
+                }
+            }
+            if (constValue != 0) {
+                when (condition.operator) {
+                    "==" -> return translateLongExprEqualsNumber(condition.left, constValue, falseLabel)
+                    "!=" -> return translateLongExprNotEqualsNumber(condition.left, constValue, falseLabel)
+                }
+            }
+        }
+        val variable = condition.right as? PtIdentifier
+        if(variable!=null) {
+            when (condition.operator) {
+                "==" -> return translateLongExprEqualsVariable(condition.left, variable, falseLabel)
+                "!=" -> return translateLongExprNotEqualsVariable(condition.left, variable, falseLabel)
+            }
+        }
+
+        // TODO don't store condition as expression result but just use the flags, like a normal PtIfElse translation does
+        asmgen.assignConditionValueToRegisterAndTest(condition)
+        asmgen.out("  beq  $falseLabel")
+    }
+
     private fun translateIfExpressionFloatConditionBranch(condition: PtBinaryExpression, elseLabel: String) {
         val constValue = (condition.right as? PtNumber)?.number
         if(constValue==0.0) {
@@ -339,6 +365,91 @@ internal class IfExpressionAsmGen(private val asmgen: AsmGen6502Internal, privat
         }
     }
 
+    private fun translateLongExprEqualsNumber(expr: PtExpression, number: Int, falseLabel: String) {
+        // if L==number
+        // TODO reuse code from ifElse?
+        val hex = number.toLongHex()
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                cmp  #$${hex.substring(6, 8)}
+                bne  $falseLabel
+                lda  $varname+1
+                cmp  #$${hex.substring(4, 6)}
+                bne  $falseLabel
+                lda  $varname+2
+                cmp  #$${hex.substring(2, 4)}
+                bne  $falseLabel
+                lda  $varname+3
+                cmp  #$${hex.take(2)}
+                bne  $falseLabel""")
+        } else {
+            // TODO cannot easily preserve R14:R15 on stack because we need the status flags of the comparison in between...
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.R14R15_32, expr.type.isSigned)
+            asmgen.out("""
+                lda  cx16.r14
+                cmp  #$${hex.substring(6, 8)}
+                bne  $falseLabel
+                lda  cx16.r14+1
+                cmp  #$${hex.substring(4, 6)}
+                bne  $falseLabel
+                lda  cx16.r14+2
+                cmp  #$${hex.substring(2, 4)}
+                bne  $falseLabel
+                lda  cx16.r14+3
+                cmp  #$${hex.take(2)}
+                bne  $falseLabel""")
+        }
+    }
+
+    private fun translateLongExprEqualsVariable(expr: PtExpression, variable: PtIdentifier, falseLabel: String) {
+        // if L==variable
+        // TODO reuse code from ifElse?
+        val varname2 = asmgen.asmVariableName(variable)
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                cmp  $varname2
+                bne  $falseLabel
+                lda  $varname+1
+                cmp  $varname2+1
+                bne  $falseLabel
+                lda  $varname+2
+                cmp  $varname2+2
+                bne  $falseLabel
+                lda  $varname+3
+                cmp  $varname2+3
+                bne  $falseLabel""")
+        } else {
+            // TODO cannot easily preserve R14:R15 on stack because we need the status flags of the comparison in between...
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.R14R15_32, expr.type.isSigned)
+            asmgen.out("""
+                lda  cx16.r14
+                cmp  $varname2
+                bne  $falseLabel
+                lda  cx16.r14+1
+                cmp  $varname2+1
+                bne  $falseLabel
+                lda  cx16.r14+2
+                cmp  $varname2+2
+                bne  $falseLabel
+                lda  cx16.r14+3
+                cmp  $varname2+3
+                bne  $falseLabel""")
+        }
+    }
+
+    private fun translateLongExprNotEqualsNumber(expr: PtExpression, number: Int, falseLabel: String) {
+        TODO("if expression LONG != number ${expr.position}")
+    }
+
+    private fun translateLongExprNotEqualsVariable(expr: PtExpression, variable: PtIdentifier, falseLabel: String) {
+        // if L!=variable
+        TODO("if expression LONG != variable ${expr.position}")
+    }
+
     private fun translateWordExprIsNotZero(expr: PtExpression, falseLabel: String) {
         // if w!=0
         // TODO reuse code from ifElse?
@@ -369,6 +480,56 @@ internal class IfExpressionAsmGen(private val asmgen: AsmGen6502Internal, privat
         }
     }
 
+    private fun translateLongExprIsZero(expr: PtExpression, falseLabel: String) {
+        // if L==0
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                ora  $varname+1
+                ora  $varname+2
+                ora  $varname+3
+                bne  $falseLabel""")
+        } else {
+            asmgen.pushLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.R14R15_32, expr.type.isSigned)
+            asmgen.out("""
+                lda  cx16.r14
+                ora  cx16.r14+1
+                ora  cx16.r14+2
+                ora  cx16.r14+3
+                sta  P8ZP_SCRATCH_REG""")
+            asmgen.popLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.out("  lda  P8ZP_SCRATCH_REG |  bne  $falseLabel")
+        }
+    }
+
+    private fun translateLongExprIsNotZero(expr: PtExpression, falseLabel: String) {
+        // if L!=0
+        // TODO reuse code from ifElse?
+        if(expr is PtIdentifier) {
+            val varname = asmgen.asmVariableName(expr)
+            asmgen.out("""
+                lda  $varname
+                ora  $varname+1
+                ora  $varname+2
+                ora  $varname+3
+                beq  $falseLabel""")
+        } else {
+            asmgen.pushLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.assignExpressionToRegister(expr, RegisterOrPair.R14R15_32, expr.type.isSigned)
+            asmgen.out("""
+                lda  cx16.r14
+                ora  cx16.r14+1
+                ora  cx16.r14+2
+                ora  cx16.r14+3
+                sta  P8ZP_SCRATCH_REG""")
+            asmgen.popLongRegisters(RegisterOrPair.R14R15_32, 1)
+            asmgen.out("  lda  P8ZP_SCRATCH_REG |  beq  $falseLabel")
+        }
+    }
+
     private fun translateIfCompareWithZeroByteBranch(condition: PtBinaryExpression, signed: Boolean, falseLabel: String) {
         // optimized code for byte comparisons with 0
 

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

@@ -210,7 +210,7 @@ internal class ProgramAndVarsGen(
     private fun memorySlabs() {
         if(symboltable.allMemorySlabs.isNotEmpty()) {
             asmgen.out("; memory slabs\n  .section BSS_SLABS")
-            asmgen.out("prog8_slabs\t.block")
+            asmgen.out("$StMemorySlabBlockName\t.block")
             for (slab in symboltable.allMemorySlabs) {
                 if (slab.align > 1u)
                     asmgen.out("\t.align  ${slab.align.toHex()}")
@@ -377,6 +377,7 @@ internal class ProgramAndVarsGen(
             dt.isSignedByte -> ".char"
             dt.isUnsignedWord || dt.isPointer -> ".word"
             dt.isSignedWord -> ".sint"
+            dt.isLong -> ".dint"
             dt.isFloat -> ".byte"
             else -> {
                 throw AssemblyError("weird dt")
@@ -418,14 +419,7 @@ internal class ProgramAndVarsGen(
             structtype.fields.withIndex().forEach { (index, field) ->
                 val dt = field.first
                 val varname = "f${index}"
-                val type = when {
-                    dt.isBool || dt.isUnsignedByte -> ".byte"
-                    dt.isSignedByte -> ".char"
-                    dt.isUnsignedWord || dt.isPointer -> ".word"
-                    dt.isSignedWord -> ".sint"
-                    dt.isFloat -> ".byte"   // TODO check that float bytes are passed as an array parameter
-                    else -> throw AssemblyError("weird dt")
-                }
+                val type = asmTypeString(dt)
                 asmgen.out("p8v_${field.second}  $type  \\$varname")        // note: struct field symbol prefixing done here because that is a lot simpler than fixing up all expressions in the AST
             }
             asmgen.out("    .endstruct\n")
@@ -434,6 +428,7 @@ internal class ProgramAndVarsGen(
         val (instancesNoInit, instances) = symboltable.allStructInstances.partition { it.initialValues.isEmpty() }
         asmgen.out("; struct instances without initialization values, as BSS zeroed at startup\n")
         asmgen.out("    .section BSS\n")
+        asmgen.out("${StStructInstanceBlockName}_bss  .block\n")
         instancesNoInit.forEach {
             val structtype: StStruct = symboltable.lookup(it.structName) as StStruct
             val zerovalues = structtype.fields.map { field ->
@@ -445,11 +440,17 @@ internal class ProgramAndVarsGen(
             }
             asmgen.out("${it.name}    .dstruct  ${it.structName}, ${zerovalues.joinToString(",")}\n")
         }
+        asmgen.out("    .endblock\n")
         asmgen.out("    .send BSS\n")
 
         asmgen.out("; struct instances with initialization values\n")
         asmgen.out("    .section STRUCTINSTANCES\n")
-        instances.forEach { asmgen.out("${it.name}    .dstruct  ${it.structName}, ${initValues(it).joinToString(",")}\n") }
+        asmgen.out("$StStructInstanceBlockName  .block\n")
+        instances.forEach {
+            val instancename = it.name.substringAfter('.')
+            asmgen.out("$instancename    .dstruct  ${it.structName}, ${initValues(it).joinToString(",")}\n")
+        }
+        asmgen.out("    .endblock\n")
         asmgen.out("    .send STRUCTINSTANCES\n")
     }
 
@@ -564,6 +565,7 @@ internal class ProgramAndVarsGen(
             else when(dt) {
                 BaseDataType.UBYTE, BaseDataType.BYTE, BaseDataType.BOOL -> asmgen.out("$name    .byte  ?")
                 BaseDataType.UWORD, BaseDataType.WORD -> asmgen.out("$name    .word  ?")
+                BaseDataType.LONG -> asmgen.out("$name    .sint  ?")
                 BaseDataType.FLOAT -> asmgen.out("$name    .fill  ${options.compTarget.FLOAT_MEM_SIZE}")
                 else -> throw AssemblyError("weird dt for extravar $dt")
             }
@@ -774,6 +776,7 @@ internal class ProgramAndVarsGen(
             dt.isSignedByte -> asmgen.out("${variable.name}\t.char  ?")
             dt.isUnsignedWord -> asmgen.out("${variable.name}\t.word  ?")
             dt.isSignedWord -> asmgen.out("${variable.name}\t.sint  ?")
+            dt.isLong -> asmgen.out("${variable.name}\t.dint  ?")
             dt.isFloat -> asmgen.out("${variable.name}\t.fill  ${compTarget.FLOAT_MEM_SIZE}")
             dt.isSplitWordArray -> {
                 alignVar(variable.align)
@@ -788,7 +791,7 @@ internal class ProgramAndVarsGen(
             }
             dt.isPointer -> asmgen.out("${variable.name}\t.word  ?")        // a pointer is just an uword address
             dt.isPointerArray -> {
-                TODO("pointers are not supported yet")
+                TODO("pointers are not supported yet for uninitialized array ${variable.astNode?.position}")
             }
             else -> {
                 throw AssemblyError("weird dt")
@@ -822,6 +825,7 @@ internal class ProgramAndVarsGen(
 //            dt.isSignedByte -> asmgen.out("${variable.name}\t.char  $initialValue")
 //            dt.isUnsignedWord -> asmgen.out("${variable.name}\t.word  ${initialValue.toHex()}")
 //            dt.isSignedWord -> asmgen.out("${variable.name}\t.sint  $initialValue")
+//            dt.isLong -> asmgen.out("${variable.name}\t.dint  $initialValue")
 //            dt.isFloat -> {
 //                if(initialValue==0) {
 //                    asmgen.out("${variable.name}\t.byte  0,0,0,0,0  ; float")
@@ -866,7 +870,7 @@ internal class ProgramAndVarsGen(
                 }
             }
             dt.isSplitWordArray -> {
-                if(dt.elementType().isUnsignedWord) {
+                if(dt.elementType().isUnsignedWord || dt.elementType().isPointer) {
                     val data = makeArrayFillDataUnsigned(dt, value, orNumberOfZeros)
                     asmgen.out("_array_$varname := ${data.joinToString()}")
                     asmgen.out("${varname}_lsb\t.byte <_array_$varname")
@@ -898,6 +902,16 @@ internal class ProgramAndVarsGen(
                         asmgen.out("  .sint  " + chunk.joinToString())
                 }
             }
+            dt.isLongArray -> {
+                val data = makeArrayFillDataSigned(dt, value, orNumberOfZeros)
+                if (data.size <= 16)
+                    asmgen.out("$varname\t.dint  ${data.joinToString()}")
+                else {
+                    asmgen.out(varname)
+                    for (chunk in data.chunked(16))
+                        asmgen.out("  .dint  " + chunk.joinToString())
+                }
+            }
             dt.isFloatArray -> {
                 val array = value ?: zeroFilledArray(orNumberOfZeros!!)
                 val floatFills = array.map {
@@ -914,7 +928,7 @@ internal class ProgramAndVarsGen(
     private fun zeroFilledArray(numElts: Int): StArray {
         val values = mutableListOf<StArrayElement>()
         repeat(numElts) {
-            values.add(StArrayElement(0.0, null, null))
+            values.add(StArrayElement(0.0, null, null,null,null))
         }
         return values
     }
@@ -924,7 +938,7 @@ internal class ProgramAndVarsGen(
             asmgen.out("  ${it.name} = ${it.address.toHex()}")
         }
         consts.sortedBy { it.name }.forEach {
-            if(it.dt==BaseDataType.FLOAT)
+            if(it.dt.isFloat)
                 asmgen.out("  ${it.name} = ${it.value}")
             else
                 asmgen.out("  ${it.name} = ${it.value.toHex()}")
@@ -972,7 +986,7 @@ internal class ProgramAndVarsGen(
                     val number = it.number!!.toInt()
                     "$"+number.toString(16).padStart(2, '0')
                 }
-            dt.isArray && dt.elementType().isUnsignedWord -> array.map {
+            dt.isArray && (dt.elementType().isUnsignedWord || dt.elementType().isPointer) -> array.map {
                 if(it.number!=null) {
                     "$" + it.number!!.toInt().toString(16).padStart(4, '0')
                 }
@@ -984,8 +998,15 @@ internal class ProgramAndVarsGen(
                     else
                         asmgen.asmSymbolName(addrOfSymbol)
                 }
-                else
+                else if(it.structInstance!=null) {
+                    asmgen.asmSymbolName("${StStructInstanceBlockName}.${it.structInstance!!}")
+                }
+                else if(it.structInstanceUninitialized!=null) {
+                    asmgen.asmSymbolName("${StStructInstanceBlockName}_bss.${it.structInstanceUninitialized!!}")
+                }
+                else {
                     throw AssemblyError("weird array elt")
+                }
             }
             else -> throw AssemblyError("invalid dt")
         }
@@ -1013,7 +1034,7 @@ internal class ProgramAndVarsGen(
                 val number = it.number!!.toInt()
                 "$" + number.toString(16).padStart(4, '0')
             }
-            dt.isArray && dt.elementType().isSignedWord -> array.map {
+            dt.isSignedWordArray -> array.map {
                 val number = it.number!!.toInt()
                 val hexnum = number.absoluteValue.toString(16).padStart(4, '0')
                 if(number>=0)
@@ -1021,6 +1042,14 @@ internal class ProgramAndVarsGen(
                 else
                     "-$$hexnum"
             }
+            dt.isLongArray -> array.map {
+                val number = it.number!!.toInt()
+                val hexnum = number.absoluteValue.toLongHex()
+                if(number>=0)
+                    "$$hexnum"
+                else
+                    "-$$hexnum"
+            }
             else -> throw AssemblyError("invalid dt")
         }
     }

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

@@ -42,6 +42,12 @@ internal class AnyExprAsmGen(
                 }
                 throw AssemblyError("expression should have been handled otherwise: word ${expr.operator} at ${expr.position}")
             }
+            expr.type.isLong -> {
+                require(expr.left.type.isLong && expr.right.type.isLong) {
+                    "both operands must be longs"
+                }
+                throw AssemblyError("expression should have been handled otherwise: long ${expr.operator} at ${expr.position}")
+            }
             expr.type.isFloat -> {
                 require(expr.left.type.isFloat && expr.right.type.isFloat) {
                     "both operands must be floats"
@@ -179,7 +185,7 @@ internal class AnyExprAsmGen(
                 asmgen.assignRegister(RegisterOrPair.A, assign.target)
                 return true
             }
-            else -> TODO("float expression operator ${expr.operator}")
+            else -> TODO("float expression operator ${expr.operator}  ${expr.position}")
         }
     }
 

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

@@ -10,8 +10,8 @@ internal enum class TargetStorageKind {
     ARRAY,
     MEMORY,
     REGISTER,
-    POINTER,        // wherever the pointer variable points to
-    VOID            // assign nothing - used in multi-value assigns for void placeholders
+    POINTER,
+    VOID              // assign nothing - used in multi-value assigns for void placeholders
 }
 
 internal enum class SourceStorageKind {
@@ -43,7 +43,7 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
             variableAsmName!!
         else {
             if(array.variable==null)
-                TODO("asmVarname for array with pointer")
+                TODO("asmVarname for array with pointer $position")
             asmgen.asmVariableName(array.variable!!)
         }
     }
@@ -114,25 +114,17 @@ internal class AsmAssignTarget(val kind: TargetStorageKind,
                     RegisterOrPair.FAC2 -> {
                         AsmAssignTarget(TargetStorageKind.REGISTER, asmgen, DataType.FLOAT, scope, pos, 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 -> {
+                    in Cx16VirtualRegisters -> {
                         val dt = if(signed) DataType.WORD else DataType.UWORD
                         AsmAssignTarget(TargetStorageKind.REGISTER, asmgen, dt, scope, pos, register = registers)
                     }
+                    in combinedLongRegisters -> {
+                        val dt = if(signed) DataType.LONG
+                            else
+                                TODO("unsigned long $pos")
+                        AsmAssignTarget(TargetStorageKind.REGISTER, asmgen, dt, scope, pos, register = registers)
+                    }
+                    else -> throw AssemblyError("weird register $registers")
                 }
     }
 

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

@@ -4,6 +4,7 @@ import prog8.code.ast.*
 import prog8.code.core.*
 import prog8.codegen.cpu6502.AsmGen6502Internal
 import prog8.codegen.cpu6502.VariableAllocator
+import prog8.codegen.cpu6502.toLongHex
 
 
 internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
@@ -106,13 +107,24 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                             }
                         }
                     }
+                    target.datatype.isLong -> {
+                        when(value.kind) {
+                            SourceStorageKind.LITERALBOOLEAN -> inplacemodificationLongWithLiteralval(target.asmVarname, operator, value.boolean!!.asInt())
+                            SourceStorageKind.LITERALNUMBER -> inplacemodificationLongWithLiteralval(target.asmVarname, operator, value.number!!.number.toInt())
+                            SourceStorageKind.VARIABLE -> inplacemodificationLongWithVariable(target.asmVarname, operator, value.asmVarname)
+                            SourceStorageKind.EXPRESSION -> inplacemodificationLongWithExpression(target.asmVarname, operator, value.expression!!)
+                            SourceStorageKind.REGISTER -> TODO("32 bits register inplace modification? ${target.position}")
+                            SourceStorageKind.ARRAY -> TODO("inplace modify long with array ${target.position}")
+                            SourceStorageKind.MEMORY -> TODO("memread into long ${target.position}")
+                        }
+                    }
                     target.datatype.isFloat -> {
                         when(value.kind) {
                             SourceStorageKind.LITERALBOOLEAN -> inplacemodificationFloatWithLiteralval(target.asmVarname, operator, value.boolean!!.asInt().toDouble())
                             SourceStorageKind.LITERALNUMBER -> inplacemodificationFloatWithLiteralval(target.asmVarname, operator, value.number!!.number)
                             SourceStorageKind.VARIABLE -> inplacemodificationFloatWithVariable(target.asmVarname, operator, value.asmVarname)
                             SourceStorageKind.REGISTER -> inplacemodificationFloatWithVariable(target.asmVarname, operator, regName(value))
-                            SourceStorageKind.MEMORY -> TODO("memread into float")
+                            SourceStorageKind.MEMORY -> TODO("memread into float ${target.position}")
                             SourceStorageKind.ARRAY -> inplacemodificationFloatWithValue(target.asmVarname, operator, value.array!!)
                             SourceStorageKind.EXPRESSION -> {
                                 if(value.expression is PtTypeCast) {
@@ -156,7 +168,7 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                             SourceStorageKind.LITERALNUMBER -> inplacemodificationBytePointerWithLiteralval(pointer, operator, value.number!!.number.toInt())
                             SourceStorageKind.VARIABLE -> inplacemodificationBytePointerWithVariable(pointer, operator, value.asmVarname)
                             SourceStorageKind.REGISTER -> inplacemodificationBytePointerWithVariable(pointer, operator, regName(value))
-                            SourceStorageKind.MEMORY -> TODO("memread into pointer")
+                            SourceStorageKind.MEMORY -> TODO("memread into pointer ${target.position}")
                             SourceStorageKind.ARRAY -> inplacemodificationBytePointerWithValue(pointer, operator, value.array!!)
                             SourceStorageKind.EXPRESSION -> {
                                 if(value.expression is PtTypeCast) {
@@ -291,13 +303,32 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                             }
                         }
 
+                        target.datatype.isLong -> {
+                            when(value.kind) {
+                                SourceStorageKind.LITERALBOOLEAN -> inplacemodificationLongWithLiteralval(targetVarName, operator, value.boolean!!.asInt())
+                                SourceStorageKind.LITERALNUMBER -> inplacemodificationLongWithLiteralval(targetVarName, operator, value.number!!.number.toInt())
+                                SourceStorageKind.VARIABLE -> inplacemodificationLongWithVariable(targetVarName, operator, value.asmVarname)
+                                SourceStorageKind.REGISTER -> inplacemodificationLongWithVariable(targetVarName, operator, regName(value))
+                                SourceStorageKind.MEMORY -> TODO("inplace long modifiication ${target.position}")
+                                SourceStorageKind.ARRAY -> TODO("inplace long modifiication ${target.position}")
+                                SourceStorageKind.EXPRESSION -> {
+                                    if(value.expression is PtTypeCast) {
+                                        if (tryInplaceModifyWithRemovedRedundantCast(value.expression, target, operator)) return
+                                        TODO("inplace long modifiication ${target.position}")
+                                    } else {
+                                        TODO("inplace long modifiication ${target.position}")
+                                    }
+                                }
+                            }
+                        }
+
                         target.datatype.isFloat -> {
                             when(value.kind) {
                                 SourceStorageKind.LITERALBOOLEAN -> inplacemodificationFloatWithLiteralval(targetVarName, operator, value.boolean!!.asInt().toDouble())
                                 SourceStorageKind.LITERALNUMBER -> inplacemodificationFloatWithLiteralval(targetVarName, operator, value.number!!.number)
                                 SourceStorageKind.VARIABLE -> inplacemodificationFloatWithVariable(targetVarName, operator, value.asmVarname)
                                 SourceStorageKind.REGISTER -> inplacemodificationFloatWithVariable(targetVarName, operator, regName(value))
-                                SourceStorageKind.MEMORY -> TODO("memread into float array")
+                                SourceStorageKind.MEMORY -> TODO("memread into float array ${target.position}")
                                 SourceStorageKind.ARRAY -> inplacemodificationFloatWithValue(targetVarName, operator, value.array!!)
                                 SourceStorageKind.EXPRESSION -> {
                                     if(value.expression is PtTypeCast) {
@@ -488,7 +519,7 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                                 SourceStorageKind.LITERALNUMBER -> inplacemodificationFloatWithLiteralval(tempvar, operator, value.number!!.number)
                                 SourceStorageKind.VARIABLE -> inplacemodificationFloatWithVariable(tempvar, operator, value.asmVarname)
                                 SourceStorageKind.REGISTER -> inplacemodificationFloatWithVariable(tempvar, operator, regName(value))
-                                SourceStorageKind.MEMORY -> TODO("memread into float")
+                                SourceStorageKind.MEMORY -> TODO("memread into float ${target.position}")
                                 SourceStorageKind.ARRAY -> inplacemodificationFloatWithValue(tempvar, operator, value.array!!)
                                 SourceStorageKind.EXPRESSION -> {
                                     if(value.expression is PtTypeCast) {
@@ -523,6 +554,449 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
         }
     }
 
+    internal fun inplacemodificationLongWithExpression(targetVar: String, operator: String, value: PtExpression) {
+        // it's not an expression so no need to preserve R14:R15
+        assignmentAsmGen.assignExpressionToRegister(value, RegisterOrPair.R14R15_32, value.type.isSigned)
+        inplacemodificationLongWithVariable(targetVar, operator, "cx16.r14")
+    }
+
+    internal fun inplacemodificationLongWithVariable(targetVar: String, operator: String, sourceVar: String) {
+        when(operator) {
+            "+" -> {
+                asmgen.out("""
+                    clc
+                    lda  $targetVar
+                    adc  $sourceVar
+                    sta  $targetVar
+                    lda  $targetVar+1
+                    adc  $sourceVar+1
+                    sta  $targetVar+1
+                    lda  $targetVar+2
+                    adc  $sourceVar+2
+                    sta  $targetVar+2
+                    lda  $targetVar+3
+                    adc  $sourceVar+3
+                    sta  $targetVar+3""")
+            }
+            "-" -> {
+                asmgen.out("""
+                    sec
+                    lda  $targetVar
+                    sbc  $sourceVar
+                    sta  $targetVar
+                    lda  $targetVar+1
+                    sbc  $sourceVar+1
+                    sta  $targetVar+1
+                    lda  $targetVar+2
+                    sbc  $sourceVar+2
+                    sta  $targetVar+2
+                    lda  $targetVar+3
+                    sbc  $sourceVar+3
+                    sta  $targetVar+3""")
+            }
+            "<<" -> {
+                asmgen.out("""
+                    ldy  $sourceVar
+-                   asl  $targetVar
+                    rol  $targetVar+1
+                    rol  $targetVar+2
+                    rol  $targetVar+3
+                    dey
+                    bne  -""")
+            }
+            ">>" -> {
+                asmgen.out("""
+                    ldy  $sourceVar
+-                   lda  $targetVar+3
+                    asl  a    ; save sign bit
+                    ror  $targetVar+3
+                    ror  $targetVar+2
+                    ror  $targetVar+1
+                    ror  $targetVar
+                    dey
+                    bne  -""")
+            }
+            "|" -> {
+                asmgen.out("""
+                    lda  $targetVar
+                    ora  $sourceVar
+                    sta  $targetVar  
+                    lda  $targetVar+1
+                    ora  $sourceVar+1
+                    sta  $targetVar+1
+                    lda  $targetVar+2
+                    ora  $sourceVar+2
+                    sta  $targetVar+2
+                    lda  $targetVar+3
+                    ora  $sourceVar+3
+                    sta  $targetVar+3""")
+            }
+            "&" -> {
+                asmgen.out("""
+                    lda  $targetVar
+                    and  $sourceVar
+                    sta  $targetVar  
+                    lda  $targetVar+1
+                    and  $sourceVar+1
+                    sta  $targetVar+1
+                    lda  $targetVar+2
+                    and  $sourceVar+2
+                    sta  $targetVar+2
+                    lda  $targetVar+3
+                    and  $sourceVar+3
+                    sta  $targetVar+3""")
+            }
+            "^" -> {
+                asmgen.out("""
+                    lda  $targetVar
+                    eor  $sourceVar
+                    sta  $targetVar  
+                    lda  $targetVar+1
+                    eor  $sourceVar+1
+                    sta  $targetVar+1
+                    lda  $targetVar+2
+                    eor  $sourceVar+2
+                    sta  $targetVar+2
+                    lda  $targetVar+3
+                    eor  $sourceVar+3
+                    sta  $targetVar+3""")
+            }
+            else -> {
+                TODO("in-place modify LONG with variable")
+            }
+        }
+    }
+
+    internal fun inplacemodificationLongWithLiteralval(variable: String, operator: String, value: Int) {
+
+        fun inplaceLongShiftLeft() {
+            when {
+                value in 0..2 -> {
+                    repeat(value) {
+                        asmgen.out("""
+                            asl  $variable
+                            rol  $variable+1
+                            rol  $variable+2
+                            rol  $variable+3""")
+                    }
+                }
+                value in 3..7 -> {
+                    asmgen.out("""
+                        ldy  #$value
+-                       asl  $variable
+                        rol  $variable+1
+                        rol  $variable+2
+                        rol  $variable+3
+                        dey
+                        bne  -""")
+                }
+                value == 8 -> {
+                    asmgen.out("""
+                        lda  $variable+2
+                        sta  $variable+3
+                        lda  $variable+1
+                        sta  $variable+2
+                        lda  $variable
+                        sta  $variable+1
+                        lda  #0
+                        sta  $variable""")
+                }
+                value in 9..15 -> {
+                    val shift = value - 8
+                    asmgen.out("""
+                        lda  $variable+2
+                        sta  $variable+3
+                        lda  $variable+1
+                        sta  $variable+2
+                        lda  $variable
+                        sta  $variable+1
+                        lda  #0
+                        sta  $variable
+                        ldy  #$shift
+-                       asl  $variable+1
+                        rol  $variable+2
+                        rol  $variable+3
+                        dey
+                        bne  -""")
+                }
+                value == 16 -> {
+                    asmgen.out("""
+                        lda  $variable+1
+                        sta  $variable+3
+                        lda  $variable
+                        sta  $variable+2
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1""")
+                }
+                value in 17..23 -> {
+                    val shift = value-16
+                    asmgen.out("""
+                        lda  $variable+1
+                        sta  $variable+3
+                        lda  $variable
+                        sta  $variable+2
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1
+                        ldy  #$shift
+-                       asl  $variable+2
+                        rol  $variable+3
+                        dey
+                        bne  -""")
+                }
+                value == 24 -> {
+                    asmgen.out("""
+                        lda  $variable
+                        sta  $variable+3
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1
+                        sta  $variable+2""")
+                }
+                value <= 31 -> {
+                    val shift = value-24
+                    asmgen.out("""
+                        lda  $variable
+                        ldy  #$shift
+-                       asl  a
+                        dey
+                        bne  -
+                        sta  $variable+3
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1
+                        sta  $variable+2""")
+                }
+                else -> {
+                    asmgen.out("""
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1
+                        sta  $variable+2
+                        sta  $variable+3""")
+                }
+            }
+        }
+
+        fun inplaceLongShiftRight() {
+            when {
+                value in 0..2 -> {
+                    repeat(value) {
+                        asmgen.out("""
+                            lda  $variable+3
+                            asl  a      ; save sign bit
+                            ror  $variable+3
+                            ror  $variable+2
+                            ror  $variable+1
+                            ror  $variable""")
+                    }
+                }
+                // TODO optimize for more cases 8, 16, 24 etc but don't forget to take the sign bit into account!
+                value <= 31 -> {
+                    asmgen.out("""
+                        ldy  #$value
+-                       lda  $variable+3
+                        asl  a      ; save sign bit
+                        ror  $variable+3
+                        ror  $variable+2
+                        ror  $variable+1
+                        ror  $variable
+                        dey
+                        bne  -""")
+                }
+                else -> {
+                    asmgen.out("""
+                        lda  #0
+                        sta  $variable
+                        sta  $variable+1
+                        sta  $variable+2
+                        sta  $variable+3""")
+                }
+            }
+        }
+
+        when(operator) {
+            "+" -> {
+                when(value) {
+                    0 -> {}
+                    1 -> {
+                        asmgen.out("""
+                            inc  $variable
+                            bne  +
+                            inc  $variable+1
+                            bne  +
+                            inc  $variable+2
+                            bne  +
+                            inc  $variable+3
++""")
+                    }
+                    else -> {
+                        if(value in 1..255) {
+                            asmgen.out("""
+                                clc
+                                lda  $variable
+                                adc  #$value
+                                sta  $variable
+                                bcc  +
+                                inc  $variable+1
+                                bne  +
+                                inc  $variable+2
+                                bne  +
+                                inc  $variable+3
++""")
+                        } else if(value in 1..65535) {
+                            asmgen.out("""
+                                clc
+                                lda  $variable
+                                adc  #<$value
+                                sta  $variable
+                                lda  $variable+1
+                                adc  #>$value
+                                sta  $variable+1
+                                bcc  +
+                                inc  $variable+2
+                                bne  +
+                                inc  $variable+3
++""")
+                        } else {
+                            val hex = value.toLongHex()
+                            asmgen.out("""
+                                clc
+                                lda  $variable
+                                adc  #$${hex.substring(6,8)}
+                                sta  $variable
+                                lda  $variable+1
+                                adc  #$${hex.substring(4, 6)}
+                                sta  $variable+1
+                                lda  $variable+2
+                                adc  #$${hex.substring(2, 4)}
+                                sta  $variable+2
+                                lda  $variable+3
+                                adc  #$${hex.take(2)}
+                                sta  $variable+3""")
+                        }
+                    }
+                }
+            }
+            "-" -> {
+                when(value) {
+                    0 -> {}
+                    1 -> {
+                        asmgen.out("""
+                            lda  $variable
+                            bne  +
+                            dec  $variable+1
+                            bne  +
+                            dec  $variable+2
+                            bne  +
+                            dec  $variable+3
++                           dec  $variable""")
+                    }
+                    else -> {
+                        if(value in 1..255) {
+                            asmgen.out("""
+                                lda  $variable
+                                sec
+                                sbc  #$value
+                                sta  $variable
+                                bcs  +
+                                dec  $variable+1
+                                bne  +
+                                dec  $variable+2
+                                bne  +
+                                dec  $variable+3
++""")
+                        } else if(value in 1..65535) {
+                            asmgen.out("""
+                                lda  $variable
+                                sec
+                                sbc  #<$value
+                                sta  $variable
+                                lda  $variable+1
+                                sbc  #>$value
+                                sta  $variable+1
+                                bcs  +
+                                dec  $variable+2
+                                bne  +
+                                dec  $variable+3
++""")
+                        } else {
+                            val hex = value.toLongHex()
+                            asmgen.out("""
+                                sec
+                                lda  $variable
+                                sbc  #$${hex.substring(6,8)}
+                                sta  $variable
+                                lda  $variable+1
+                                sbc  #$${hex.substring(4, 6)}
+                                sta  $variable+1
+                                lda  $variable+2
+                                sbc  #$${hex.substring(2, 4)}
+                                sta  $variable+2
+                                lda  $variable+3
+                                sbc  #$${hex.take(2)}
+                                sta  $variable+3""")
+                        }
+                    }
+                }
+            }
+            "<<" -> if (value > 0) inplaceLongShiftLeft()
+            ">>" -> if (value > 0) inplaceLongShiftRight()
+            "|" -> {
+                val hex = value.toLongHex()
+                asmgen.out("""
+                    lda  $variable
+                    ora  #$${hex.substring(6,8)}
+                    sta  $variable  
+                    lda  $variable+1
+                    ora  #$${hex.substring(4, 6)}
+                    sta  $variable+1
+                    lda  $variable+2
+                    ora  #$${hex.substring(2, 4)}
+                    sta  $variable+2
+                    lda  $variable+3
+                    ora  #$${hex.take(2)}
+                    sta  $variable+3""")
+            }
+            "&" -> {
+                val hex = value.toLongHex()
+                asmgen.out("""
+                    lda  $variable
+                    and  #$${hex.substring(6,8)}
+                    sta  $variable  
+                    lda  $variable+1
+                    and  #$${hex.substring(4, 6)}
+                    sta  $variable+1
+                    lda  $variable+2
+                    and  #$${hex.substring(2, 4)}
+                    sta  $variable+2
+                    lda  $variable+3
+                    and  #$${hex.take(2)}
+                    sta  $variable+3""")
+            }
+            "^" -> {
+                val hex = value.toLongHex()
+                asmgen.out("""
+                    lda  $variable
+                    eor  #$${hex.substring(6,8)}
+                    sta  $variable  
+                    lda  $variable+1
+                    eor  #$${hex.substring(4, 6)}
+                    sta  $variable+1
+                    lda  $variable+2
+                    eor  #$${hex.substring(2, 4)}
+                    sta  $variable+2
+                    lda  $variable+3
+                    eor  #$${hex.take(2)}
+                    sta  $variable+3""")
+            }
+            else -> {
+                TODO("inplace long $operator $value")
+            }
+        }
+    }
+
     private fun tryIndexedIncDec(array: PtArrayIndexer, operator: String): Boolean {
         val arrayVar = array.variable
         if(arrayVar==null) {
@@ -966,7 +1440,7 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
 +""")
             }
             // pretty uncommon, who's going to assign a comparison boolean expression to a pointer?
-            "<", "<=", ">", ">=" -> TODO("byte-var-to-pointer comparisons")
+            "<", "<=", ">", ">=" -> TODO("byte-var-to-pointer comparisons ${pointervar.position}")
             else -> throw AssemblyError("invalid operator for in-place modification $operator")
         }
         asmgen.storeAIntoZpPointerVar(sourceName, false)
@@ -1089,7 +1563,7 @@ internal class AugmentableAssignmentAsmGen(private val program: PtProgram,
                 asmgen.storeAIntoZpPointerVar(sourceName, false)
             }
             // pretty uncommon, who's going to assign a comparison boolean expression to a pointer?:
-            "<", "<=", ">", ">=" -> TODO("byte-litval-to-pointer comparisons")
+            "<", "<=", ">", ">=" -> TODO("byte-litval-to-pointer comparisons ${pointervar.position}")
             else -> throw AssemblyError("invalid operator for in-place modification $operator")
         }
     }
@@ -1851,7 +2325,7 @@ $shortcutLabel:""")
                 if(value in asmgen.optimizedWordMultiplications) {
                     asmgen.out("  lda  $lsb |  ldy  $msb |  jsr  prog8_math.mul_word_$value |  sta  $lsb |  sty  $msb")
                 } else {
-                    if(block?.options?.veraFxMuls==true)
+                    if(block?.options?.veraFxMuls==true) {
                         // cx16 verafx hardware mul
                         asmgen.out("""
                             lda  $lsb
@@ -1862,9 +2336,10 @@ $shortcutLabel:""")
                             ldy  #>$value
                             sta  cx16.r1
                             sty  cx16.r1+1
-                            jsr  verafx.muls
+                            jsr  verafx.muls16
                             sta  $lsb
                             sty  $msb""")
+                    }
                     else
                         asmgen.out("""
                             lda  $lsb
@@ -2349,9 +2824,10 @@ $shortcutLabel:""")
                                 ldy  $name+1
                                 sta  cx16.r0
                                 sty  cx16.r0+1
-                                jsr  verafx.muls
+                                jsr  verafx.muls16
                                 sta  $name
                                 sty  $name+1""")
+
                         } else {
                             if(valueDt.isUnsignedByte) {
                                 asmgen.out("  lda  $otherName |  sta  prog8_math.multiply_words.multiplier")
@@ -2494,7 +2970,7 @@ $shortcutLabel:""")
                     "+" -> asmgen.out("  lda  $name |  clc |  adc  $otherName |  sta  $name |  lda  $name+1 |  adc  $otherName+1 |  sta  $name+1")
                     "-" -> asmgen.out("  lda  $name |  sec |  sbc  $otherName |  sta  $name |  lda  $name+1 |  sbc  $otherName+1 |  sta  $name+1")
                     "*" -> {
-                        if(block?.options?.veraFxMuls==true)
+                        if(block?.options?.veraFxMuls==true) {
                             // cx16 verafx hardware muls
                             asmgen.out("""
                                 lda  $name
@@ -2505,9 +2981,10 @@ $shortcutLabel:""")
                                 ldy  $otherName+1
                                 sta  cx16.r1
                                 sty  cx16.r1+1
-                                jsr  verafx.muls
+                                jsr  verafx.muls16
                                 sta  $name
                                 sty  $name+1""")
+                        }
                         else
                             asmgen.out("""
                                 lda  $otherName
@@ -2698,7 +3175,7 @@ $shortcutLabel:""")
     private fun inplacemodificationWordWithValue(name: String, dt: DataType, operator: String, value: PtExpression, block: PtBlock?) {
         require(dt.isWord)
         fun multiplyVarByWordInAX() {
-            if(block?.options?.veraFxMuls==true)
+            if(block?.options?.veraFxMuls==true) {
                 // cx16 verafx hardware muls
                 asmgen.out("""
                     sta  cx16.r1
@@ -2707,9 +3184,10 @@ $shortcutLabel:""")
                     ldx  $name+1
                     sta  cx16.r0
                     stx  cx16.r0+1
-                    jsr  verafx.muls
+                    jsr  verafx.muls16
                     sta  $name
                     sty  $name+1""")
+            }
             else
                 asmgen.out("""
                     sta  prog8_math.multiply_words.multiplier
@@ -3003,7 +3481,7 @@ $shortcutLabel:""")
                         if(value is PtNumber && value.number<=255) {
                             TODO("shift a word var by ${value.number}")
                         } else {
-                            throw AssemblyError("shift by a word value not supported, max is a byte")
+                            throw AssemblyError("bit shift value can not be larger than a byte")
                         }
                     }
                     "&" -> {

codeGenCpu6502/test/Dummies.kt

@@ -12,12 +12,14 @@ internal object DummyMemsizer : IMemSizer {
             return when(dt.sub) {
                 BaseDataType.BOOL, BaseDataType.BYTE, BaseDataType.UBYTE -> numElements
                 BaseDataType.UWORD, BaseDataType.WORD -> numElements*2
+                BaseDataType.LONG -> numElements*4
                 BaseDataType.FLOAT -> numElements*5
                 else -> throw IllegalArgumentException("invalid sub type")
             }
         }
         return when {
             dt.isByteOrBool -> 1 * (numElements ?: 1)
+            dt.isLong -> 4 * (numElements ?: 1)
             dt.isFloat -> 5 * (numElements ?: 1)
             else -> 2 * (numElements ?: 1)
         }

codeGenCpu6502/test/TestCodegen.kt

@@ -31,6 +31,7 @@ class TestCodegen: FunSpec({
             noSysInit = false,
             romable = false,
             compTarget = target,
+            compilerVersion="99.99",
             loadAddress = target.PROGRAM_LOAD_ADDRESS,
             memtopAddress = 0xffffu
         )

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

@@ -24,7 +24,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             if(extsub!=null) {
                 require(funcCall.multipleResultRegs.size + funcCall.multipleResultFpRegs.size >= 2)
                 if (funcCall.multipleResultFpRegs.isNotEmpty())
-                    TODO("deal with (multiple?) FP return registers")
+                    TODO("deal with (multiple?) FP return registers ${assignment.position}")
                 if (extsub.returns.size == assignmentTargets.size) {
                     // Targets and values match. Assign all the things. Skip 'void' targets.
                     extsub.returns.zip(assignmentTargets).zip(funcCall.multipleResultRegs).forEach {
@@ -76,6 +76,11 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             RegisterOrPair.AY -> addInstr(result, IRInstruction(Opcode.LOADHAY, IRDataType.WORD, reg1=regNum), null)
             RegisterOrPair.XY -> addInstr(result, IRInstruction(Opcode.LOADHXY, IRDataType.WORD, reg1=regNum), null)
             in Cx16VirtualRegisters -> addInstr(result, IRInstruction(Opcode.LOADM, irType(returns.type), reg1=regNum, labelSymbol = "cx16.${returns.register.registerOrPair.toString().lowercase()}"), null)
+            in combinedLongRegisters -> {
+                require(returns.type.isLong)
+                val startreg = returns.register.registerOrPair!!.startregname()
+                addInstr(result, IRInstruction(Opcode.LOADM, IRDataType.LONG, reg1=regNum, labelSymbol = "cx16.${startreg}"), null)
+            }
             RegisterOrPair.FAC1 -> addInstr(result, IRInstruction(Opcode.LOADHFACZERO, IRDataType.FLOAT, fpReg1 = regNum), null)
             RegisterOrPair.FAC2 -> addInstr(result, IRInstruction(Opcode.LOADHFACONE, IRDataType.FLOAT, fpReg1 = regNum), null)
             null -> if(returns.register.statusflag!=null)
@@ -118,6 +123,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
     }
 
     internal fun translate(augAssign: PtAugmentedAssign): IRCodeChunks {
+        if(augAssign.target.type.isString)
+            throw AssemblyError("cannot assign to str type ${augAssign.position}")
+
         // augmented assignment always has just a single target
         if (augAssign.target.children.single() is PtIrRegister)
             throw AssemblyError("assigning to a register should be done by just evaluating the expression into resultregister")
@@ -137,63 +145,75 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             val inplaceInstrs = mutableListOf<IRCodeChunkBase>()
             val (addressReg, fieldOffset) = codeGen.evaluatePointerAddressIntoReg(inplaceInstrs, pointerDeref)
             val oldvalueReg = codeGen.registers.next(targetDt)
-            var operandTr = ExpressionCodeResult(emptyList(), IRDataType.BYTE, -1, -1)
-            if(augAssign.operator!="or=" && augAssign.operator!="and=") {
-                // for everything except the shortcircuit boolean operators, we can evaluate the value here unconditionally
-                operandTr = expressionEval.translateExpression(value)
-                inplaceInstrs += operandTr.chunks
-            }
-            if(targetDt== IRDataType.FLOAT) {
-                if(fieldOffset>0u)
-                    addInstr(inplaceInstrs, IRInstruction(Opcode.LOADFIELD, targetDt, fpReg1 = oldvalueReg, reg1 = addressReg, immediate = fieldOffset.toInt()), null)
-                else
-                    addInstr(inplaceInstrs, IRInstruction(Opcode.LOADI, targetDt, fpReg1 = oldvalueReg, reg1 = addressReg), null)
-                when(augAssign.operator) {
-                    "+=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ADDR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
-                    "-=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.SUBR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
-                    "*=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MULR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
-                    "/=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.DIVSR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
-                    "%=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MODR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
-                    "+" -> { /* inplace + is a no-op */ }
-                    else -> throw AssemblyError("invalid augmented assign operator for floats ${augAssign.operator}")
+
+            if((augAssign.operator=="+=" || augAssign.operator=="-=") && value.asConstInteger()==1 || value.asConstInteger()==2) {
+                // INC/DEC optimization instead of ADD/SUB
+
+                loadfield(inplaceInstrs, addressReg, fieldOffset, targetDt, oldvalueReg)
+                val instr = if(augAssign.operator=="+=") Opcode.INC else Opcode.DEC
+                repeat(value.asConstInteger()!!) {
+                    addInstr(inplaceInstrs, IRInstruction(instr, targetDt, reg1 = oldvalueReg), null)
                 }
+
             } else {
-                if(fieldOffset>0u)
-                    addInstr(inplaceInstrs, IRInstruction(Opcode.LOADFIELD, targetDt, reg1 = oldvalueReg, reg2 = addressReg, immediate = fieldOffset.toInt()), null)
-                else
-                    addInstr(inplaceInstrs, IRInstruction(Opcode.LOADI, targetDt, reg1 = oldvalueReg, reg2 = addressReg), null)
-                when(augAssign.operator) {
-                    "+=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ADDR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "-=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.SUBR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "*=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MULR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "/=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.DIVR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "%=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MODR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "|=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ORR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "&=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ANDR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "^=", "xor=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.XORR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "<<=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.LSLN, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    ">>=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.LSRN, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
-                    "or=" -> {
-                        val shortcutLabel = codeGen.createLabelName()
-                        addInstr(inplaceInstrs, IRInstruction(Opcode.BSTNE, labelSymbol = shortcutLabel), null)
-                        val valueTr = expressionEval.translateExpression(value)
-                        inplaceInstrs += valueTr.chunks
-                        addInstr(inplaceInstrs, IRInstruction(Opcode.ORR, targetDt, reg1=oldvalueReg, reg2=valueTr.resultReg), null)
-                        inplaceInstrs += IRCodeChunk(shortcutLabel, null)
+
+                var operandTr = ExpressionCodeResult(emptyList(), IRDataType.BYTE, -1, -1)
+                if(augAssign.operator!="or=" && augAssign.operator!="and=") {
+                    // for everything except the shortcircuit boolean operators, we can evaluate the value here unconditionally
+                    operandTr = expressionEval.translateExpression(value)
+                    // note: the instructions to load the value will be placed after the LOADFIELD instruction so that later optimizations about what modification is actually done, are easier
+                }
+                if(targetDt== IRDataType.FLOAT) {
+
+                    loadfield(inplaceInstrs, addressReg, fieldOffset, targetDt, oldvalueReg)
+                    inplaceInstrs += operandTr.chunks
+                    when(augAssign.operator) {
+                        "+=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ADDR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
+                        "-=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.SUBR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
+                        "*=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MULR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
+                        "/=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.DIVSR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
+                        "%=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MODR, targetDt, fpReg1 = oldvalueReg, fpReg2 = operandTr.resultFpReg), null)
+                        "+" -> { /* inplace + is a no-op */ }
+                        else -> throw AssemblyError("invalid augmented assign operator for floats ${augAssign.operator}")
                     }
-                    "and=" -> {
-                        val shortcutLabel = codeGen.createLabelName()
-                        addInstr(inplaceInstrs, IRInstruction(Opcode.BSTEQ, labelSymbol = shortcutLabel), null)
-                        val valueTr = expressionEval.translateExpression(value)
-                        inplaceInstrs += valueTr.chunks
-                        addInstr(inplaceInstrs, IRInstruction(Opcode.ANDR, targetDt, reg1=oldvalueReg, reg2=valueTr.resultReg), null)
-                        inplaceInstrs += IRCodeChunk(shortcutLabel, null)
+
+                } else {
+
+                    loadfield(inplaceInstrs, addressReg, fieldOffset, targetDt, oldvalueReg)
+                    inplaceInstrs += operandTr.chunks
+                    when(augAssign.operator) {
+                        "+=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ADDR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "-=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.SUBR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "*=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MULR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "/=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.DIVR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "%=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.MODR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "|=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ORR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "&=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.ANDR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "^=", "xor=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.XORR, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "<<=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.LSLN, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        ">>=" -> addInstr(inplaceInstrs, IRInstruction(Opcode.LSRN, targetDt, reg1 = oldvalueReg, reg2 = operandTr.resultReg), null)
+                        "or=" -> {
+                            val shortcutLabel = codeGen.createLabelName()
+                            addInstr(inplaceInstrs, IRInstruction(Opcode.BSTNE, labelSymbol = shortcutLabel), null)
+                            val valueTr = expressionEval.translateExpression(value)
+                            inplaceInstrs += valueTr.chunks
+                            addInstr(inplaceInstrs, IRInstruction(Opcode.ORR, targetDt, reg1=oldvalueReg, reg2=valueTr.resultReg), null)
+                            inplaceInstrs += IRCodeChunk(shortcutLabel, null)
+                        }
+                        "and=" -> {
+                            val shortcutLabel = codeGen.createLabelName()
+                            addInstr(inplaceInstrs, IRInstruction(Opcode.BSTEQ, labelSymbol = shortcutLabel), null)
+                            val valueTr = expressionEval.translateExpression(value)
+                            inplaceInstrs += valueTr.chunks
+                            addInstr(inplaceInstrs, IRInstruction(Opcode.ANDR, targetDt, reg1=oldvalueReg, reg2=valueTr.resultReg), null)
+                            inplaceInstrs += IRCodeChunk(shortcutLabel, null)
+                        }
+                        "-" -> addInstr(inplaceInstrs, IRInstruction(Opcode.NEG, targetDt, reg1 = oldvalueReg), null)
+                        "~" -> addInstr(inplaceInstrs, IRInstruction(Opcode.INV, targetDt, reg1 = oldvalueReg), null)
+                        "not" -> addInstr(inplaceInstrs, IRInstruction(Opcode.XOR, targetDt, reg1 = oldvalueReg, immediate = 1), null)
+                        "+" -> { /* inplace + is a no-op */ }
+                        else -> throw AssemblyError("invalid augmented assign operator ${augAssign.operator}")
                     }
-                    "-" -> addInstr(inplaceInstrs, IRInstruction(Opcode.NEG, targetDt, reg1 = oldvalueReg), null)
-                    "~" -> addInstr(inplaceInstrs, IRInstruction(Opcode.INV, targetDt, reg1 = oldvalueReg), null)
-                    "not" -> addInstr(inplaceInstrs, IRInstruction(Opcode.XOR, targetDt, reg1 = oldvalueReg, immediate = 1), null)
-                    "+" -> { /* inplace + is a no-op */ }
-                    else -> throw AssemblyError("invalid augmented assign operator ${augAssign.operator}")
                 }
             }
 
@@ -222,6 +242,42 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         return chunks
     }
 
+    private fun loadfield(
+        inplaceInstrs: MutableList<IRCodeChunkBase>,
+        addressReg: Int,
+        fieldOffset: UByte,
+        targetDt: IRDataType,
+        oldvalueReg: Int
+    ) {
+        if (targetDt == IRDataType.FLOAT) {
+            if (fieldOffset > 0u)
+                addInstr(
+                    inplaceInstrs,
+                    IRInstruction(Opcode.LOADFIELD, targetDt, fpReg1 = oldvalueReg, reg1 = addressReg, immediate = fieldOffset.toInt()),
+                    null
+                )
+            else
+                addInstr(
+                    inplaceInstrs,
+                    IRInstruction(Opcode.LOADI, targetDt, fpReg1 = oldvalueReg, reg1 = addressReg),
+                    null
+                )
+        } else {
+            if (fieldOffset > 0u)
+                addInstr(
+                    inplaceInstrs,
+                    IRInstruction(Opcode.LOADFIELD, targetDt, reg1 = oldvalueReg, reg2 = addressReg, immediate = fieldOffset.toInt()),
+                    null
+                )
+            else
+                addInstr(
+                    inplaceInstrs,
+                    IRInstruction(Opcode.LOADI, targetDt, reg1 = oldvalueReg, reg2 = addressReg),
+                    null
+                )
+        }
+    }
+
     private fun fallbackAssign(origAssign: PtAugmentedAssign): IRCodeChunks {
         val value: PtExpression
         if(origAssign.operator in PrefixOperators) {
@@ -414,11 +470,11 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
     }
 
     private fun translateRegularAssign(assignment: PtAssignment): IRCodeChunks {
-        // note: assigning array and string values is done via an explicit memcopy/stringcopy function call.
-        val targetIdent = assignment.target.identifier
-        val targetMemory = assignment.target.memory
-        val targetArray = assignment.target.array
-        val targetPointerDeref = assignment.target.pointerDeref
+        if(assignment.target.type.isString) {
+            // assigning array and string values is done via an explicit memcopy/stringcopy function calls.
+            throw AssemblyError("cannot assign to str type ${assignment.position}")
+        }
+
         val valueDt = irType(assignment.value.type)
         val targetDt = irType(assignment.target.type)
         val result = mutableListOf<IRCodeChunkBase>()
@@ -459,98 +515,104 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             }
         }
 
-        if(targetIdent!=null) {
-            val instruction = if(zero) {
-                IRInstruction(Opcode.STOREZM, targetDt, labelSymbol = targetIdent.name)
-            } else {
-                if (targetDt == IRDataType.FLOAT) {
-                    require(valueFpRegister>=0)
-                    IRInstruction(Opcode.STOREM, targetDt, fpReg1 = valueFpRegister, labelSymbol = targetIdent.name)
-                }
-                else {
-                    require(valueRegister>=0)
-                    IRInstruction(Opcode.STOREM, targetDt, reg1 = valueRegister, labelSymbol = targetIdent.name)
-                }
-            }
-            result += IRCodeChunk(null, null).also { it += instruction }
-            return result
-        }
-        else if(targetArray!=null) {
-            val eltSize = codeGen.program.memsizer.memorySize(targetArray.type, null)
-            val variable = targetArray.variable
-            if(variable==null)
-                translateRegularAssignPointerIndexed(result, targetArray.pointerderef!!, eltSize, targetArray, zero, targetDt, valueRegister, valueFpRegister)
-            else if(variable.type.isPointer)
-                assignToIndexedSimplePointer(result, variable, eltSize, targetArray, zero, targetDt, valueRegister, valueFpRegister)
-            else
-                translateRegularAssignArrayIndexed(result, variable.name, eltSize, targetArray, zero, targetDt, valueRegister, valueFpRegister)
-            return result
-        }
-        else if(targetMemory!=null) {
-            require(targetDt == IRDataType.BYTE) { "must be byte type ${targetMemory.position}"}
-            if(zero) {
-                if(targetMemory.address is PtNumber) {
-                    val chunk = IRCodeChunk(null, null).also { it += IRInstruction(Opcode.STOREZM, targetDt, address = (targetMemory.address as PtNumber).number.toInt()) }
-                    result += chunk
-                } else {
-                    val tr = expressionEval.translateExpression(targetMemory.address)
-                    val addressReg = tr.resultReg
-                    addToResult(result, tr, tr.resultReg, -1)
-                    result += IRCodeChunk(null, null).also {
-                        it += IRInstruction(Opcode.STOREZI, targetDt, reg1=addressReg)
-                    }
-                }
-            } else {
-                val constAddress = targetMemory.address as? PtNumber
-                if(constAddress!=null) {
-                    addInstr(result, IRInstruction(Opcode.STOREM, targetDt, reg1=valueRegister, address=constAddress.number.toInt()), null)
-                    return result
-                }
-                val ptrWithOffset = targetMemory.address as? PtBinaryExpression
-                if(ptrWithOffset!=null) {
-                    if(ptrWithOffset.operator=="+" && ptrWithOffset.left is PtIdentifier) {
-                        val constOffset = (ptrWithOffset.right as? PtNumber)?.number?.toInt()
-                        if(constOffset in 0..255) {
-                            val ptrName = (ptrWithOffset.left as PtIdentifier).name
-                            val pointerReg = codeGen.registers.next(IRDataType.WORD)
-                            result += IRCodeChunk(null, null).also {
-                                it += IRInstruction(Opcode.LOADM, IRDataType.WORD, reg1=pointerReg, labelSymbol = ptrName)
-                                it += IRInstruction(Opcode.STOREFIELD, IRDataType.BYTE, reg1=valueRegister, reg2=pointerReg, immediate = constOffset)
-                            }
-                            return result
+        with(assignment.target) {
+            when {
+                identifier != null -> {
+                    val instruction = if(zero) {
+                        IRInstruction(Opcode.STOREZM, targetDt, labelSymbol = identifier!!.name)
+                    } else {
+                        if (targetDt == IRDataType.FLOAT) {
+                            require(valueFpRegister>=0)
+                            IRInstruction(Opcode.STOREM, targetDt, fpReg1 = valueFpRegister, labelSymbol = identifier!!.name)
+                        }
+                        else {
+                            require(valueRegister>=0)
+                            IRInstruction(Opcode.STOREM, targetDt, reg1 = valueRegister, labelSymbol = identifier!!.name)
                         }
                     }
-                    val offsetTypecast = ptrWithOffset.right as? PtTypeCast
-                    if(ptrWithOffset.operator=="+" && ptrWithOffset.left is PtIdentifier && (ptrWithOffset.right.type.isByte || offsetTypecast?.value?.type?.isByte==true)) {
-                        // STOREIX only works with byte index.
-                        val tr = if(offsetTypecast?.value?.type?.isByte==true)
-                            expressionEval.translateExpression(offsetTypecast.value)
-                        else
-                            expressionEval.translateExpression(ptrWithOffset.right)
+                    result += IRCodeChunk(null, null).also { it += instruction }
+                    return result
+                }
+                memory != null -> {
+                    require(targetDt == IRDataType.BYTE) { "must be byte type ${memory!!.position}"}
+                    if(zero) {
+                        if(memory!!.address is PtNumber) {
+                            val chunk = IRCodeChunk(null, null).also { it += IRInstruction(Opcode.STOREZM, targetDt, address = (memory!!.address as PtNumber).number.toInt()) }
+                            result += chunk
+                        } else {
+                            val tr = expressionEval.translateExpression(memory!!.address)
+                            val addressReg = tr.resultReg
+                            addToResult(result, tr, tr.resultReg, -1)
+                            result += IRCodeChunk(null, null).also {
+                                it += IRInstruction(Opcode.STOREZI, targetDt, reg1=addressReg)
+                            }
+                        }
+                    } else {
+                        val constAddress = memory!!.address as? PtNumber
+                        if(constAddress!=null) {
+                            addInstr(result, IRInstruction(Opcode.STOREM, targetDt, reg1=valueRegister, address=constAddress.number.toInt()), null)
+                            return result
+                        }
+                        val ptrWithOffset = memory!!.address as? PtBinaryExpression
+                        if(ptrWithOffset!=null) {
+                            if(ptrWithOffset.operator=="+" && ptrWithOffset.left is PtIdentifier) {
+                                val constOffset = (ptrWithOffset.right as? PtNumber)?.number?.toInt()
+                                if(constOffset in 0..255) {
+                                    val ptrName = (ptrWithOffset.left as PtIdentifier).name
+                                    val pointerReg = codeGen.registers.next(IRDataType.WORD)
+                                    result += IRCodeChunk(null, null).also {
+                                        it += IRInstruction(Opcode.LOADM, IRDataType.WORD, reg1=pointerReg, labelSymbol = ptrName)
+                                        it += IRInstruction(Opcode.STOREFIELD, IRDataType.BYTE, reg1=valueRegister, reg2=pointerReg, immediate = constOffset)
+                                    }
+                                    return result
+                                }
+                            }
+                            val offsetTypecast = ptrWithOffset.right as? PtTypeCast
+                            if(ptrWithOffset.operator=="+" && ptrWithOffset.left is PtIdentifier && (ptrWithOffset.right.type.isByte || offsetTypecast?.value?.type?.isByte==true)) {
+                                // STOREIX only works with byte index.
+                                val tr = if(offsetTypecast?.value?.type?.isByte==true)
+                                    expressionEval.translateExpression(offsetTypecast.value)
+                                else
+                                    expressionEval.translateExpression(ptrWithOffset.right)
+                                addToResult(result, tr, tr.resultReg, -1)
+                                val ptrName = (ptrWithOffset.left as PtIdentifier).name
+                                addInstr(result, IRInstruction(Opcode.STOREIX, IRDataType.BYTE, reg1=valueRegister, reg2=tr.resultReg, labelSymbol = ptrName), null)
+                                return result
+                            }
+                        }
+
+                        val tr = expressionEval.translateExpression(memory!!.address)
+                        val addressReg = tr.resultReg
                         addToResult(result, tr, tr.resultReg, -1)
-                        val ptrName = (ptrWithOffset.left as PtIdentifier).name
-                        addInstr(result, IRInstruction(Opcode.STOREIX, IRDataType.BYTE, reg1=valueRegister, reg2=tr.resultReg, labelSymbol = ptrName), null)
+                        addInstr(result, IRInstruction(Opcode.STOREI, targetDt, reg1=valueRegister, reg2=addressReg), null)
                         return result
                     }
+
+                    return result
                 }
+                array != null -> {
+                    val eltSize = codeGen.program.memsizer.memorySize(array!!.type, null)
+                    val variable = array!!.variable
+                    if(variable==null)
+                        translateRegularAssignPointerIndexed(result, array!!.pointerderef!!, eltSize, array!!, zero, targetDt, valueRegister, valueFpRegister)
+                    else if(variable.type.isPointer)
+                        assignToIndexedSimplePointer(result, variable, eltSize, array!!, zero, targetDt, valueRegister, valueFpRegister)
+                    else
+                        translateRegularAssignArrayIndexed(result, variable.name, eltSize, array!!, zero, targetDt, valueRegister, valueFpRegister)
+                    return result
+                }
+                pointerDeref != null -> {
+                    val (addressReg, offset) = codeGen.evaluatePointerAddressIntoReg(result, pointerDeref!!)
+                    val actualValueReg = if(pointerDeref!!.type.isFloat) valueFpRegister else valueRegister
+                    codeGen.storeValueAtPointersLocation(result, addressReg, offset, pointerDeref!!.type, zero, actualValueReg)
+                    return result
 
-                val tr = expressionEval.translateExpression(targetMemory.address)
-                val addressReg = tr.resultReg
-                addToResult(result, tr, tr.resultReg, -1)
-                addInstr(result, IRInstruction(Opcode.STOREI, targetDt, reg1=valueRegister, reg2=addressReg), null)
-                return result
+                }
+                else -> {
+                    throw AssemblyError("weird assigntarget")
+                }
             }
-
-            return result
         }
-        else if(targetPointerDeref!=null) {
-            val (addressReg, offset) = codeGen.evaluatePointerAddressIntoReg(result, targetPointerDeref)
-            val actualValueReg = if(targetPointerDeref.type.isFloat) valueFpRegister else valueRegister
-            codeGen.storeValueAtPointersLocation(result, addressReg, offset, targetPointerDeref.type, zero, actualValueReg)
-            return result
-        }
-        else
-            throw AssemblyError("weird assigntarget")
     }
 
     private fun assignToIndexedSimplePointer(
@@ -647,9 +709,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                     val chunk = IRCodeChunk(null, null).also {
                         if(targetArray.splitWords) {
                             val lsbmsbReg = codeGen.registers.next(IRDataType.BYTE)
-                            it += IRInstruction(Opcode.LSIG, IRDataType.BYTE, reg1 = lsbmsbReg, reg2 = valueRegister)
+                            it += IRInstruction(Opcode.LSIGB, IRDataType.WORD, reg1 = lsbmsbReg, reg2 = valueRegister)
                             it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = lsbmsbReg, immediate = arrayLength, labelSymbol = "${variable}_lsb", symbolOffset = fixedIndex)
-                            it += IRInstruction(Opcode.MSIG, IRDataType.BYTE, reg1 = lsbmsbReg, reg2 = valueRegister)
+                            it += IRInstruction(Opcode.MSIGB, IRDataType.WORD, reg1 = lsbmsbReg, reg2 = valueRegister)
                             it += IRInstruction(Opcode.STOREM, IRDataType.BYTE, reg1 = lsbmsbReg, immediate = arrayLength, labelSymbol = "${variable}_msb", symbolOffset = fixedIndex)
                         }
                         else
@@ -662,9 +724,9 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                     result += IRCodeChunk(null, null).also {
                         if(targetArray.splitWords) {
                             val lsbmsbReg = codeGen.registers.next(IRDataType.BYTE)
-                            it += IRInstruction(Opcode.LSIG, IRDataType.BYTE, reg1 = lsbmsbReg, reg2 = valueRegister)
+                            it += IRInstruction(Opcode.LSIGB, IRDataType.WORD, reg1 = lsbmsbReg, reg2 = valueRegister)
                             it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1 = lsbmsbReg, reg2=indexReg, immediate = arrayLength, labelSymbol = "${variable}_lsb")
-                            it += IRInstruction(Opcode.MSIG, IRDataType.BYTE, reg1 = lsbmsbReg, reg2 = valueRegister)
+                            it += IRInstruction(Opcode.MSIGB, IRDataType.WORD, reg1 = lsbmsbReg, reg2 = valueRegister)
                             it += IRInstruction(Opcode.STOREX, IRDataType.BYTE, reg1 = lsbmsbReg, reg2=indexReg, immediate = arrayLength, labelSymbol = "${variable}_msb")
                         }
                         else
@@ -914,7 +976,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
 
     private fun operatorDivideInplace(symbol: String?, array: PtArrayIndexer?, constAddress: Int?, memory: PtMemoryByte?, vmDt: IRDataType, operand: PtExpression, signed: Boolean): IRCodeChunks? {
         if(array!=null) {
-            TODO("/ in array")
+            TODO("/ in array ${array.position}")
         }
         if(constAddress==null && memory!=null)
             return null  // TODO("optimized memory in-place /"")
@@ -1054,14 +1116,14 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         if(constAddress==null && memory!=null)
             return null  // TODO("optimized memory in-place -"")
 
+        val constValue = (operand as? PtNumber)?.number
         val result = mutableListOf<IRCodeChunkBase>()
         if(vmDt==IRDataType.FLOAT) {
-            if((operand as? PtNumber)?.number==1.0) {
+            if(constValue==1.0) {
                 addInstr(result, if(constAddress!=null)
                     IRInstruction(Opcode.DECM, vmDt, address = constAddress)
                 else
-                    IRInstruction(Opcode.DECM, vmDt, labelSymbol = symbol)
-                    , null)
+                    IRInstruction(Opcode.DECM, vmDt, labelSymbol = symbol), null)
             }
             else {
                 val tr = expressionEval.translateExpression(operand)
@@ -1069,16 +1131,14 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 addInstr(result, if(constAddress!=null)
                     IRInstruction(Opcode.SUBM, vmDt, fpReg1=tr.resultFpReg, address = constAddress)
                 else
-                    IRInstruction(Opcode.SUBM, vmDt, fpReg1=tr.resultFpReg, labelSymbol = symbol)
-                    , null)
+                    IRInstruction(Opcode.SUBM, vmDt, fpReg1=tr.resultFpReg, labelSymbol = symbol), null)
             }
         } else {
-            if((operand as? PtNumber)?.number==1.0) {
+            if(constValue==1.0) {
                 addInstr(result, if(constAddress!=null)
                     IRInstruction(Opcode.DECM, vmDt, address = constAddress)
                 else
-                    IRInstruction(Opcode.DECM, vmDt, labelSymbol = symbol)
-                    , null)
+                    IRInstruction(Opcode.DECM, vmDt, labelSymbol = symbol), null)
             }
             else {
                 val tr = expressionEval.translateExpression(operand)
@@ -1086,8 +1146,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                 addInstr(result, if(constAddress!=null)
                     IRInstruction(Opcode.SUBM, vmDt, reg1=tr.resultReg, address = constAddress)
                 else
-                    IRInstruction(Opcode.SUBM, vmDt, reg1=tr.resultReg, labelSymbol = symbol)
-                    , null)
+                    IRInstruction(Opcode.SUBM, vmDt, reg1=tr.resultReg, labelSymbol = symbol), null)
             }
         }
         return result
@@ -1153,9 +1212,10 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
             return null  // TODO("optimized memory in-place +"")
 
         val result = mutableListOf<IRCodeChunkBase>()
+        val constValue = (operand as? PtNumber)?.number
 
         if(vmDt==IRDataType.FLOAT) {
-            if((operand as? PtNumber)?.number==1.0) {
+            if(constValue==1.0) {
                 addInstr(result, if (constAddress != null)
                     IRInstruction(Opcode.INCM, vmDt, address = constAddress)
                 else
@@ -1170,7 +1230,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
                     IRInstruction(Opcode.ADDM, vmDt, fpReg1 = tr.resultFpReg, labelSymbol = symbol) , null)
             }
         } else {
-            if((operand as? PtNumber)?.number==1.0) {
+            if(constValue==1.0) {
                 addInstr(result, if (constAddress != null)
                     IRInstruction(Opcode.INCM, vmDt, address = constAddress)
                 else
@@ -1259,11 +1319,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         } else {
             val shiftTr = expressionEval.translateExpression(operand)
             addToResult(result, shiftTr, shiftTr.resultReg, -1)
-            var shiftReg = shiftTr.resultReg
-            if(vmDt==IRDataType.WORD && shiftTr.dt==IRDataType.BYTE) {
-                shiftReg = codeGen.registers.next(IRDataType.WORD)
-                addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=shiftReg, reg2=shiftTr.resultReg), null)
-            }
+            val shiftReg = shiftTr.resultReg
             val opc = if (signed) Opcode.ASRNM else Opcode.LSRNM
             val ins = if(constAddress!=null)
                 IRInstruction(opc, vmDt, reg1 = shiftReg, address = constAddress)
@@ -1321,11 +1377,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
         } else {
             val shiftTr = expressionEval.translateExpression(operand)
             addToResult(result, shiftTr, shiftTr.resultReg, -1)
-            var shiftReg = shiftTr.resultReg
-            if(vmDt==IRDataType.WORD && shiftTr.dt==IRDataType.BYTE) {
-                shiftReg = codeGen.registers.next(IRDataType.WORD)
-                addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=shiftReg, reg2=shiftTr.resultReg), null)
-            }
+            val shiftReg = shiftTr.resultReg
             addInstr(result, if(constAddress!=null)
                 IRInstruction(Opcode.LSLNM, vmDt, reg1=shiftReg, address = constAddress)
             else
@@ -1380,7 +1432,7 @@ internal class AssignmentGen(private val codeGen: IRCodeGen, private val express
 
     private fun operatorModuloInplace(symbol: String?, array: PtArrayIndexer?, constAddress: Int?, memory: PtMemoryByte?, vmDt: IRDataType, operand: PtExpression): IRCodeChunks? {
         if(array!=null) {
-            TODO("% in array")
+            TODO("% in array  ${array.position}")
         }
         if(constAddress==null && memory!=null)
             return null  // TODO("optimized memory in-place %"")

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

@@ -1,5 +1,7 @@
 package prog8.codegen.intermediate
 
+import prog8.code.StMemorySlabBlockName
+import prog8.code.StStructInstanceBlockName
 import prog8.code.SymbolTable
 import prog8.code.ast.*
 import prog8.code.core.AssemblyError
@@ -12,35 +14,40 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
 
     fun translate(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         return when(call.name) {
-            "abs__byte", "abs__word", "abs__float" -> funcAbs(call)
+            "abs__byte", "abs__word", "abs__long", "abs__float" -> funcAbs(call)
             "cmp" -> funcCmp(call)
             "sgn" -> funcSgn(call)
-            "sqrt__ubyte", "sqrt__uword", "sqrt__float" -> funcSqrt(call)
+            "sqrt__ubyte", "sqrt__uword", "sqrt__long", "sqrt__float" -> funcSqrt(call)
             "divmod__ubyte" -> funcDivmod(call, IRDataType.BYTE)
             "divmod__uword" -> funcDivmod(call, IRDataType.WORD)
             "rsave", "rrestore" -> ExpressionCodeResult.EMPTY  // vm doesn't have registers to save/restore
             "callfar" -> funcCallfar(call)
             "callfar2" -> funcCallfar2(call)
             "call" -> funcCall(call)
-            "lsw" -> throw AssemblyError("lsw() should have been removed or replaced by a const value")
-            "msw" -> throw AssemblyError("msw() should have been removed or replaced by a const value")
-            "msb" -> funcMsb(call)
-            "lsb" -> funcLsb(call)
+            "msw" -> funcMsw(call)
+            "lsw" -> funcLsw(call)
+            "msb" -> funcMsb(call, false)
+            "msb__long" -> funcMsb(call, true)
+            "lsb" -> funcLsb(call, false)
+            "lsb__long" -> funcLsb(call, true)
             "memory" -> funcMemory(call)
             "peek" -> funcPeek(call, IRDataType.BYTE)
             "peekbool" -> funcPeek(call, IRDataType.BYTE)
             "peekw" -> funcPeek(call, IRDataType.WORD)
+            "peekl" -> funcPeek(call, IRDataType.LONG)
             "peekf" -> funcPeek(call, IRDataType.FLOAT)
             "poke" -> funcPoke(call, IRDataType.BYTE)
             "pokebool" -> funcPoke(call, IRDataType.BYTE)
             "pokebowl" -> funcPoke(call, IRDataType.BYTE)
             "pokew" -> funcPoke(call, IRDataType.WORD)
+            "pokel" -> funcPoke(call, IRDataType.LONG)
             "pokef" -> funcPoke(call, IRDataType.FLOAT)
             "pokemon" -> funcPokemon(call)
             "mkword" -> funcMkword(call)
-            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword" -> funcClamp(call)
-            "min__byte", "min__ubyte", "min__word", "min__uword" -> funcMin(call)
-            "max__byte", "max__ubyte", "max__word", "max__uword" -> funcMax(call)
+            "mklong", "mklong2" -> funcMklong(call)
+            "clamp__byte", "clamp__ubyte", "clamp__word", "clamp__uword", "clamp__long" -> funcClamp(call)
+            "min__byte", "min__ubyte", "min__word", "min__uword", "min__long" -> funcMin(call)
+            "max__byte", "max__ubyte", "max__word", "max__uword", "max__long" -> funcMax(call)
             "setlsb" -> funcSetLsbMsb(call, false)
             "setmsb" -> funcSetLsbMsb(call, true)
             "rol" -> funcRolRor(call)
@@ -52,6 +59,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
             "prog8_lib_square_word" -> funcSquare(call, IRDataType.WORD)
             "prog8_lib_structalloc" -> funcStructAlloc(call)
             "sizeof" -> throw AssemblyError("sizeof must have been replaced with a constant")
+            "offsetof" -> throw AssemblyError("offsetof must have been replaced with a constant")
             else -> throw AssemblyError("missing builtinfunc for ${call.name}")
         }
     }
@@ -201,6 +209,17 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 result += IRCodeChunk(notNegativeLabel, null)
                 return ExpressionCodeResult(result, IRDataType.WORD, tr.resultReg, -1)
             }
+            BaseDataType.LONG -> {
+                val notNegativeLabel = codeGen.createLabelName()
+                val compareReg = codeGen.registers.next(IRDataType.LONG)
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.LOADR, IRDataType.LONG, reg1=compareReg, reg2=tr.resultReg)
+                    it += IRInstruction(Opcode.BSTPOS, labelSymbol = notNegativeLabel)
+                    it += IRInstruction(Opcode.NEG, IRDataType.LONG, reg1=tr.resultReg)
+                }
+                result += IRCodeChunk(notNegativeLabel, null)
+                return ExpressionCodeResult(result, IRDataType.LONG, tr.resultReg, -1)
+            }
             BaseDataType.FLOAT -> {
                 val resultFpReg = codeGen.registers.next(IRDataType.FLOAT)
                 addInstr(result, IRInstruction(Opcode.FABS, IRDataType.FLOAT, fpReg1 = resultFpReg, fpReg2 = tr.resultFpReg), null)
@@ -213,7 +232,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
     private fun funcSgn(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val tr = exprGen.translateExpression(call.args.single())
-        val resultReg = codeGen.registers.next(tr.dt)
+        val resultReg = codeGen.registers.next(IRDataType.BYTE)
 
         if(tr.dt==IRDataType.FLOAT) {
             addToResult(result, tr, -1, tr.resultFpReg)
@@ -250,6 +269,14 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 }
                 return ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
             }
+            BaseDataType.LONG -> {
+                addToResult(result, tr, tr.resultReg, -1)
+                val resultReg = codeGen.registers.next(IRDataType.WORD)     // sqrt of a long still produces just a word result
+                result += IRCodeChunk(null, null).also {
+                    it += IRInstruction(Opcode.SQRT, IRDataType.LONG, reg1=resultReg, reg2=tr.resultReg)
+                }
+                return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
+            }
             BaseDataType.FLOAT -> {
                 addToResult(result, tr, -1, tr.resultFpReg)
                 val resultFpReg = codeGen.registers.next(IRDataType.FLOAT)
@@ -280,6 +307,52 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
     }
 
+    private fun funcMklong(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val resultReg = codeGen.registers.next(IRDataType.LONG)
+        if(call.args.size==2) {
+            // mklong2(word, word)
+            if((call.args[0] as? PtNumber)?.number == 0.0) {
+                // msw is 0, use EXT
+                val lswTr = exprGen.translateExpression(call.args[1])
+                addToResult(result, lswTr, lswTr.resultReg, -1)
+                addInstr(result, IRInstruction(Opcode.EXT, IRDataType.WORD, reg1=resultReg, reg2 = lswTr.resultReg), null)
+            } else {
+                val mswTr = exprGen.translateExpression(call.args[0])
+                addToResult(result, mswTr, mswTr.resultReg, -1)
+                val lswTr = exprGen.translateExpression(call.args[1])
+                addToResult(result, lswTr, lswTr.resultReg, -1)
+                addInstr(result, IRInstruction(Opcode.CONCAT, IRDataType.WORD, reg1=resultReg, reg2 = mswTr.resultReg, reg3 = lswTr.resultReg), null)
+            }
+        } else {
+            // mklong(msb, b3, b2, lsb)
+            if((call.args[0] as? PtNumber)?.number == 0.0 && (call.args[1] as? PtNumber)?.number == 0.0 && (call.args[2] as? PtNumber)?.number == 0.0) {
+                // use EXT.b + EXT.w
+                val lsbTr = exprGen.translateExpression(call.args[3])
+                addToResult(result, lsbTr, lsbTr.resultReg, -1)
+                val wordReg = codeGen.registers.next(IRDataType.WORD)
+                addInstr(result, IRInstruction(Opcode.EXT, IRDataType.BYTE, reg1=wordReg, reg2 = lsbTr.resultReg), null)
+                addInstr(result, IRInstruction(Opcode.EXT, IRDataType.WORD, reg1=resultReg, reg2 = wordReg), null)
+            } else {
+                val msbTr = exprGen.translateExpression(call.args[0])
+                val b2Tr = exprGen.translateExpression(call.args[1])
+                val b1Tr = exprGen.translateExpression(call.args[2])
+                val lsbTr = exprGen.translateExpression(call.args[3])
+                addToResult(result, msbTr, msbTr.resultReg, -1)
+                addToResult(result, b2Tr, b2Tr.resultReg, -1)
+                addToResult(result, b1Tr, b1Tr.resultReg, -1)
+                addToResult(result, lsbTr, lsbTr.resultReg, -1)
+                val lswReg = codeGen.registers.next(IRDataType.WORD)
+                val mswReg = codeGen.registers.next(IRDataType.WORD)
+                addInstr(result, IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=mswReg, reg2 = msbTr.resultReg, reg3 = b2Tr.resultReg), null)
+                addInstr(result, IRInstruction(Opcode.CONCAT, IRDataType.BYTE, reg1=lswReg, reg2 = b1Tr.resultReg, reg3 = lsbTr.resultReg), null)
+                addInstr(result, IRInstruction(Opcode.CONCAT, IRDataType.WORD, reg1=resultReg, reg2 = mswReg, reg3 = lswReg), null)
+            }
+
+        }
+        return ExpressionCodeResult(result, IRDataType.LONG, resultReg, -1)
+    }
+
     private fun funcClamp(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val type = irType(call.type)
@@ -304,6 +377,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
                 BaseDataType.BYTE -> IMSyscall.CLAMP_BYTE
                 BaseDataType.UWORD -> IMSyscall.CLAMP_UWORD
                 BaseDataType.WORD -> IMSyscall.CLAMP_WORD
+                BaseDataType.LONG -> IMSyscall.CLAMP_LONG
                 else -> throw AssemblyError("invalid dt")
             }
             result += codeGen.makeSyscall(syscall, listOf(
@@ -500,7 +574,7 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val name = (call.args[0] as PtString).value
         val code = IRCodeChunk(null, null)
         val resultReg = codeGen.registers.next(IRDataType.WORD)
-        code += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=resultReg, labelSymbol = "$StMemorySlabPrefix.prog8_memoryslab_$name")
+        code += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=resultReg, labelSymbol = "$StMemorySlabBlockName.memory_$name")
         return ExpressionCodeResult(code, IRDataType.WORD, resultReg, -1)
     }
 
@@ -508,30 +582,54 @@ internal class BuiltinFuncGen(private val codeGen: IRCodeGen, private val exprGe
         val code = IRCodeChunk(null, null)
         val resultReg = codeGen.registers.next(IRDataType.WORD)
         val labelname = SymbolTable.labelnameForStructInstance(call)
-        code += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=resultReg, labelSymbol = labelname)
+        code += IRInstruction(Opcode.LOAD, IRDataType.WORD, reg1=resultReg, labelSymbol = "${StStructInstanceBlockName}.$labelname")
         return ExpressionCodeResult(code, IRDataType.WORD, resultReg, -1)
     }
 
-    private fun funcLsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+    private fun funcLsb(call: PtBuiltinFunctionCall, fromLong: Boolean): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val tr = exprGen.translateExpression(call.args.single())
         addToResult(result, tr, tr.resultReg, -1)
         val resultReg = codeGen.registers.next(IRDataType.BYTE)
-        addInstr(result, IRInstruction(Opcode.LSIG, IRDataType.BYTE, reg1 = resultReg, reg2 = tr.resultReg), null)
+        if(fromLong)
+            addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.LONG, reg1 = resultReg, reg2 = tr.resultReg), null)
+        else
+            addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.WORD, reg1 = resultReg, reg2 = tr.resultReg), null)
         // note: if a word result is needed, the upper byte is cleared by the typecast that follows. No need to do it here.
         return ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
     }
 
-    private fun funcMsb(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+    private fun funcLsw(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val tr = exprGen.translateExpression(call.args.single())
+        addToResult(result, tr, tr.resultReg, -1)
+        val resultReg = codeGen.registers.next(IRDataType.WORD)
+        addInstr(result, IRInstruction(Opcode.LSIGW, IRDataType.LONG, reg1 = resultReg, reg2 = tr.resultReg), null)
+        return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
+    }
+
+    private fun funcMsb(call: PtBuiltinFunctionCall, fromLong: Boolean): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val tr = exprGen.translateExpression(call.args.single())
         addToResult(result, tr, tr.resultReg, -1)
         val resultReg = codeGen.registers.next(IRDataType.BYTE)
-        addInstr(result, IRInstruction(Opcode.MSIG, IRDataType.BYTE, reg1 = resultReg, reg2 = tr.resultReg), null)
+        if(fromLong)
+            addInstr(result, IRInstruction(Opcode.MSIGB, IRDataType.LONG, reg1 = resultReg, reg2 = tr.resultReg), null)
+        else
+            addInstr(result, IRInstruction(Opcode.MSIGB, IRDataType.WORD, reg1 = resultReg, reg2 = tr.resultReg), null)
         // note: if a word result is needed, the upper byte is cleared by the typecast that follows. No need to do it here.
         return ExpressionCodeResult(result, IRDataType.BYTE, resultReg, -1)
     }
 
+    private fun funcMsw(call: PtBuiltinFunctionCall): ExpressionCodeResult {
+        val result = mutableListOf<IRCodeChunkBase>()
+        val tr = exprGen.translateExpression(call.args.single())
+        addToResult(result, tr, tr.resultReg, -1)
+        val resultReg = codeGen.registers.next(IRDataType.WORD)
+        addInstr(result, IRInstruction(Opcode.MSIGW, IRDataType.LONG, reg1 = resultReg, reg2 = tr.resultReg), null)
+        return ExpressionCodeResult(result, IRDataType.WORD, resultReg, -1)
+    }
+
     private fun funcRolRor(call: PtBuiltinFunctionCall): ExpressionCodeResult {
         val result = mutableListOf<IRCodeChunkBase>()
         val arg = call.args[0]

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

@@ -99,7 +99,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
         var pointerReg: Int
 
         if(deref.startpointer.type.isStructInstance) {
-            TODO("translate structinstance deref???")
+            TODO("translate structinstance deref??? ${deref.position}")
 /*
             val arrayIndexer = deref.startpointer as? PtArrayIndexer
             if(arrayIndexer==null)
@@ -446,7 +446,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 else (it as PtNumber).number.toInt()
             }
             when {
-                elementDt.isIntegerOrBool -> {
+                elementDt.isWordOrByteOrBool -> {
                     if (elementDt.isByteOrBool) require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_BYTE)
                     if (elementDt.isWord) require(haystack.size in 0..PtContainmentCheck.MAX_SIZE_FOR_INLINE_CHECKS_WORD)
                     val gottemLabel = codeGen.createLabelName()
@@ -686,6 +686,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                         addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.WORD, reg1=tr.resultReg, immediate = 0), null)
                         actualResultReg2 = loadStatusAsBooleanResult(Opcode.BSTNE, result)
                     }
+                    valueDt.isLong -> {
+                        addInstr(result, IRInstruction(Opcode.CMPI, IRDataType.LONG, reg1=tr.resultReg, immediate = 0), null)
+                        actualResultReg2 = loadStatusAsBooleanResult(Opcode.BSTNE, result)
+                    }
                     valueDt.isFloat -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
                         result += IRCodeChunk(null, null).also {
@@ -703,7 +707,11 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     }
                     BaseDataType.UWORD, BaseDataType.WORD -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
-                        addInstr(result, IRInstruction(Opcode.LSIG, IRDataType.BYTE, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                        addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.WORD, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                    }
+                    BaseDataType.LONG -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
+                        addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.LONG, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
                     }
                     BaseDataType.FLOAT -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
@@ -719,7 +727,11 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     }
                     BaseDataType.UWORD, BaseDataType.WORD -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
-                        addInstr(result, IRInstruction(Opcode.LSIG, IRDataType.BYTE, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                        addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.WORD, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                    }
+                    BaseDataType.LONG -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
+                        addInstr(result, IRInstruction(Opcode.LSIGB, IRDataType.LONG, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
                     }
                     BaseDataType.FLOAT -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.BYTE)
@@ -743,6 +755,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     BaseDataType.WORD -> {
                         actualResultReg2 = tr.resultReg
                     }
+                    BaseDataType.LONG -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.WORD)
+                        addInstr(result, IRInstruction(Opcode.LSIGW, IRDataType.LONG, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                    }
                     BaseDataType.FLOAT -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.WORD)
                         addInstr(result, IRInstruction(Opcode.FTOUW, IRDataType.FLOAT, reg1=actualResultReg2, fpReg1 = tr.resultFpReg), null)
@@ -768,6 +784,10 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     BaseDataType.UWORD -> {
                         actualResultReg2 = tr.resultReg
                     }
+                    BaseDataType.LONG -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.WORD)
+                        addInstr(result, IRInstruction(Opcode.LSIGW, IRDataType.LONG, reg1=actualResultReg2, reg2=tr.resultReg, immediate = 0), null)
+                    }
                     BaseDataType.FLOAT -> {
                         actualResultReg2 = codeGen.registers.next(IRDataType.WORD)
                         addInstr(result, IRInstruction(Opcode.FTOSW, IRDataType.FLOAT, reg1=actualResultReg2, fpReg1 = tr.resultFpReg), null)
@@ -775,6 +795,37 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     else -> throw AssemblyError("weird cast value type ${cast.position}")
                 }
             }
+            BaseDataType.LONG -> {
+                when(valueDt.base) {
+                    BaseDataType.UBYTE, BaseDataType.BOOL -> {
+                        // ubyte to long: double sign extend
+                        val wordreg = codeGen.registers.next(IRDataType.WORD)
+                        actualResultReg2 = codeGen.registers.next(IRDataType.LONG)
+                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.BYTE, reg1 = wordreg, reg2=tr.resultReg), null)
+                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.WORD, reg1 = actualResultReg2, reg2=wordreg), null)
+                    }
+                    BaseDataType.BYTE -> {
+                        // byte to long: double sign extend
+                        val wordreg = codeGen.registers.next(IRDataType.WORD)
+                        actualResultReg2 = codeGen.registers.next(IRDataType.LONG)
+                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.BYTE, reg1 = wordreg, reg2=tr.resultReg), null)
+                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.WORD, reg1 = actualResultReg2, reg2=wordreg), null)
+                    }
+                    BaseDataType.UWORD, BaseDataType.POINTER -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.LONG)
+                        addInstr(result, IRInstruction(Opcode.EXT, type = IRDataType.WORD, reg1 = actualResultReg2, reg2=tr.resultReg), null)
+                    }
+                    BaseDataType.WORD -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.LONG)
+                        addInstr(result, IRInstruction(Opcode.EXTS, type = IRDataType.WORD, reg1 = actualResultReg2, reg2=tr.resultReg), null)
+                    }
+                    BaseDataType.FLOAT -> {
+                        actualResultReg2 = codeGen.registers.next(IRDataType.LONG)
+                        addInstr(result, IRInstruction(Opcode.FTOSL, IRDataType.FLOAT, reg1=actualResultReg2, fpReg1 = tr.resultFpReg), null)
+                    }
+                    else -> throw AssemblyError("weird cast $valueDt to long ${cast.position}")
+                }
+            }
             BaseDataType.FLOAT -> {
                 actualResultFpReg2 = codeGen.registers.next(IRDataType.FLOAT)
                 when(valueDt.base) {
@@ -790,6 +841,9 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                     BaseDataType.WORD -> {
                         addInstr(result, IRInstruction(Opcode.FFROMSW, IRDataType.FLOAT, reg1=tr.resultReg, fpReg1 = actualResultFpReg2), null)
                     }
+                    BaseDataType.LONG ->  {
+                        addInstr(result, IRInstruction(Opcode.FFROMSL, IRDataType.FLOAT, reg1=tr.resultReg, fpReg1 = actualResultFpReg2), null)
+                    }
                     else -> throw AssemblyError("weird cast value type ${cast.position}")
                 }
             }
@@ -799,7 +853,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 // no further conversion required, pointers are all just uwords
             }
             BaseDataType.ARRAY_POINTER -> {
-                TODO("typecast to array of pointers $valueDt -> ${cast.type}")
+                TODO("typecast to array of pointers $valueDt -> ${cast.type}  ${cast.position}")
             }
             else -> throw AssemblyError("weird cast value type ${cast.position}")
         }
@@ -1137,7 +1191,7 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                         fcallArgs = FunctionCallArgs(argRegisters, returnRegisters)
                     )
                 }
-                else TODO("extsub with banked address got called ${callTarget.name}")
+                else TODO("extsub with banked address got called ${callTarget.name}  ${fcall.position}")
             }
         addInstr(result, call, null)
         val resultRegs = returnRegisters.filter{it.dt!=IRDataType.FLOAT}.map{it.registerNum}
@@ -1715,9 +1769,9 @@ internal class ExpressionGen(private val codeGen: IRCodeGen) {
                 result += pointerTr.chunks
                 pointerReg = pointerTr.resultReg
             } else if(left.pointerderef!=null) {
-                TODO("get pointer from deref $left")
+                TODO("get pointer from deref $left  ${left.position}")
             } else {
-                throw AssemblyError("weird arrayindexer $left")
+                throw AssemblyError("weird arrayindexer $left  ${left.position}")
             }
             val struct = left.type.subType!! as StStruct
             val constindex = left.index as? PtNumber

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

@@ -47,6 +47,7 @@ class IRCodeGen(
         ensureFirstChunkLabels(irProg)
         irProg.linkChunks()
         irProg.convertAsmChunks()
+        irProg.splitSSAchunks()
 
         // the optimizer also does 1 essential step regardless of optimizations: joining adjacent chunks.
         val optimizer = IRPeepholeOptimizer(irProg)
@@ -156,6 +157,7 @@ class IRCodeGen(
                             is IRCodeChunk -> {
                                 val replacement = IRCodeChunk(sub.label, first.next)
                                 replacement.instructions += first.instructions
+                                replacement.appendSrcPositions(first.sourceLinesPositions)
                                 replacement
                             }
                             is IRInlineAsmChunk -> IRInlineAsmChunk(sub.label, first.assembly, first.isIR, first.next)
@@ -271,7 +273,7 @@ class IRCodeGen(
             is PtSub -> throw AssemblyError("nested subroutines should have been flattened $node")
             is PtStructDecl -> emptyList()
             is PtSubSignature -> emptyList()
-            else -> TODO("missing codegen for $node")
+            else -> TODO("missing codegen for $node  ${node.position}")
         }
 
         val nonEmptyChunks = chunks.filter { it.isNotEmpty() || it.label != null }
@@ -409,6 +411,7 @@ class IRCodeGen(
             is IRCodeChunk -> {
                 val newChunk = IRCodeChunk(label, first.next)
                 newChunk.instructions += first.instructions
+                newChunk.appendSrcPositions(first.sourceLinesPositions)
                 newChunk
             }
             is IRInlineAsmChunk -> {
@@ -1141,7 +1144,7 @@ class IRCodeGen(
             if(identifier!=null && !isIndirectJump(goto))
                 IRInstruction(branchOpcode, labelSymbol = identifier.name)
             else
-                TODO("JUMP to expression address ${goto.target}")
+                TODO("JUMP to expression address ${goto.target}  ${goto.position}")
         }
     }
 
@@ -1859,7 +1862,7 @@ class IRCodeGen(
                 is PtNop -> { /* nothing */ }
                 is PtAssignment, is PtAugmentedAssign -> { /* global variable initialization is done elsewhere */ }
                 is PtVariable, is PtConstant, is PtMemMapped -> { /* vars should be looked up via symbol table */ }
-                is PtAlign -> TODO("ir support for inline %align")
+                is PtAlign -> TODO("ir support for inline %align  ${child.position}")
                 is PtSub -> {
                     val sub = IRSubroutine(child.name, translateParameters(child.signature.children), child.signature.returns, child.position)
                     for (subchild in child.children) {
@@ -1911,7 +1914,7 @@ class IRCodeGen(
                     }
                 }
                 is PtStructDecl -> { /* do nothing, should be found in the symbol table */ }
-                else -> TODO("weird block child node $child")
+                else -> TODO("weird block child node $child  ${child.position}")
             }
         }
         return irBlock
@@ -1975,12 +1978,17 @@ class IRCodeGen(
             in Cx16VirtualRegisters -> {
                 chunk += IRInstruction(Opcode.STOREM, paramDt, reg1=resultReg, labelSymbol = "cx16.${registerOrFlag.registerOrPair.toString().lowercase()}")
             }
+            in combinedLongRegisters -> {
+                require(paramDt==IRDataType.LONG)
+                val startreg = registerOrFlag.registerOrPair!!.startregname()
+                chunk += IRInstruction(Opcode.STOREM, paramDt, reg1=resultReg, labelSymbol = "cx16.${startreg}")
+            }
             null -> when(registerOrFlag.statusflag) {
                 // TODO: do the statusflag argument as last
                 Statusflag.Pc -> chunk += IRInstruction(Opcode.LSR, paramDt, reg1=resultReg)
                 else -> throw AssemblyError("unsupported statusflag as param")
             }
-            else -> throw AssemblyError("unsupported register arg")
+            else -> throw AssemblyError("unsupported register arg $registerOrFlag")
         }
         return chunk
     }

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

@@ -161,10 +161,10 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
             if(chunk.label!=null)
                 return false
             if(previous is IRCodeChunk && chunk is IRCodeChunk) {
-                // if the previous chunk doesn't end in a jump or a return, flow continues into the next chunk
+                // if the previous chunk doesn't end in a SSA branching instruction, flow continues into the next chunk, so they may be joined
                 val lastInstruction = previous.instructions.lastOrNull()
                 if(lastInstruction!=null)
-                    return lastInstruction.opcode !in OpcodesThatJump
+                    return lastInstruction.opcode !in OpcodesThatEndSSAblock
                 return true
             }
             return false
@@ -300,16 +300,16 @@ class IRPeepholeOptimizer(private val irprog: IRProgram) {
             // remove useless RETURN
             if(idx>0 && (ins.opcode == Opcode.RETURN || ins.opcode==Opcode.RETURNR || ins.opcode==Opcode.RETURNI)) {
                 val previous = chunk.instructions[idx-1]
-                if(previous.opcode in OpcodesThatJump) {
+                if(previous.opcode in OpcodesThatBranchUnconditionally) {
                     chunk.instructions.removeAt(idx)
                     changed = true
                 }
             }
 
             // replace subsequent opcodes that jump by just the first
-            if(idx>0 && (ins.opcode in OpcodesThatJump)) {
+            if(idx>0 && (ins.opcode in OpcodesThatBranchUnconditionally)) {
                 val previous = chunk.instructions[idx-1]
-                if(previous.opcode in OpcodesThatJump) {
+                if(previous.opcode in OpcodesThatBranchUnconditionally) {
                     chunk.instructions.removeAt(idx)
                     changed = true
                 }
@@ -362,7 +362,7 @@ jump p8_label_gen_2
          */
         var changed=false
         indexedInstructions.reversed().forEach { (idx, ins) ->
-            if(idx>=2 && ins.opcode in OpcodesThatJump) {
+            if(idx>=2 && ins.opcode in OpcodesThatBranchUnconditionally) {
                 val previous = indexedInstructions[idx-1].value
                 val previous2 = indexedInstructions[idx-2].value
                 if(previous.opcode==Opcode.LOADR && previous2.opcode in OpcodesThatLoad) {
@@ -415,13 +415,19 @@ jump p8_label_gen_2
                             chunk.instructions.removeAt(idx)
                             changed = true
                         }
+                        -1 if ins.type == IRDataType.LONG -> {
+                            chunk.instructions.removeAt(idx)
+                            changed = true
+                        }
                     }
                 }
                 Opcode.OR -> {
                     if (ins.immediate == 0) {
                         chunk.instructions.removeAt(idx)
                         changed = true
-                    } else if ((ins.immediate == 255 && ins.type == IRDataType.BYTE) || (ins.immediate == 65535 && ins.type == IRDataType.WORD)) {
+                    } else if ((ins.immediate == 255 && ins.type == IRDataType.BYTE) ||
+                               (ins.immediate == 65535 && ins.type == IRDataType.WORD) ||
+                               (ins.immediate == -1 && ins.type == IRDataType.LONG)) {
                         chunk.instructions[idx] = IRInstruction(Opcode.LOAD, ins.type, reg1 = ins.reg1, immediate = ins.immediate)
                         changed = true
                     }
@@ -434,6 +440,54 @@ jump p8_label_gen_2
                 }
                 else -> {}
             }
+
+            fun optimizeImmediateLoadAssociative(replacementOpcode: Opcode) {
+
+                fun getImmediateLoad(reg: Int): Pair<Int, Int>? {
+                    // look if the given register gets an immediate value 1 or 2 istructions back
+                    // returns (index of load instruction, immediate value) or null.
+                    if(idx>=1) {
+                        val previous = indexedInstructions[idx-1].value
+                        if(previous.opcode==Opcode.LOAD && previous.reg1==reg)
+                            return idx-1 to previous.immediate!!
+                    }
+                    if(idx>=2) {
+                        val previous = indexedInstructions[idx-2].value
+                        if(previous.opcode==Opcode.LOAD && previous.reg1==reg)
+                            return idx-2 to previous.immediate!!
+                    }
+                    return null
+                }
+
+                if(ins.reg1!=null) {
+                    val immediate1 = getImmediateLoad(ins.reg1!!)
+                    if(immediate1!=null) {
+                        chunk.instructions[idx] = IRInstruction(replacementOpcode, ins.type, reg1 = ins.reg2, immediate = immediate1.second)
+                        chunk.instructions.removeAt(immediate1.first)
+                        changed=true
+                    } else {
+                        val immediate2 = getImmediateLoad(ins.reg2!!)
+                        if (immediate2 != null) {
+                            chunk.instructions[idx] = IRInstruction(replacementOpcode, ins.type, reg1 = ins.reg1, immediate = immediate2.second)
+                            chunk.instructions.removeAt(immediate2.first)
+                            changed=true
+                        }
+                    }
+                }
+            }
+
+            // try to use immediate arithmetic instruction if possible
+            when(ins.opcode) {
+                Opcode.ADDR -> optimizeImmediateLoadAssociative(Opcode.ADD)
+                Opcode.MULR -> optimizeImmediateLoadAssociative(Opcode.MUL)
+                Opcode.MULSR -> optimizeImmediateLoadAssociative(Opcode.MULS)
+//                Opcode.SUBR -> TODO("ir peephole Subr")
+//                Opcode.DIVR -> TODO("ir peephole Divr")
+//                Opcode.DIVSR -> TODO("ir peephole Divsr")
+//                Opcode.MODR -> TODO("ir peephole Modr")
+//                Opcode.DIVMODR -> TODO("ir peephole DivModr")
+                else -> {}
+            }
         }
         return changed
     }

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

@@ -7,6 +7,7 @@ internal class RegisterPool {
     // everything from 99000 onwards is reserved for special purposes:
     // 99000 - 99099 : WORD registers for syscall arguments and response value(s)
     // 99100 - 99199 : BYTE registers for syscall arguments and response value(s)
+    // 99200 - 99299 : LONG registers for syscall arguments and response value(s)
 
     private var nextRegister: Int=1
     private val registerTypes: MutableMap<Int, IRDataType> = mutableMapOf()
@@ -18,6 +19,8 @@ internal class RegisterPool {
             registerTypes[i] = IRDataType.WORD
         for(i in 99100..99199)
             registerTypes[i] = IRDataType.BYTE
+        for(i in 99200..99299)
+            registerTypes[i] = IRDataType.LONG
     }
 
     fun next(type: IRDataType): Int {

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

@@ -47,10 +47,16 @@ private fun convert(struct: StStruct): IRStStructDef =
     IRStStructDef(struct.scopedNameString, struct.fields, struct.size)
 
 
-private fun convertArrayElt(elt: StArrayElement): IRStArrayElement = if(elt.boolean!=null)
-    IRStArrayElement(elt.boolean, null, elt.addressOfSymbol)
-else
-    IRStArrayElement(null, elt.number, elt.addressOfSymbol)
+private fun convertArrayElt(elt: StArrayElement): IRStArrayElement {
+    return if (elt.boolean != null)
+        IRStArrayElement(elt.boolean, null, null)
+    else if(elt.number!=null)
+        IRStArrayElement(null, elt.number, null)
+    else {
+        val symbol = elt.addressOfSymbol ?: (StStructInstanceBlockName + "." + (elt.structInstance ?: elt.structInstanceUninitialized))
+        IRStArrayElement(null, null, symbol)
+    }
+}
 
 
 private fun convert(variable: StStaticVariable): IRStStaticVariable {
@@ -72,7 +78,8 @@ private fun convert(variable: StStaticVariable): IRStStaticVariable {
             val newArray = mutableListOf<IRStArrayElement>()
             array.forEach {
                 if(it.addressOfSymbol!=null) {
-                    val target = variable.lookup(it.addressOfSymbol!!) ?: throw NoSuchElementException("can't find variable ${it.addressOfSymbol}")
+                    val target = variable.lookup(it.addressOfSymbol!!) ?:
+                        throw NoSuchElementException("can't find variable ${it.addressOfSymbol}")
                     newArray.add(IRStArrayElement(null, null, target.scopedNameString))
                 } else {
                     newArray.add(convertArrayElt(it))
@@ -115,7 +122,6 @@ private fun convert(variable: StMemVar): IRStMemVar {
 
 
 private fun convert(constant: StConstant): IRStConstant {
-    val dt = DataType.forDt(constant.dt)
     val scopedName = if('.' in constant.name) {
         constant.name
     } else {
@@ -125,15 +131,15 @@ private fun convert(constant: StConstant): IRStConstant {
             constant.name
         }
     }
-    return IRStConstant(scopedName, dt, constant.value)
+    return IRStConstant(scopedName, constant.dt, constant.value)
 }
 
 
-private fun convert(variable: StMemorySlab): IRStMemorySlab {
-    return if('.' in variable.name)
-        IRStMemorySlab(variable.name, variable.size, variable.align)
+private fun convert(mem: StMemorySlab): IRStMemorySlab {
+    return if('.' in mem.name)
+        IRStMemorySlab(mem.name, mem.size, mem.align)
     else
-        IRStMemorySlab("$StMemorySlabPrefix.${variable.name}", variable.size, variable.align)
+        IRStMemorySlab("$StMemorySlabBlockName.${mem.name}", mem.size, mem.align)
 }
 
 
@@ -142,8 +148,8 @@ private fun convert(instance: StStructInstance, fields: Iterable<Pair<DataType,
         val elt = convertArrayElt(value)
         IRStructInitValue(field.first.base, elt)
     }
-    return IRStStructInstance(instance.name, instance.structName, values, instance.size)
+    return if('.' in instance.name)
+        IRStStructInstance(instance.name, instance.structName, values, instance.size)
+    else
+        IRStStructInstance("${StStructInstanceBlockName}.${instance.name}", instance.structName, values, instance.size)
 }
-
-
-internal const val StMemorySlabPrefix = "prog8_slabs"       // TODO also add  ".prog8_memoryslab_"  ?

codeGenIntermediate/test/Dummies.kt

@@ -10,12 +10,14 @@ internal object DummyMemsizer : IMemSizer {
             return when(dt.sub) {
                 BaseDataType.BOOL, BaseDataType.BYTE, BaseDataType.UBYTE -> numElements
                 BaseDataType.UWORD, BaseDataType.WORD -> numElements*2
+                BaseDataType.LONG -> numElements*4
                 BaseDataType.FLOAT -> numElements*5
                 else -> throw IllegalArgumentException("invalid sub type")
             }
         }
         return when {
             dt.isByteOrBool -> 1 * (numElements ?: 1)
+            dt.isLong -> 4 * (numElements ?: 1)
             dt.isFloat -> 5 * (numElements ?: 1)
             else -> 2 * (numElements ?: 1)
         }

codeGenIntermediate/test/TestIRPeepholeOpt.kt

@@ -23,6 +23,7 @@ class TestIRPeepholeOpt: FunSpec({
             noSysInit = true,
             romable = false,
             compTarget = target,
+            compilerVersion="99.99",
             loadAddress = target.PROGRAM_LOAD_ADDRESS,
             memtopAddress = 0xffffu
         )
@@ -204,21 +205,27 @@ class TestIRPeepholeOpt: FunSpec({
         val irProg = makeIRProgram(listOf(
             IRInstruction(Opcode.AND, IRDataType.BYTE, reg1=42, immediate = 0),
             IRInstruction(Opcode.AND, IRDataType.WORD, reg1=42, immediate = 0),
+            IRInstruction(Opcode.AND, IRDataType.LONG, reg1=42, immediate = 0),
             IRInstruction(Opcode.OR, IRDataType.BYTE, reg1=42, immediate = 255),
-            IRInstruction(Opcode.OR, IRDataType.WORD, reg1=42, immediate = 65535)
+            IRInstruction(Opcode.OR, IRDataType.WORD, reg1=42, immediate = 65535),
+            IRInstruction(Opcode.OR, IRDataType.LONG, reg1=42, immediate = -1)
         ))
-        irProg.chunks().single().instructions.size shouldBe 4
+        irProg.chunks().single().instructions.size shouldBe 6
         val opt = IRPeepholeOptimizer(irProg)
         opt.optimize(true, ErrorReporterForTests())
         val instr = irProg.chunks().single().instructions
-        instr.size shouldBe 4
+        instr.size shouldBe 6
         instr[0].opcode shouldBe Opcode.LOAD
         instr[1].opcode shouldBe Opcode.LOAD
         instr[2].opcode shouldBe Opcode.LOAD
         instr[3].opcode shouldBe Opcode.LOAD
+        instr[4].opcode shouldBe Opcode.LOAD
+        instr[5].opcode shouldBe Opcode.LOAD
         instr[0].immediate shouldBe 0
         instr[1].immediate shouldBe 0
-        instr[2].immediate shouldBe 255
-        instr[3].immediate shouldBe 65535
+        instr[2].immediate shouldBe 0
+        instr[3].immediate shouldBe 255
+        instr[4].immediate shouldBe 65535
+        instr[5].immediate shouldBe -1
     }
 })

codeGenIntermediate/test/TestVmCodeGen.kt

@@ -25,6 +25,7 @@ class TestVmCodeGen: FunSpec({
             noSysInit = false,
             romable = false,
             compTarget = target,
+            compilerVersion="99.99",
             loadAddress = target.PROGRAM_LOAD_ADDRESS,
             memtopAddress = 0xffffu
         )
@@ -340,7 +341,7 @@ class TestVmCodeGen: FunSpec({
         val errors = ErrorReporterForTests()
         val result = codegen.generate(program, st, options, errors) as VmAssemblyProgram
         val irChunks = (result.irProgram.blocks.first().children.single() as IRSubroutine).chunks
-        irChunks.size shouldBe 1
+        irChunks.size shouldBe 2
     }
 
     test("integer comparison expressions against zero") {
@@ -536,7 +537,7 @@ class TestVmCodeGen: FunSpec({
         val errors = ErrorReporterForTests()
         val result = codegen.generate(program, st, options, errors) as VmAssemblyProgram
         val irChunks = (result.irProgram.blocks.first().children.single() as IRSubroutine).chunks
-        irChunks.size shouldBe 1
+        irChunks.size shouldBe 2
     }
 
     test("extsub allowed in ir-codegen") {

codeOptimizers/src/prog8/optimizer/ConstantFoldingOptimizer.kt

@@ -39,7 +39,10 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
             // see if LONG values may be reduced to something smaller
             val smaller = NumericLiteral.optimalInteger(numLiteral.number.toInt(), numLiteral.position)
             if(smaller.type!=BaseDataType.LONG) {
-                return listOf(IAstModification.ReplaceNode(numLiteral, smaller, parent))
+                if(parent !is Assignment || !parent.target.inferType(program).isLong) {
+                    // do NOT reduce the type if the target of the assignment is a long
+                    return listOf(IAstModification.ReplaceNode(numLiteral, smaller, parent))
+                }
             }
         }
 
@@ -85,6 +88,8 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
      *        (X + c1) - c2  ->  X + (c1-c2)
      */
     override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        if(expr.operator==".")
+            return noModifications
         val modifications = mutableListOf<IAstModification>()
         val leftconst = expr.left.constValue(program)
         val rightconst = expr.right.constValue(program)
@@ -344,7 +349,7 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
                         }
                     }
                 } else if(arrayIndexedExpression.pointerderef!=null) {
-                    TODO("constant fold pointer[i]")
+                    TODO("constant fold pointer[i]  ${arrayIndexedExpression.position}")
                 }
             }
         }
@@ -438,7 +443,7 @@ class ConstantFoldingOptimizer(private val program: Program, private val errors:
         val numval = decl.value as? NumericLiteral
         if(decl.type== VarDeclType.CONST && numval!=null) {
             val valueDt = numval.inferType(program)
-            if(valueDt issimpletype BaseDataType.LONG) {
+            if(valueDt issimpletype BaseDataType.LONG || decl.datatype.isLong) {
                 return noModifications  // this is handled in the numericalvalue case
             }
             if(!(valueDt istype decl.datatype)) {

codeOptimizers/src/prog8/optimizer/ConstantIdentifierReplacer.kt

@@ -26,19 +26,15 @@ class VarConstantValueTypeAdjuster(
     override fun after(decl: VarDecl, parent: Node): Iterable<IAstModification> {
 
         if(decl.parent is AnonymousScope)
-            throw FatalAstException("vardecl may no longer occur in anonymousscope")
+            throw FatalAstException("vardecl may no longer occur in anonymousscope ${decl.position}")
 
-        try {
-            val declConstValue = decl.value?.constValue(program)
-            if(declConstValue!=null && (decl.type== VarDeclType.VAR || decl.type==VarDeclType.CONST)
-                && declConstValue.type != decl.datatype.base) {
-                if(decl.datatype.isInteger && declConstValue.type == BaseDataType.FLOAT) {
-                    // avoid silent float roundings
-                    errors.err("refused truncating of float to avoid loss of precision", decl.value!!.position)
-                }
+        val declConstValue = decl.value?.constValue(program)
+        if(declConstValue!=null && (decl.type== VarDeclType.VAR || decl.type==VarDeclType.CONST)
+            && declConstValue.type != decl.datatype.base) {
+            if(decl.datatype.isInteger && declConstValue.type == BaseDataType.FLOAT) {
+                // avoid silent float roundings
+                errors.err("refused truncating of float to avoid loss of precision", decl.value!!.position)
             }
-        } catch (x: UndefinedSymbolError) {
-            errors.err(x.message, x.position)
         }
 
         // replace variables by constants, if possible
@@ -173,7 +169,12 @@ class VarConstantValueTypeAdjuster(
                 val replaceFunc = if(t1.isBytes) {
                     if(t1 issimpletype BaseDataType.BYTE) "clamp__byte" else "clamp__ubyte"
                 } else if(t1.isInteger) {
-                    if(t1 issimpletype BaseDataType.WORD) "clamp__word" else "clamp__uword"
+                    when {
+                        t1 issimpletype BaseDataType.WORD -> "clamp__word"
+                        t1 issimpletype BaseDataType.UWORD -> "clamp__uword"
+                        t1 issimpletype BaseDataType.LONG -> "clamp__long"
+                        else -> throw FatalAstException("clamp type")
+                    }
                 } else {
                     errors.err("clamp builtin not supported for floats, use floats.clamp", functionCallExpr.position)
                     return noModifications
@@ -195,10 +196,12 @@ class VarConstantValueTypeAdjuster(
                     else
                         "${funcName}__ubyte"
                 } else if(t1.isInteger && t2.isInteger) {
-                    replaceFunc = if(t1 issimpletype BaseDataType.WORD || t2 issimpletype BaseDataType.WORD)
-                        "${funcName}__word"
-                    else
-                        "${funcName}__uword"
+                    replaceFunc = when {
+                        t1 issimpletype BaseDataType.LONG || t2 issimpletype BaseDataType.LONG -> "${funcName}__long"
+                        t1 issimpletype BaseDataType.WORD || t2 issimpletype BaseDataType.WORD -> "${funcName}__word"
+                        t1 issimpletype BaseDataType.UWORD || t2 issimpletype BaseDataType.UWORD -> "${funcName}__uword"
+                        else -> throw FatalAstException("min/max type")
+                    }
                 } else if(t1.isNumeric && t2.isNumeric) {
                     errors.err("min/max not supported for floats", functionCallExpr.position)
                     return noModifications
@@ -218,6 +221,7 @@ class VarConstantValueTypeAdjuster(
                 val replaceFunc = when {
                     dt.isSignedByte -> "abs__byte"
                     dt.isSignedWord -> "abs__word"
+                    dt.isLong -> "abs__long"
                     dt.isFloat -> "abs__float"
                     dt.isUnsignedByte || dt.isUnsignedWord -> {
                         return listOf(IAstModification.ReplaceNode(functionCallExpr, functionCallExpr.args[0], parent))
@@ -240,8 +244,16 @@ class VarConstantValueTypeAdjuster(
                     dt.isUnsignedByte -> "sqrt__ubyte"
                     dt.isUnsignedWord -> "sqrt__uword"
                     dt.isFloat -> "sqrt__float"
+                    dt.isLong -> {
+                        val value = functionCallExpr.args[0].constValue(program)?.number
+                        if(value!=null && value<0) {
+                            errors.err("expected unsigned or float numeric argument", functionCallExpr.args[0].position)
+                            return noModifications
+                        }
+                        "sqrt__long"
+                    }
                     else -> {
-                        errors.err("expected unsigned or float numeric argument", functionCallExpr.args[0].position)
+                        errors.err("expected numeric argument", functionCallExpr.args[0].position)
                         return noModifications
                     }
                 }
@@ -250,6 +262,15 @@ class VarConstantValueTypeAdjuster(
                     functionCallExpr))
             }
         }
+        else if(func==listOf("lsb") || func==listOf("msb")) {
+            val t1 = functionCallExpr.args[0].inferType(program)
+            if(t1.isLong) {
+                val replaceFunc = func[0]+"__long"
+                return listOf(IAstModification.SetExpression({functionCallExpr.target = it as IdentifierReference},
+                    IdentifierReference(listOf(replaceFunc), functionCallExpr.target.position),
+                    functionCallExpr))
+            }
+        }
         return noModifications
     }
 
@@ -313,41 +334,36 @@ internal class ConstantIdentifierReplacer(
         if(!dt.isKnown || !dt.isNumeric && !dt.isBool)
             return noModifications
 
-        try {
-            val cval = identifier.constValue(program) ?: return noModifications
-            val arrayIdx = identifier.parent as? ArrayIndexedExpression
-            if(arrayIdx!=null && cval.type.isNumeric) {
-                // special case when the identifier is used as a pointer var
-                // var = constpointer[x] --> var = @(constvalue+x) [directmemoryread]
-                // constpointer[x] = var -> @(constvalue+x) [directmemorywrite] = var
-                val add = BinaryExpression(NumericLiteral(cval.type, cval.number, identifier.position), "+", arrayIdx.indexer.indexExpr, identifier.position)
-                return if(arrayIdx.parent is AssignTarget) {
-                    val memwrite = DirectMemoryWrite(add, identifier.position)
-                    val assignTarget = AssignTarget(null, null, memwrite, null, false, position = identifier.position)
-                    listOf(IAstModification.ReplaceNode(arrayIdx.parent, assignTarget, arrayIdx.parent.parent))
-                } else {
-                    val memread = DirectMemoryRead(add, identifier.position)
-                    listOf(IAstModification.ReplaceNode(arrayIdx, memread, arrayIdx.parent))
-                }
+        val cval = identifier.constValue(program) ?: return noModifications
+        val arrayIdx = identifier.parent as? ArrayIndexedExpression
+        if(arrayIdx!=null && cval.type.isNumeric) {
+            // special case when the identifier is used as a pointer var
+            // var = constpointer[x] --> var = @(constvalue+x) [directmemoryread]
+            // constpointer[x] = var -> @(constvalue+x) [directmemorywrite] = var
+            val add = BinaryExpression(NumericLiteral(cval.type, cval.number, identifier.position), "+", arrayIdx.indexer.indexExpr, identifier.position)
+            return if(arrayIdx.parent is AssignTarget) {
+                val memwrite = DirectMemoryWrite(add, identifier.position)
+                val assignTarget = AssignTarget(null, null, memwrite, null, false, position = identifier.position)
+                listOf(IAstModification.ReplaceNode(arrayIdx.parent, assignTarget, arrayIdx.parent.parent))
+            } else {
+                val memread = DirectMemoryRead(add, identifier.position)
+                listOf(IAstModification.ReplaceNode(arrayIdx, memread, arrayIdx.parent))
             }
-            when {
-                cval.type.isNumericOrBool -> {
-                    if(parent is AddressOf)
-                        return noModifications      // cannot replace the identifier INSIDE the addr-of here, let's do it later.
-                    return listOf(
-                        IAstModification.ReplaceNode(
-                            identifier,
-                            NumericLiteral(cval.type, cval.number, identifier.position),
-                            identifier.parent
-                        )
+        }
+        when {
+            cval.type.isNumericOrBool -> {
+                if(parent is AddressOf)
+                    return noModifications      // cannot replace the identifier INSIDE the addr-of here, let's do it later.
+                return listOf(
+                    IAstModification.ReplaceNode(
+                        identifier,
+                        NumericLiteral(cval.type, cval.number, identifier.position),
+                        identifier.parent
                     )
-                }
-                cval.type.isPassByRef -> throw InternalCompilerException("pass-by-reference type should not be considered a constant")
-                else -> return noModifications
+                )
             }
-        } catch (x: UndefinedSymbolError) {
-            errors.err(x.message, x.position)
-            return noModifications
+            cval.type.isPassByRef -> throw InternalCompilerException("pass-by-reference type should not be considered a constant")
+            else -> return noModifications
         }
     }
 

codeOptimizers/src/prog8/optimizer/ExpressionSimplifier.kt

@@ -9,6 +9,7 @@ import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
 import prog8.code.core.*
 import kotlin.math.abs
+import kotlin.math.floor
 import kotlin.math.log2
 
 
@@ -90,6 +91,8 @@ class ExpressionSimplifier(private val program: Program, private val errors: IEr
     }
 
     override fun after(expr: BinaryExpression, parent: Node): Iterable<IAstModification> {
+        if(expr.operator==".")
+            return noModifications
         val newExpr = applyAbsorptionLaws(expr)
         if(newExpr!=null)
             return listOf(IAstModification.ReplaceNode(expr, newExpr, parent))
@@ -244,15 +247,19 @@ class ExpressionSimplifier(private val program: Program, private val errors: IEr
             if (rightVal?.number == 1.0) {
                 if (rightDt != leftDt && !(leftDt.isPointer && rightDt.isUnsignedWord)) {
                     val dt = if(leftDt.isPointer) BaseDataType.UWORD else leftDt.base
-                    val right = NumericLiteral(dt, rightVal.number, rightVal.position)
-                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    if(!dt.isLong) {
+                        val right = NumericLiteral(dt, rightVal.number, rightVal.position)
+                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    }
                 }
             }
             else if (rightVal?.number == 0.0) {
                 if (rightDt != leftDt && !(leftDt.isPointer && rightDt.isUnsignedWord)) {
                     val dt = if(leftDt.isPointer) BaseDataType.UWORD else leftDt.base
-                    val right = NumericLiteral(dt, rightVal.number, rightVal.position)
-                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    if(!dt.isLong) {
+                        val right = NumericLiteral(dt, rightVal.number, rightVal.position)
+                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    }
                 }
             }
         }
@@ -266,15 +273,19 @@ class ExpressionSimplifier(private val program: Program, private val errors: IEr
             if (rightVal?.number == 1.0) {
                 if(rightDt!=leftDt && !(leftDt.isPointer && rightDt.isUnsignedWord)) {
                     val dt = if(leftDt.isPointer) BaseDataType.UWORD else leftDt.base
-                    val right = NumericLiteral(dt, rightVal.number, rightVal.position)
-                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    if(!dt.isLong) {
+                        val right = NumericLiteral(dt, rightVal.number, rightVal.position)
+                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    }
                 }
             }
             else if (rightVal?.number == 0.0) {
                 if(rightDt!=leftDt && !(leftDt.isPointer && rightDt.isUnsignedWord)) {
                     val dt = if(leftDt.isPointer) BaseDataType.UWORD else leftDt.base
-                    val right = NumericLiteral(dt, rightVal.number, rightVal.position)
-                    return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    if(!dt.isLong) {
+                        val right = NumericLiteral(dt, rightVal.number, rightVal.position)
+                        return listOf(IAstModification.ReplaceNode(expr.right, right, expr))
+                    }
                 }
             }
         }
@@ -424,19 +435,92 @@ class ExpressionSimplifier(private val program: Program, private val errors: IEr
             }
         }
 
+//        fun isPowerOfTwo(number: Double): Boolean {
+//            val intValue = number.toInt()
+//            return intValue > 0 && (intValue and (intValue - 1)) == 0
+//        }
+
+        fun isFactorOf256(number: Double): Boolean {
+            val intValue = number.toInt()
+            return intValue >= 256 && (intValue and 0xFF) == 0
+        }
+
+        if (leftDt.isUnsignedWord && rightVal!=null) {
+            if (expr.operator == ">" && rightVal.number == 255.0 || expr.operator == ">=" && rightVal.number == 256.0) {
+                // uword > 255  -->  msb(value)!=0
+                // uword >= 256 -->  msb(value)!=0
+                expr.operator = "!="
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, 0.0, expr.right.position)
+                expr.linkParents(parent)
+            }
+            else if(expr.operator==">=" && isFactorOf256(rightVal.number)) {
+                // uword >= $xx00 --> msb(value)>=xx
+                val msbFraction = floor(rightVal.number / 256.0)
+                expr.operator = ">="
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, msbFraction, expr.right.position)
+                expr.linkParents(parent)
+            }
+            else if(expr.operator==">" && isFactorOf256(rightVal.number+1)) {
+                // uword > $xxFF --> msb(value)>xx
+                val msbFraction = floor((rightVal.number) / 256.0)
+                expr.operator = ">"
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, msbFraction, expr.right.position)
+                expr.linkParents(parent)
+            }
+            else if(expr.operator == "<" && rightVal.number == 256.0 || expr.operator == "<=" && rightVal.number == 255.0) {
+                // uword < 256  -->  msb(value)==0
+                // uword <= 255 -->  msb(value)==0
+                expr.operator = "=="
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, 0.0, expr.right.position)
+                expr.linkParents(parent)
+            }
+            else if(expr.operator == "<" && isFactorOf256(rightVal.number)) {
+                // uword < $xx00 --> msb(value)<xx
+                val msbFraction = floor(rightVal.number / 256.0)
+                expr.operator = "<"
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, msbFraction, expr.right.position)
+                expr.linkParents(parent)
+            }
+            else if(expr.operator=="<=" && isFactorOf256(rightVal.number+1)) {
+                // uword <= $xxFF --> msb(value)<=xx
+                val msbFraction = floor((rightVal.number) / 256.0)
+                expr.operator = "<="
+                expr.left = FunctionCallExpression(IdentifierReference(listOf("msb"), expr.left.position), mutableListOf(expr.left), expr.left.position)
+                expr.right = NumericLiteral(BaseDataType.UBYTE, msbFraction, expr.right.position)
+                expr.linkParents(parent)
+            }
+
+        }
+
         return noModifications
     }
 
     override fun after(arrayIndexedExpression: ArrayIndexedExpression, parent: Node): Iterable<IAstModification> {
         if(arrayIndexedExpression.indexer.constIndex()==0) {
             if(arrayIndexedExpression.plainarrayvar!=null) {
-                if((arrayIndexedExpression.parent as? BinaryExpression)?.operator !=".") {
+                val binexprParent = arrayIndexedExpression.parent as? BinaryExpression
+                if(binexprParent?.operator!=".") {
                     val dt = arrayIndexedExpression.plainarrayvar!!.inferType(program).getOrUndef()
                     if(dt.isPointer) {
                         // pointer[0]  -->   pointer^^
                         val deref = PtrDereference(arrayIndexedExpression.plainarrayvar!!.nameInSource, true, arrayIndexedExpression.plainarrayvar!!.position)
                         return listOf(IAstModification.ReplaceNode(arrayIndexedExpression,deref, parent))
                     }
+                } else if(arrayIndexedExpression.pointerderef==null) {
+                    // possibly     pointer[0].field   -->  pointer.field
+                    val target = arrayIndexedExpression.plainarrayvar!!.targetVarDecl()
+                    if(target?.datatype?.isPointer==true) {
+                        val field = (binexprParent.right as? IdentifierReference)?.nameInSource
+                        if(field!=null) {
+                            val deref = PtrDereference(arrayIndexedExpression.plainarrayvar!!.nameInSource + field, false, arrayIndexedExpression.plainarrayvar!!.position)
+                            return listOf(IAstModification.ReplaceNode(arrayIndexedExpression.parent, deref, arrayIndexedExpression.parent.parent))
+                        }
+                    }
                 }
             }
             val ptrDeref = arrayIndexedExpression.pointerderef

codeOptimizers/src/prog8/optimizer/UnusedCodeRemover.kt

@@ -8,7 +8,7 @@ import prog8.ast.expressions.TypecastExpression
 import prog8.ast.statements.*
 import prog8.ast.walk.AstWalker
 import prog8.ast.walk.IAstModification
-import prog8.code.INTERNED_STRINGS_MODULENAME
+import prog8.code.PROG8_CONTAINER_MODULES
 import prog8.code.core.ICompilationTarget
 import prog8.code.core.IErrorReporter
 import prog8.compiler.CallGraph
@@ -93,7 +93,7 @@ class UnusedCodeRemover(private val program: Program,
     override fun after(block: Block, parent: Node): Iterable<IAstModification> {
         if("force_output" !in block.options()) {
             if (block.containsNoCodeNorVars) {
-                if (block.name != INTERNED_STRINGS_MODULENAME && "ignore_unused" !in block.options()) {
+                if (block.name !in PROG8_CONTAINER_MODULES && "ignore_unused" !in block.options()) {
                     if (!block.statements.any { it is Subroutine && it.hasBeenInlined })
                         errors.info("removing unused block '${block.name}'", block.position)
                 }

compiler/build.gradle.kts

@@ -2,7 +2,8 @@ plugins {
     id("application")
     kotlin("jvm")
     // id("com.github.johnrengelman.shadow") version "8.1.1"
-    id("io.github.goooler.shadow") version "8.1.8"
+    // id("io.github.goooler.shadow") version "8.1.8"
+    id("com.gradleup.shadow") version "9.2.2"
     id("com.peterabeles.gversion") version "1.10.3"
 }
 

compiler/res/prog8lib/bcd.p8

@@ -0,0 +1,144 @@
+bcd {
+    ; Decimal addition an subtraction routines.
+    ; For CPUs that support BCD mode (binary coded decimal) (not all 6502 variants support this mode...)
+    ; This is useful for example for counting decimal score in a game, to avoid costly conversion to a decimal display string. Just print the hex representation.
+
+    sub addb(byte a, byte b) -> byte {
+        setbcd()
+        a += b
+        clearbcd()
+        return a
+    }
+
+    sub addub(ubyte a, ubyte b) -> ubyte {
+        setbcd()
+        a += b
+        clearbcd()
+        return a
+    }
+
+    sub addw(word a, word b) -> word {
+        setbcd()
+        a += b
+        clearbcd()
+        return a
+    }
+
+    sub adduw(uword a, uword b) -> uword {
+        setbcd()
+        a += b
+        clearbcd()
+        return a
+    }
+
+    sub addl(long a, long b) -> long {
+        setbcd()
+        a += b
+        clearbcd()
+        return a
+    }
+
+    sub addtol(^^long a, long b) {
+        ; -- inplace long BCD addition (avoids copying long values by value)
+        setbcd()
+        ;; NOT YET IMPLEMENTED IN PROG8: a^^ += b, so inline asm
+        %asm {{
+            lda  p8v_a
+            ldy  p8v_a+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+            clc
+            lda  (P8ZP_SCRATCH_W1),y
+            adc  p8v_b
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            adc  p8v_b+1
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            adc  p8v_b+2
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            adc  p8v_b+3
+            sta  (P8ZP_SCRATCH_W1),y
+        }}
+        clearbcd()
+    }
+
+    sub subb(byte a, byte b) -> byte {
+        setbcd()
+        a -= b
+        clearbcd()
+        return a
+    }
+
+    sub subub(ubyte a, ubyte b) -> ubyte {
+        setbcd()
+        a -= b
+        clearbcd()
+        return a
+    }
+
+    sub subuw(uword a, uword b) -> uword {
+        setbcd()
+        a -= b
+        clearbcd()
+        return a
+    }
+
+    sub subl(long a, long b) -> long {
+        setbcd()
+        a -= b
+        clearbcd()
+        return a
+    }
+
+    sub subfroml(^^long a, long b) {
+        ; -- inplace long BCD subtraction (avoids copying long values by value)
+        setbcd()
+        ;; NOT YET IMPLEMENTED IN PROG8: a^^ -= b, so inline asm
+        %asm {{
+            lda  p8v_a
+            ldy  p8v_a+1
+            sta  P8ZP_SCRATCH_W1
+            sty  P8ZP_SCRATCH_W1+1
+            ldy  #0
+            sec
+            lda  (P8ZP_SCRATCH_W1),y
+            sbc  p8v_b
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            sbc  p8v_b+1
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            sbc  p8v_b+2
+            sta  (P8ZP_SCRATCH_W1),y
+            iny
+            lda  (P8ZP_SCRATCH_W1),y
+            sbc  p8v_b+3
+            sta  (P8ZP_SCRATCH_W1),y
+        }}
+        clearbcd()
+    }
+
+
+    inline asmsub setbcd() {
+        ; be safe and 6502 compatible and prohibit interrupts during BCD mode
+        %asm {{
+            php
+            sei
+            sed
+        }}
+    }
+
+    inline asmsub clearbcd() {
+        %asm {{
+            plp
+        }}
+    }
+}

compiler/res/prog8lib/buffers.p8

@@ -70,7 +70,65 @@ smallringbuffer {
     }
 }
 
-; note: for a "small stack"  (256 bytes size) just use the CPU stack via sys.push[w] / sys.pop[w].
+
+smallstack {
+    ; -- A small stack (LIFO) that uses just 256 bytes and is independent of the CPU stack. Stack is growing downward from the top of the buffer.
+    ;    You can store and retrieve bytes and words. There are no guards against stack over/underflow.
+    ; note: for a "small stack"  (256 bytes size) you might also perhaps just use the CPU stack via sys.push[w] / sys.pop[w].
+
+    ubyte[256] buffer
+    ubyte sp = 255
+
+    sub init() {
+        sp = 255
+    }
+
+    sub size() -> ubyte {
+        return 255-sp
+    }
+
+    sub free() -> ubyte {
+        return sp
+    }
+
+    sub isfull() -> bool {
+        return sp==0
+    }
+
+    sub isempty() -> bool {
+        return sp==255
+    }
+
+    sub push(ubyte value) {
+        ; -- put a byte on the stack
+        buffer[sp] = value
+        sp--
+    }
+
+    sub pushw(uword value) {
+        ; -- put a word on the stack (lsb first then msb)
+        buffer[sp] = lsb(value)
+        sp--
+        buffer[sp] = msb(value)
+        sp--
+    }
+
+    sub pop() -> ubyte {
+        ; -- pops a byte off the stack
+        sp++
+        return buffer[sp]
+    }
+
+    sub popw() -> uword {
+        ; -- pops a word off the stack.
+        sp++
+        cx16.r0H = buffer[sp]
+        sp++
+        cx16.r0L = buffer[sp]
+        return cx16.r0
+    }
+}
+
 
 stack {
     ; -- A stack (LIFO) that uses a block of 8 KB of memory. Growing downward from the top of the buffer.

compiler/res/prog8lib/c128/syslib.p8

@@ -133,6 +133,29 @@ asmsub RDTIM16() clobbers(X) -> uword @AY {
     }}
 }
 
+asmsub SETTIML(long jiffies @R0R1_32) {
+    ; -- just like SETTIM, but with a single 32 bit (lower 24 bits used) argument.
+    %asm {{
+        lda  cx16.r0
+        ldx  cx16.r0+1
+        ldy  cx16.r0+2
+        jmp  SETTIM
+    }}
+}
+
+asmsub RDTIML() clobbers(X) -> long @R0R1_32 {
+    ; --  like RDTIM() and returning the timer value as a 32 bit (lower 24 bits used) value.
+    %asm {{
+        jsr  RDTIM
+        sta  cx16.r0
+        stx  cx16.r0+1
+        sty  cx16.r0+2
+        lda  #0
+        sta  cx16.r0+3
+        rts
+    }}
+}
+
 sub CLEARST() {
     ; -- Set the ST status variable back to 0. (there's no direct kernal call for this)
     ;    Note: a drive error state (blinking led) isn't cleared! You can use diskio.status() to clear that.
@@ -335,7 +358,7 @@ extsub $FF71 = JMPFAR() clobbers(A,X)                           ; jump without r
 extsub $FF74 = INDFET(ubyte zpaddr @A, ubyte bank @X, ubyte offset @Y) clobbers(X) -> ubyte @A ; fetch byte from another bank (address in zp, ptr in A)
 extsub $FF77 = INDSTA(ubyte value @A, ubyte bank @X, ubyte offset @Y) clobbers(X) ; store byte to another bank (address in zp, ptr in $02b9)
 extsub $FF7A = INDCMP(ubyte value @A, ubyte bank @X, ubyte offset @Y) clobbers(X) -> bool @Pz, bool @Pc, bool @Pv; compare byte in another bank (address in zp, ptr in $02c8)
-extsub $FF7D = PRIMM()                                          ; print immediate string
+extsub $FF7D = PRIMM()                                          ; print immediate string  (the 0-terminated string immediately following the JSR PRIMM instruction)
 
 
 ; ---- C128 specific system utility routines: ----
@@ -506,6 +529,7 @@ save_SCRATCH_ZPWORD2	.word  ?
 
 asmsub  set_irq(uword handler @AY) clobbers(A)  {
 	%asm {{
+	    php
 		sei
         sta  _vector
         sty  _vector+1
@@ -513,7 +537,7 @@ asmsub  set_irq(uword handler @AY) clobbers(A)  {
 		sta  cbm.CINV
 		lda  #>_irq_handler
 		sta  cbm.CINV+1
-		cli
+		plp
 		rts
 _irq_handler
         jsr  sys.save_prog8_internals
@@ -545,6 +569,7 @@ _vector	.word ?
 
 asmsub  restore_irq() clobbers(A) {
 	%asm {{
+	    php
 		sei
 		lda  #<cbm.IRQDFRT
 		sta  cbm.CINV
@@ -554,24 +579,34 @@ asmsub  restore_irq() clobbers(A) {
 		sta  c64.IREQMASK	; enable raster irq
 		lda  #%10000001
 		sta  c64.CIA1ICR	; restore CIA1 irq
-		cli
+		plp
 		rts
 	}}
 }
 
 asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
 	%asm {{
+	    php
         sei
-        sta  _vector
-        sty  _vector+1
+        sta  user_vector
+        sty  user_vector+1
+
+        lda  #%01111111
+        sta  c64.CIA1ICR    ; "switch off" interrupts signals from cia-1
+        sta  c64.CIA2ICR    ; "switch off" interrupts signals from cia-2
+        lda  c64.CIA1ICR    ; ack previous irq
+        lda  c64.CIA2ICR    ; ack previous irq
         lda  cx16.r0
         ldy  cx16.r0+1
-        jsr  _setup_raster_irq
+        jsr  sys.set_rasterline
+        lda  #%00000001
+        sta  c64.IREQMASK   ; enable raster interrupt signals from vic
+
         lda  #<_raster_irq_handler
         sta  cbm.CINV
         lda  #>_raster_irq_handler
         sta  cbm.CINV+1
-        cli
+        plp
         rts
 
 _raster_irq_handler
@@ -593,34 +628,46 @@ _raster_irq_handler
 		pla
 		rti
 _run_custom
-		jmp  (_vector)
+		jmp  (user_vector)
 		.section BSS
-_vector	.word ?
+user_vector	.word ?
 		.send BSS
 
-
-_setup_raster_irq
-		pha
-		lda  #%01111111
-		sta  c64.CIA1ICR    ; "switch off" interrupts signals from cia-1
-		sta  c64.CIA2ICR    ; "switch off" interrupts signals from cia-2
-		and  c64.SCROLY
-		sta  c64.SCROLY     ; clear most significant bit of raster position
-		lda  c64.CIA1ICR    ; ack previous irq
-		lda  c64.CIA2ICR    ; ack previous irq
-		pla
-		sta  c64.RASTER     ; set the raster line number where interrupt should occur
-		cpy  #0
-		beq  +
-		lda  c64.SCROLY
-		ora  #%10000000
-		sta  c64.SCROLY     ; set most significant bit of raster position
-+		lda  #%00000001
-		sta  c64.IREQMASK   ; enable raster interrupt signals from vic
-		rts
+        ; !notreached!
 	}}
 }
 
+    asmsub update_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
+        ; -- just update the IRQ handler and raster line position for the raster IRQ
+        ;    this is much more efficient than calling set_rasterirq() again every time.
+        ;    (but you have to call that one initially at least once to setup the prog8 handler itself)
+        %asm {{
+            php
+            sei
+            sta  sys.set_rasterirq.user_vector
+            sty  sys.set_rasterirq.user_vector+1
+            lda  cx16.r0L
+            ldy  cx16.r0H
+            jsr  sys.set_rasterline
+            plp
+            rts
+        }}
+    }
+
+    asmsub  set_rasterline(uword line @AY) {
+        ; -- only set a new raster line for the raster IRQ
+        %asm {{
+            sta  c64.RASTER     ; set the raster line number where interrupt should occur
+            lda  c64.SCROLY
+            and  #%01111111
+            cpy  #0
+            beq  +
+            ora  #%10000000
+    +       sta  c64.SCROLY     ; clear most significant bit of raster position
+            rts
+        }}
+    }
+
     asmsub  reset_system()  {
         ; Soft-reset the system back to initial power-on Basic prompt.
         %asm {{
@@ -680,6 +727,26 @@ _loop       lda  P8ZP_SCRATCH_W1
         }}
     }
 
+    asmsub waitrasterline(uword line @AY) {
+        ; -- CPU busy wait until the given raster line is reached
+        %asm {{
+            cpy  #0
+            bne  _larger
+-           cmp  c64.RASTER
+            bne  -
+            bit  c64.SCROLY
+            bmi  -
+            rts
+_larger
+            cmp  c64.RASTER
+            bne  _larger
+            bit  c64.SCROLY
+            bpl  _larger
+            rts
+        }}
+    }
+
+
     asmsub internal_stringcopy(str source @R0, str target @AY) clobbers (A,Y) {
         ; Called when the compiler wants to assign a string value to another string.
         %asm {{
@@ -989,6 +1056,48 @@ _no_msb_size
         }}
     }
 
+    inline asmsub pushl(long value @R0R1_32) {
+        %asm {{
+            lda  cx16.r0
+            pha
+            lda  cx16.r0+1
+            pha
+            lda  cx16.r0+2
+            pha
+            lda  cx16.r0+3
+            pha
+        }}
+    }
+
+    inline asmsub popl() -> long @R0R1_32 {
+        %asm {{
+            pla
+            sta  cx16.r0+3
+            pla
+            sta  cx16.r0+2
+            pla
+            sta  cx16.r0+1
+            pla
+            sta  cx16.r0
+        }}
+    }
+
+    sub cpu_is_65816() -> bool {
+        ; Returns true when you have a 65816 cpu, false when it's a 6502.
+        ; The SuperCPU expansion for the C64/C128 contains a 65816.
+        %asm {{
+			php
+			clv
+			.byte $e2, $ea  ; SEP #$ea, should be interpreted as 2 NOPs by 6502. 65c816 will set the Overflow flag.
+			bvc +
+			lda #1
+			plp
+			rts
++			lda #0
+			plp
+			rts
+        }}
+    }
 }
 
 cx16 {
@@ -1165,12 +1274,6 @@ cx16 {
             rts
         }}
     }
-
-    sub cpu_is_65816() -> bool {
-        ; Returns true when you have a 65816 cpu, false when it's a 6502.
-        return false
-    }
-
 }
 
 p8_sys_startup {
@@ -1199,6 +1302,11 @@ asmsub  init_system()  {
         lda  #0
         sta  c64.BGCOL0
         jsr  disable_runstop_and_charsetswitch
+        ; basic is not banked in, adjust MEMTOP
+        ldx  #<$c000
+        ldy  #>$c000
+        clc
+        jsr  cbm.MEMTOP
         cli
         rts
     }}
@@ -1220,6 +1328,10 @@ asmsub  cleanup_at_exit() {
         sta  $ff00          ; default bank 15
         jsr  cbm.CLRCHN		; reset i/o channels
         jsr  enable_runstop_and_charsetswitch
+        ldx  #<$ff00
+        ldy  #>$ff00
+        clc
+        jsr  cbm.MEMTOP     ; adjust MEMTOP to original value again
         lda  _exitcarry
         lsr  a
         lda  _exitcode

compiler/res/prog8lib/c128/textio.p8

@@ -406,6 +406,7 @@ asmsub width() clobbers(X,Y) -> ubyte @A {
     ; -- returns the text screen width (number of columns)
     %asm {{
         jsr  cbm.SCREEN
+        inx
         txa
         rts
     }}
@@ -415,11 +416,22 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     ; -- returns the text screen height (number of rows)
     %asm {{
         jsr  cbm.SCREEN
+        iny
         tya
         rts
     }}
 }
 
+asmsub size() clobbers(A) -> ubyte @X, ubyte @Y {
+    ; -- returns the text screen width in X and height in Y (number of columns and rows)
+    %asm {{
+        jsr  cbm.SCREEN
+        inx
+        iny
+        rts
+    }}
+}
+
 asmsub waitkey() -> ubyte @A {
     %asm {{
 -       jsr cbm.GETIN

compiler/res/prog8lib/c64/floats.asm

@@ -2,7 +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)
+; note: don't add too many constants here because they all end up in the resulting program
+
 
 		.section BSS
 floats_temp_var         .byte  ?,?,?,?,?        ; temporary storage for a float
@@ -126,6 +127,75 @@ cast_FAC1_as_w_into_ay	.proc               ; also used for float 2 b
 		.pend
 
 
+cast_from_long          .proc
+        ; convert long pointed to by AY into a float pointed to by R0
+        ; a bit slow algorithm implemented below:  (msw(l) as word) as float * 65536.0 + (lsw(l) as float)
+        sta  P8ZP_SCRATCH_W1
+        sty  P8ZP_SCRATCH_W1+1
+        ldy  #3
+        lda  (P8ZP_SCRATCH_W1),y
+        sta  P8ZP_SCRATCH_REG
+        dey
+        lda  (P8ZP_SCRATCH_W1),y
+        ldy  P8ZP_SCRATCH_REG
+        jsr  GIVAYFAY
+        lda  #<FL_65536_const
+        ldy  #>FL_65536_const
+        jsr  CONUPK
+        jsr  FMULTT
+        jsr  pushFAC1
+        ldy  #1
+        lda  (P8ZP_SCRATCH_W1),y
+        sta  P8ZP_SCRATCH_REG
+        dey
+        lda  (P8ZP_SCRATCH_W1),y
+        ldy  P8ZP_SCRATCH_REG
+        jsr  GIVUAYFAY
+        jsr  MOVEF
+        clc
+        jsr  popFAC
+        jsr  FADDT
+        ldx  cx16.r0L
+        ldy  cx16.r0H
+        jmp  MOVMF
+
+FL_65536_const  .byte  $91, $00, $00, $00, $00  ; 65536.0
+        ; !notreached!
+        .pend
+
+cast_as_long            .proc
+        ; convert float pointed to by R0 into a long pointed to by AY
+
+        FACHO = FAC_ADDR + 1
+
+        ; save the target variable pointer on the stack
+        pha
+        tya
+        pha
+
+        lda  cx16.r0L
+        ldy  cx16.r0H
+        jsr  MOVFM
+        jsr  QINT
+
+        ; restore the target variable pointer from the stack, and put the result in it
+        pla
+        sta  P8ZP_SCRATCH_PTR+1
+        pla
+        sta  P8ZP_SCRATCH_PTR
+
+        ldx  #3
+        ldy  #0
+-       lda  FACHO,x
+        sta  (P8ZP_SCRATCH_PTR),y
+        iny
+        dex
+        bpl -
+        rts
+        .pend
+
+
+
 copy_float	.proc
 		; -- copies the 5 bytes of the mflt value pointed to by P8ZP_SCRATCH_W1,
 		;    into the 5 bytes pointed to by A/Y.  Clobbers A,Y.

compiler/res/prog8lib/c64/floats.p8

@@ -11,6 +11,7 @@ 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.
 
+const uword FAC_ADDR = $61
 
 ; 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

@@ -2,8 +2,11 @@
 
 
 func_sign_f_into_A	.proc
+		; sign in A, also sets status flags
 		jsr  MOVFM
-		jmp  SIGN
+		jsr  SIGN
+		cmp  #0
+		rts
 		.pend
 
 

compiler/res/prog8lib/c64/graphics.p8

@@ -288,29 +288,16 @@ hline_filled_right   .byte  0, %10000000, %11000000, %11100000, %11110000, %1111
     sub disc(uword xcenter, ubyte ycenter, ubyte radius) {
         ; Warning: NO BOUNDS CHECKS. Make sure circle fits in the screen.
         ; Midpoint algorithm, filled.
+        ; Overdraws horizontal lines unfortunately.
         if radius==0
             return
         ubyte @zp yy = 0
         word decisionOver2 = (1 as word)-radius
-        ubyte last_y3 = ycenter+radius
-        ubyte last_y4 = ycenter-radius
-        ubyte new_y3, new_y4
-
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*2+1)
             horizontal_line(xcenter-radius, ycenter-yy, radius*2+1)
-
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                horizontal_line(xcenter-yy, last_y3, yy*2+1)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                horizontal_line(xcenter-yy, last_y4, yy*2+1)
-                last_y4 = new_y4
-            }
-
+            horizontal_line(xcenter-yy, ycenter+radius, yy*2+1)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*2+1)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -319,10 +306,6 @@ hline_filled_right   .byte  0, %10000000, %11000000, %11100000, %11110000, %1111
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        horizontal_line(xcenter-yy, last_y3, yy*2+1)
-        horizontal_line(xcenter-yy, last_y4, yy*2+1)
     }
 
 

compiler/res/prog8lib/c64/syslib.p8

@@ -132,6 +132,29 @@ asmsub RDTIM16() clobbers(X) -> uword @AY {
     }}
 }
 
+asmsub SETTIML(long jiffies @R0R1_32) {
+    ; -- just like SETTIM, but with a single 32 bit (lower 24 bits used) argument.
+    %asm {{
+        lda  cx16.r0
+        ldx  cx16.r0+1
+        ldy  cx16.r0+2
+        jmp  SETTIM
+    }}
+}
+
+asmsub RDTIML() clobbers(X) -> long @R0R1_32 {
+    ; --  like RDTIM() and returning the timer value as a 32 bit (lower 24 bits used) value.
+    %asm {{
+        jsr  RDTIM
+        sta  cx16.r0
+        stx  cx16.r0+1
+        sty  cx16.r0+2
+        lda  #0
+        sta  cx16.r0+3
+        rts
+    }}
+}
+
 sub CLEARST() {
     ; -- Set the ST status variable back to 0. (there's no direct kernal call for this)
     ;    Note: a drive error state (blinking led) isn't cleared! You can use diskio.status() to clear that.
@@ -313,12 +336,13 @@ asmsub banks(ubyte banks @A) {
     %asm {{
         and  #%00000111
         sta  P8ZP_SCRATCH_REG
+        php
         sei
         lda  $01
         and  #%11111000
         ora  P8ZP_SCRATCH_REG
         sta  $01
-        cli
+        plp
         rts
     }}
 }
@@ -511,6 +535,7 @@ save_SCRATCH_ZPWORD2	.word  ?
 
 asmsub  set_irq(uword handler @AY) clobbers(A)  {
 	%asm {{
+	    php
 	    sei
         sta  _vector
         sty  _vector+1
@@ -518,7 +543,7 @@ asmsub  set_irq(uword handler @AY) clobbers(A)  {
 		sta  cbm.CINV
 		lda  #>_irq_handler
 		sta  cbm.CINV+1
-		cli
+		plp
 		rts
 _irq_handler
         jsr  sys.save_prog8_internals
@@ -551,6 +576,7 @@ _vector	.word ?
 
 asmsub  restore_irq() clobbers(A) {
 	%asm {{
+	    php
 		sei
 		lda  #<cbm.IRQDFRT
 		sta  cbm.CINV
@@ -560,24 +586,34 @@ asmsub  restore_irq() clobbers(A) {
 		sta  c64.IREQMASK	; disable raster irq
 		lda  #%10000001
 		sta  c64.CIA1ICR	; restore CIA1 irq
-		cli
+		plp
 		rts
 	}}
 }
 
 asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
 	%asm {{
+	    php
 	    sei
-        sta  _vector
-        sty  _vector+1
+        sta  user_vector
+        sty  user_vector+1
+
+		lda  #%01111111
+		sta  c64.CIA1ICR    ; "switch off" interrupts signals from cia-1
+		sta  c64.CIA2ICR    ; "switch off" interrupts signals from cia-2
+		lda  c64.CIA1ICR    ; ack previous irq
+		lda  c64.CIA2ICR    ; ack previous irq
         lda  cx16.r0
         ldy  cx16.r0+1
-        jsr  _setup_raster_irq
+		jsr  sys.set_rasterline
+ 		lda  #%00000001
+		sta  c64.IREQMASK   ; enable raster interrupt signals from vic
+
         lda  #<_raster_irq_handler
         sta  cbm.CINV
         lda  #>_raster_irq_handler
         sta  cbm.CINV+1
-        cli
+        plp
         rts
 
 _raster_irq_handler
@@ -600,33 +636,46 @@ _raster_irq_handler
 		rti
 
 _run_custom
-		jmp  (_vector)
+		jmp  (user_vector)
 		.section BSS
-_vector	.word ?
+user_vector	.word ?
 		.send BSS
 
-_setup_raster_irq
-		pha
-		lda  #%01111111
-		sta  c64.CIA1ICR    ; "switch off" interrupts signals from cia-1
-		sta  c64.CIA2ICR    ; "switch off" interrupts signals from cia-2
-		and  c64.SCROLY
-		sta  c64.SCROLY     ; clear most significant bit of raster position
-		lda  c64.CIA1ICR    ; ack previous irq
-		lda  c64.CIA2ICR    ; ack previous irq
-		pla
-		sta  c64.RASTER     ; set the raster line number where interrupt should occur
-		cpy  #0
-		beq  +
-		lda  c64.SCROLY
-		ora  #%10000000
-		sta  c64.SCROLY     ; set most significant bit of raster position
-+		lda  #%00000001
-		sta  c64.IREQMASK   ; enable raster interrupt signals from vic
-		rts
+		; !notreached!
 	}}
 }
 
+    asmsub update_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
+        ; -- just update the IRQ handler and raster line position for the raster IRQ
+        ;    this is much more efficient than calling set_rasterirq() again every time.
+        ;    (but you have to call that one initially at least once to setup the prog8 handler itself)
+        %asm {{
+            php
+            sei
+            sta  sys.set_rasterirq.user_vector
+            sty  sys.set_rasterirq.user_vector+1
+            lda  cx16.r0L
+            ldy  cx16.r0H
+            jsr  sys.set_rasterline
+            plp
+            rts
+        }}
+    }
+
+asmsub  set_rasterline(uword line @AY) {
+    ; -- only set a new raster line for the raster IRQ
+    %asm {{
+        sta  c64.RASTER     ; set the raster line number where interrupt should occur
+        lda  c64.SCROLY
+        and  #%01111111
+        cpy  #0
+        beq  +
+        ora  #%10000000
++       sta  c64.SCROLY     ; clear most significant bit of raster position
+        rts
+    }}
+}
+
 
     asmsub reset_system()  {
         ; Soft-reset the system back to initial power-on Basic prompt.
@@ -687,6 +736,26 @@ _loop       lda  P8ZP_SCRATCH_W1
         }}
     }
 
+    asmsub waitrasterline(uword line @AY) {
+        ; -- CPU busy wait until the given raster line is reached
+        %asm {{
+            cpy  #0
+            bne  _larger
+-           cmp  c64.RASTER
+            bne  -
+            bit  c64.SCROLY
+            bmi  -
+            rts
+_larger
+            cmp  c64.RASTER
+            bne  _larger
+            bit  c64.SCROLY
+            bpl  _larger
+            rts
+        }}
+    }
+
+
     asmsub internal_stringcopy(str source @R0, str target @AY) clobbers (A,Y) {
         ; Called when the compiler wants to assign a string value to another string.
         %asm {{
@@ -995,6 +1064,49 @@ _no_msb_size
             pla
         }}
     }
+
+    inline asmsub pushl(long value @R0R1_32) {
+        %asm {{
+            lda  cx16.r0
+            pha
+            lda  cx16.r0+1
+            pha
+            lda  cx16.r0+2
+            pha
+            lda  cx16.r0+3
+            pha
+        }}
+    }
+
+    inline asmsub popl() -> long @R0R1_32 {
+        %asm {{
+            pla
+            sta  cx16.r0+3
+            pla
+            sta  cx16.r0+2
+            pla
+            sta  cx16.r0+1
+            pla
+            sta  cx16.r0
+        }}
+    }
+
+    sub cpu_is_65816() -> bool {
+        ; Returns true when you have a 65816 cpu, false when it's a 6502.
+        ; The SuperCPU expansion for the C64/C128 contains a 65816.
+        %asm {{
+			php
+			clv
+			.byte $e2, $ea  ; SEP #$ea, should be interpreted as 2 NOPs by 6502. 65c816 will set the Overflow flag.
+			bvc +
+			lda #1
+			plp
+			rts
++			lda #0
+			plp
+			rts
+        }}
+    }
 }
 
 cx16 {
@@ -1172,12 +1284,6 @@ cx16 {
             rts
         }}
     }
-
-    sub cpu_is_65816() -> bool {
-        ; Returns true when you have a 65816 cpu, false when it's a 6502.
-        return false
-    }
-
 }
 
 p8_sys_startup {
@@ -1207,6 +1313,8 @@ asmsub  init_system()  {
         lda  #0
         sta  c64.BGCOL0
         jsr  disable_runstop_and_charsetswitch
+        lda  #$80
+        sta  650    ; enable key repeat for normal keys too
         lda  #PROG8_C64_BANK_CONFIG     ; apply bank config
         sta  $01
         and  #1
@@ -1243,6 +1351,8 @@ asmsub  cleanup_at_exit() {
         jsr  cbm.MEMTOP     ; adjust MEMTOP down again
         jsr  cbm.CLRCHN		; reset i/o channels
         jsr  enable_runstop_and_charsetswitch
+        lda  #0
+        sta  650            ; disable keyrepeats
         lda  _exitcarry
         lsr  a
         lda  _exitcode

compiler/res/prog8lib/c64/textio.p8

@@ -424,6 +424,14 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub size() clobbers(A) -> ubyte @X, ubyte @Y {
+    ; -- returns the text screen width in X and height in Y (number of columns and rows)
+    %asm {{
+        jmp  cbm.SCREEN
+    }}
+}
+
+
 asmsub waitkey() -> ubyte @A {
     %asm {{
 -       jsr cbm.GETIN

compiler/res/prog8lib/conv.p8

@@ -6,8 +6,6 @@ conv {
 
 ; ----- number conversions to decimal strings ----
 
-    ubyte[16]  @shared string_out       ; result buffer for the string conversion routines (note: uses uninitialized ARRAY instead of STR, to force it to be allocated in BSS area so it's ROM-compatible)
-
     asmsub str_ub0(ubyte value @A) clobbers(X) -> str @AY {
         ; ---- convert the ubyte in A in decimal string form, with left padding 0s (3 positions total)
         %asm {{
@@ -155,6 +153,30 @@ asmsub  str_uwhex  (uword value @ AY) -> str @AY  {
 	}}
 }
 
+sub str_ulhex(long value) -> str {
+    %asm {{
+        lda  value+3
+        jsr  internal_ubyte2hex
+        sta  string_out
+        sty  string_out+1
+        lda  value+2
+        jsr  internal_ubyte2hex
+        sta  string_out+2
+        sty  string_out+3
+        lda  value+1
+        jsr  internal_ubyte2hex
+        sta  string_out+4
+        sty  string_out+5
+        lda  value
+        jsr  internal_ubyte2hex
+        sta  string_out+6
+        sty  string_out+7
+        lda  #0
+        sta  string_out+8
+    }}
+    return &string_out
+}
+
 asmsub  str_uw0  (uword value @ AY) clobbers(X) -> str @AY  {
 	; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
 	%asm {{
@@ -228,15 +250,91 @@ asmsub  str_w  (word value @ AY) clobbers(X) -> str @AY  {
 	}}
 }
 
+sub  str_l  (long value) -> str  {
+	; ---- convert the long value into decimal string form, without left padding 0s
+        ; source: https://codebase64.net/doku.php?id=base:32_bit_hexadecimal_to_decimal_conversion
+        ; NOTE: this is a pretty slow conversion, don't use for time critical displays!
+        bool negative
+        string_out[0] = '0'
+        if value<0 {
+            negative = true
+            value = -value
+        }
+
+        %asm {{
+            jsr  hex2dec
+            ldx  #0
+-           lda  string_out+1,x
+            cmp  #'0'
+            bne  +
+            inx
+            bne  -
++           cpx  #10   ; just all 0? keep one
+            bne  +
+            dex
+
++           ; x points at the rightmost leading 0
+            lda  negative
+            beq  +
+            lda  #'-'       ; sign in front
+            sta  string_out,x
+            dex
+            bmi  _done
++           ; x still points at rightmost leading 0
+            ; remove leading 0's (there is at least one if we're here)
+            ldy  #0
+-           lda  string_out+1,x
+            sta  string_out,y
+            iny
+            inx
+            cpx  #10
+            bne  -
+            lda  #0
+            sta  string_out,y
+_done
+
+            lda  #<string_out
+            ldy  #>string_out
+            rts
+
+hex2dec:
+        ; converts 10 digits (32 bit values have max. 10 decimal digits)
+        ldx  #9+1
+l3      jsr  div10
+        ora  #$30
+        sta  string_out,x
+        dex
+        bne  l3
+        rts
+
+        ; divides a 32 bit value by 10
+        ; remainder is returned in akku
+div10
+        ldy  #32         ; 32 bits
+        lda  #0
+        clc
+l4      rol
+        cmp  #10
+        bcc  skip
+        sbc  #10
+skip    rol  value
+        rol  value+1
+        rol  value+2
+        rol  value+3
+        dey
+        bpl  l4
+        rts
+        }}
+    }
 
 ; ---- string conversion to numbers -----
 
-asmsub  any2uword(str string @AY) clobbers(Y) -> ubyte @A {
+asmsub  any2uword(str string @AY) -> uword @AY, ubyte @X {
 	; -- 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.
+	;    returns the parsed number word in AY, and the number of processed characters (including the prefix symbol) in X.
+	;    if the string was invalid, 0 will be returned as count in X (and the word value in AY will be undefined).
 	%asm {{
 	pha
 	sta  P8ZP_SCRATCH_W1
@@ -257,13 +355,7 @@ _hex	pla
 _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
+        ldx  cx16.r15
         rts
 	}}
 }
@@ -480,6 +572,85 @@ _try_iso
 	}}
 }
 
+asmsub hex2long(str string @AY) -> long @R0R1_32 {
+	; -- hexadecimal string (with or without '$') to long.
+	;    string may be in petscii or c64-screencode encoding.
+	;    stops parsing at the first character that's not a hex digit (except leading $)
+	;    result in R0:R1,  number of characters processed also remains in cx16.r15L if you want to use it!! (0 = error)
+    %asm {{
+	sta  P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_W2+1
+	ldy  #0
+	sty  cx16.r0
+	sty  cx16.r0+1
+	sty  cx16.r0+2
+	sty  cx16.r0+3
+	sty  cx16.r15H
+	lda  (P8ZP_SCRATCH_W2),y
+	beq  _stop
+	cmp  #'$'
+	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
+	and  #127
+	cmp  #97
+	bcs  _try_iso            ; maybe letter is iso:'a'-iso:'f' (97-102)
+	cmp  #'g'
+	bcs  _stop
+	cmp  #'a'
+	bcs  _add_letter
+	cmp  #'0'
+	bcc  _stop
+	cmp  #'9'+1
+	bcs  _stop
+_calc
+	asl  cx16.r0
+	rol  cx16.r0+1
+	rol  cx16.r0+2
+	rol  cx16.r0+3
+	asl  cx16.r0
+	rol  cx16.r0+1
+	rol  cx16.r0+2
+	rol  cx16.r0+3
+	asl  cx16.r0
+	rol  cx16.r0+1
+	rol  cx16.r0+2
+	rol  cx16.r0+3
+	asl  cx16.r0
+	rol  cx16.r0+1
+	rol  cx16.r0+2
+	rol  cx16.r0+3
+	and  #$0f
+	clc
+	adc  P8ZP_SCRATCH_B1
+	ora  cx16.r0
+	sta  cx16.r0
+	iny
+	bne  _loop
+_stop
+	sty  cx16.r15L
+	rts
+_add_letter
+	pha
+	lda  #9
+	sta  P8ZP_SCRATCH_B1
+	pla
+	jmp  _calc
+_try_iso
+        cmp  #103
+        bcs  _stop
+        and  #63
+        bne  _add_letter
+		; !notreached!
+    }}
+}
+
 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 %)
@@ -767,4 +938,6 @@ output		.fill 5      ; 0-terminated output buffer (to make printing easier)
 	}}
 }
 
+    ubyte[16]  @shared string_out       ; internal result buffer for the string conversion routines (note: uses uninitialized ARRAY instead of STR, to force it to be allocated in BSS area so it's ROM-compatible)
+
 }

compiler/res/prog8lib/cx16/buffers.p8

@@ -68,7 +68,65 @@ smallringbuffer {
     }
 }
 
-; note: for a "small stack"  (256 bytes size) just use the CPU stack via sys.push[w] / sys.pop[w].
+
+smallstack {
+    ; -- A small stack (LIFO) that uses just 256 bytes and is independent of the CPU stack. Stack is growing downward from the top of the buffer.
+    ;    You can store and retrieve bytes and words. There are no guards against stack over/underflow.
+    ; note: for a "small stack"  (256 bytes size) you might also perhaps just use the CPU stack via sys.push[w] / sys.pop[w].
+
+    ubyte[256] buffer
+    ubyte sp = 255
+
+    sub init() {
+        sp = 255
+    }
+
+    sub size() -> ubyte {
+        return 255-sp
+    }
+
+    sub free() -> ubyte {
+        return sp
+    }
+
+    sub isfull() -> bool {
+        return sp==0
+    }
+
+    sub isempty() -> bool {
+        return sp==255
+    }
+
+    sub push(ubyte value) {
+        ; -- put a byte on the stack
+        buffer[sp] = value
+        sp--
+    }
+
+    sub pushw(uword value) {
+        ; -- put a word on the stack (lsb first then msb)
+        buffer[sp] = lsb(value)
+        sp--
+        buffer[sp] = msb(value)
+        sp--
+    }
+
+    sub pop() -> ubyte {
+        ; -- pops a byte off the stack
+        sp++
+        return buffer[sp]
+    }
+
+    sub popw() -> uword {
+        ; -- pops a word off the stack.
+        sp++
+        cx16.r0H = buffer[sp]
+        sp++
+        cx16.r0L = buffer[sp]
+        return cx16.r0
+    }
+}
+
 
 stack {
     ; -- A stack (LIFO) that uses a block of 8 KB of memory. Growing downward from the top of the buffer.
@@ -78,7 +136,7 @@ stack {
     ubyte bank
 
     sub init(ubyte rambank) {
-        ; -- initialize the stack, must be called before use. Supply the HiRAM bank to use as buffer space.
+        ; -- initialize the stack, must be called before use. Supply the HIRAM bank to use as buffer space.
         sp = $bfff
         bank = rambank
     }
@@ -157,7 +215,7 @@ ringbuffer {
     ubyte bank = 255        ; set via init()
 
     sub init(ubyte rambank) {
-        ; -- initialize the ringbuffer, must be called before use. Supply the HiRAM bank to use as buffer space.
+        ; -- initialize the ringbuffer, must be called before use. Supply the HIRAM bank to use as buffer space.
         head = $a000
         tail = $bfff
         fill = 0

compiler/res/prog8lib/cx16/compression.p8

@@ -1,4 +1,4 @@
-; **experimental** data compression/decompression routines, API subject to change!!
+; data compression/decompression routines
 
 %import shared_compression
 

compiler/res/prog8lib/cx16/diskio.p8

@@ -566,7 +566,7 @@ return_status:
 
     sub internal_f_open_w(str filename, bool open_for_seeks) -> bool {
         f_close_w()
-        list_filename = filename
+        void strings.copy(filename, list_filename)
         str modifier = ",s,?"
         modifier[3] = 'w'
         if open_for_seeks
@@ -624,7 +624,7 @@ done:
         return list_filename
 
 io_error:
-        list_filename = "io error"
+        void strings.copy("io error", list_filename)
         goto done
     }
 
@@ -818,8 +818,13 @@ io_error:
 
     ; For use directly after a load or load_raw call (don't mess with the ram bank yet):
     ; Calculates the number of bytes loaded (files > 64Kb are truncated to 16 bits)
-    sub load_size(ubyte startbank, uword startaddress, uword endaddress) -> uword {
-        return $2000 * (cx16.getrambank() - startbank) + endaddress - startaddress
+    sub load_size(ubyte startbank, uword startaddress, uword endaddress) -> long {
+        &long banksize = &cx16.r12
+        banksize = cx16.getrambank() - startbank
+        banksize <<= 13     ; * 8192
+        banksize += endaddress
+        banksize -= startaddress
+        return banksize
     }
 
     asmsub vload(str name @R0, ubyte bank @A, uword startaddress @R1) clobbers(X, Y) -> bool @A {
@@ -994,15 +999,12 @@ io_error:
         return false
     }
 
-    sub f_seek(uword pos_hiword, uword pos_loword) {
+    sub f_seek(long position) {
         ; -- seek in the reading file opened with f_open, to the given 32-bits position
         ;    Note: this will not work if you have already read the last byte of the file! Then you must close and reopen the file first.
         ubyte[6] command = ['p',0,0,0,0,0]
         command[1] = READ_IO_CHANNEL       ; f_open uses this secondary address
-        command[2] = lsb(pos_loword)
-        command[3] = msb(pos_loword)
-        command[4] = lsb(pos_hiword)
-        command[5] = msb(pos_hiword)
+        pokel(&command+2, position)
         cbm.SETNAM(sizeof(command), &command)
         cbm.SETLFS(15, drivenumber, 15)
         void cbm.OPEN()
@@ -1011,13 +1013,10 @@ io_error:
     }
 
 
-    sub f_seek_w(uword pos_hiword, uword pos_loword) {
+    sub f_seek_w(long position) {
         ; -- seek in the output file opened with f_open_w_seek, to the given 32-bits position
         diskio.f_seek.command[1] = WRITE_IO_CHANNEL       ; f_open_w uses this secondary address
-        diskio.f_seek.command[2] = lsb(pos_loword)
-        diskio.f_seek.command[3] = msb(pos_loword)
-        diskio.f_seek.command[4] = lsb(pos_hiword)
-        diskio.f_seek.command[5] = msb(pos_hiword)
+        pokel(&diskio.f_seek.command+2, position)
         cbm.SETNAM(sizeof(diskio.f_seek.command), &diskio.f_seek.command)
         cbm.SETLFS(15, drivenumber, 15)
         void cbm.OPEN()
@@ -1025,54 +1024,63 @@ io_error:
         reset_write_channel()    ; back to the write io channel
     }
 
-    asmsub f_tell() -> uword @R0, uword @R1, uword @R2, uword @R3 {
-        ; -- Returns the current read position of the opened read file,
-        ;    in R0 and R1 (low + high words) and the file size in R2 and R3 (low + high words).
-        ;    Returns 0 as size if the command is not supported by the DOS implementation/version.
-        %asm {{
-            jmp  internal_f_tell
-        }}
-    }
-
-    sub internal_f_tell() {
-        ; gets the (32 bits) position + file size of the opened read file channel
-        ubyte[2] command = ['t',0]
-        command[1] = READ_IO_CHANNEL       ; f_open uses this secondary address
-        cbm.SETNAM(sizeof(command), &command)
-        cbm.SETLFS(15, drivenumber, 15)
-        void cbm.OPEN()
-        void cbm.CHKIN(15)        ; use #15 as input channel
+    sub f_tell() -> long, long {
+        ; -- Returns the current read position of the opened read file, and the file size.
+        ;    Returns 0,0 if the command is not supported by the DOS implementation/version.
+        ubyte[3] command = ['?','?',0]
         bool success=false
-        ; valid response starts with "07," followed by hex notations of the position and filesize
-        if cbm.CHRIN()=='0' and cbm.CHRIN()=='7' and cbm.CHRIN()==',' {
-            cx16.r1 = read4hex()
-            cx16.r0 = read4hex()        ; position in R1:R0
-            void cbm.CHRIN()            ; separator space
-            cx16.r3 = read4hex()
-            cx16.r2 = read4hex()        ; filesize in R3:R2
-            success = true
+        long filesize, filepos
+        ubyte commandoffset
+
+        command[1] = 't'
+        commandoffset = 1
+        f_tell_internal()
+        return filepos, filesize
+
+        sub f_tell_internal() {
+            ; this code is used for both the T (tell) and FL (fatlba) commands
+            command[2] = READ_IO_CHANNEL       ; f_open uses this secondary address
+            cbm.SETNAM(sizeof(command)-commandoffset, &command+commandoffset)
+            cbm.SETLFS(15, drivenumber, 15)
+            void cbm.OPEN()
+            void cbm.CHKIN(15)        ; use #15 as input channel
+
+            ; valid response starts with "07," followed by hex notations of the position and filesize
+            if cbm.CHRIN()=='0' and cbm.CHRIN()=='7' and cbm.CHRIN()==',' {
+                filepos = read8hex()
+                void cbm.CHRIN()            ; separator space
+                filesize = read8hex()
+                success = true
+            } else {
+                filesize = filepos = 0
+            }
+
+            while cbm.READST()==0 {
+                cx16.r5L = cbm.CHRIN()
+                if cx16.r5L=='\r' or cx16.r5L=='\n'
+                    break
+            }
+
+            cbm.CLOSE(15)
+            reset_read_channel()       ; back to the read io channel
         }
-
-        while cbm.READST()==0 {
-            cx16.r5L = cbm.CHRIN()
-            if cx16.r5L=='\r' or cx16.r5L=='\n'
-                break
-        }
-
-        cbm.CLOSE(15)
-        reset_read_channel()       ; back to the read io channel
-        if success
-            return
-
-        cx16.r0 = cx16.r1 = cx16.r2 = cx16.r3 = 0
     }
 
-    sub read4hex() -> uword {
-        str hex = "0000"
-        hex[0] = cbm.CHRIN()
-        hex[1] = cbm.CHRIN()
-        hex[2] = cbm.CHRIN()
-        hex[3] = cbm.CHRIN()
-        return conv.hex2uword(hex)
+    sub f_fatlba() -> long, long {
+        ; -- Return the current LBA, cluster number  (sector index and shift value are ignored for now)
+        diskio.f_tell.command[0] = 'f'
+        diskio.f_tell.command[1] = 'l'
+        diskio.f_tell.commandoffset = 0
+        diskio.f_tell.f_tell_internal()
+        return diskio.f_tell.filepos, diskio.f_tell.filesize
+    }
+
+
+    sub read8hex() -> long {
+        str hex = "00000000"
+        for cx16.r0L in 0 to 7 {
+            hex[cx16.r0L] = cbm.CHRIN()
+        }
+        return conv.hex2long(hex)
     }
 }

compiler/res/prog8lib/cx16/floats.p8

@@ -14,6 +14,8 @@ floats {
 ; note: fac1/2 might get clobbered even if not mentioned in the function's name.
 ; note: for subtraction and division, the left operand is in fac2, the right operand in fac1.
 
+const uword FAC_ADDR = $c3
+
 extsub $fe00 = AYINT() clobbers(A,X,Y)          ; fac1-> signed word in 'facmo' and 'faclo', MSB FIRST. DON'T USE THIS, USE WRAPPER 'AYINT2' INSTEAD.  (might throw ILLEGAL QUANTITY)
 
 ; GIVAYF: signed word in Y/A (note different lsb/msb order) -> float in fac1

compiler/res/prog8lib/cx16/gfx_hires.p8

@@ -433,25 +433,12 @@ gfx_hires {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
 
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
-
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
-                last_y4 = new_y4
-            }
-
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -460,11 +447,6 @@ gfx_hires {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-
-        ; draw the final two spans
-        yy--
-        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
@@ -474,23 +456,12 @@ gfx_hires {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
 
         while radius>=yy {
             safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
             safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
-                last_y4 = new_y4
-            }
+            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
+            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -499,10 +470,6 @@ gfx_hires {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-        safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
     }
 
     sub plot(uword @zp xx, uword @zp yy, ubyte @zp color) {

compiler/res/prog8lib/cx16/gfx_lores.p8

@@ -505,25 +505,11 @@ gfx_lores {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        ubyte last_y3 = ycenter+radius
-        ubyte last_y4 = ycenter-radius
-        ubyte new_y3, new_y4
-
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, color)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, color)
-
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
-                last_y4 = new_y4
-            }
-
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, color)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, color)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -532,11 +518,6 @@ gfx_lores {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-
-        ; draw the final two spans
-        yy--
-        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, color)
-        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, color)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, ubyte color) {
@@ -546,29 +527,22 @@ gfx_lores {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
 
         while radius>=yy {
             uword liney = ycenter+yy
             if msb(liney)==0
-                safe_horizontal_line(xcenter-radius, lsb(ycenter+yy), radius*$0002+1, color)
+                safe_horizontal_line(xcenter-radius, lsb(liney), radius*$0002+1, color)
             liney = ycenter-yy
             if msb(liney)==0
-                safe_horizontal_line(xcenter-radius, lsb(ycenter-yy), radius*$0002+1, color)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                if msb(last_y3)==0
-                    safe_horizontal_line(xcenter-yy, lsb(last_y3), yy*$0002+1, color)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                if msb(last_y4)==0
-                    safe_horizontal_line(xcenter-yy, lsb(last_y4), yy*$0002+1, color)
-                last_y4 = new_y4
-            }
+                safe_horizontal_line(xcenter-radius, lsb(liney), radius*$0002+1, color)
+
+            liney = ycenter+radius
+            if msb(liney)==0
+                safe_horizontal_line(xcenter-yy, lsb(liney), yy*$0002+1, color)
+            liney = ycenter-radius
+            if msb(liney)==0
+                safe_horizontal_line(xcenter-yy, lsb(liney), yy*$0002+1, color)
+
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -577,12 +551,6 @@ gfx_lores {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        if msb(last_y3)==0
-            safe_horizontal_line(xcenter-yy, lsb(last_y3), yy*$0002+1, color)
-        if msb(last_y4)==0
-            safe_horizontal_line(xcenter-yy, lsb(last_y4), yy*$0002+1, color)
     }
 
     asmsub plot(uword x @AX, ubyte y @Y, ubyte color @R0) {

compiler/res/prog8lib/cx16/monogfx.p8

@@ -595,23 +595,12 @@ drawmode:               ora  cx16.r15L
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
 
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
-                last_y4 = new_y4
-            }
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -620,10 +609,6 @@ drawmode:               ora  cx16.r15L
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
@@ -634,23 +619,12 @@ drawmode:               ora  cx16.r15L
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
 
         while radius>=yy {
             safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
             safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
-                last_y4 = new_y4
-            }
+            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -659,10 +633,6 @@ drawmode:               ora  cx16.r15L
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-        safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub plot(uword @zp xx, uword @zp yy, bool @zp draw) {
@@ -1038,7 +1008,7 @@ _doplot         beq  +
 
     ; y*40 lookup table. Pretty compact because it all fits in a word and we only need 240 y positions.
     ; a y*80 lookup table would be very large (lo,mid,hi for 480 values...)
-    uword[240] @split @shared times40 = [
+    uword[240] @shared times40 = [
         0, 40, 80, 120, 160, 200, 240, 280, 320, 360, 400, 440, 480, 520, 560, 600,
         640, 680, 720, 760, 800, 840, 880, 920, 960, 1000, 1040, 1080, 1120, 1160,
         1200, 1240, 1280, 1320, 1360, 1400, 1440, 1480, 1520, 1560, 1600, 1640, 1680,

compiler/res/prog8lib/cx16/palette.p8

@@ -283,7 +283,7 @@ palette {
     ; NOTE: this routine requires rom version 49+
     alias set_default = cx16.set_default_palette
 
-    ; get the bank and address of the word-array containing the 256 default palette colors
+    ; get the bank and address (in A, and XY) of the word-array containing the 256 default palette colors
     ; NOTE: this routine requires rom version 49+
     alias get_default = cx16.get_default_palette
 }

compiler/res/prog8lib/cx16/sprites.p8

@@ -1,10 +1,13 @@
 ; Simple routines to control sprites.
 
 ; They're not written for high performance, but for simplicity.
-; That's why they control 1 sprite at a time. \
+; That's why they control 1 sprite at a time.
 ; The exceptions are pos_batch() and pos_batch_split(); these are quite efficient
 ; to update sprite positions of multiple sprites in one call.
 
+; HIGH PERFORMANCE sprite handling would probably have a copy of the sprite registers for each sprite instead,
+; and a unrolled loop that copies those into the VERA registers when needed.
+
 ; note: sprites z-order will be in front of all layers.
 ; note: collision mask is not supported here yet.
 ; note: "palette offset" is counted as 0-15  (vera multiplies the offset by 16 to get at the actual color index)

compiler/res/prog8lib/cx16/syslib.p8

@@ -127,6 +127,28 @@ asmsub RDTIM16() clobbers(X) -> uword @AY {
     }}
 }
 
+asmsub SETTIML(long jiffies @R0R1_32) {
+    ; -- just like SETTIM, but with a single 32 bit (lower 24 bits used) argument.
+    %asm {{
+        lda  cx16.r0
+        ldx  cx16.r0+1
+        ldy  cx16.r0+2
+        jmp  SETTIM
+    }}
+}
+
+asmsub RDTIML() clobbers(X) -> long @R0R1_32 {
+    ; --  like RDTIM_safe() and returning the timer value as a 32 bit (lower 24 bits used) value.
+    %asm {{
+        jsr  RDTIM_safe
+        sta  cx16.r0
+        stx  cx16.r0+1
+        sty  cx16.r0+2
+        stz  cx16.r0+3
+        rts
+    }}
+}
+
 sub CLEARST() {
     ; -- Set the ST status variable back to 0. (there's no direct kernal call for this)
     ;    Note: a drive error state (blinking led) isn't cleared! You can use diskio.status() to clear that.
@@ -441,7 +463,7 @@ extsub $ff62 = screen_set_charset(ubyte charset @A, uword charsetptr @XY)  clobb
 extsub $ff6e = JSRFAR()  ; following word = address to call, byte after that=rom/ram bank it is in
 extsub $ff74 = fetch(ubyte zp_startaddr @A, ubyte bank @X, ubyte index @Y)  clobbers(X) -> ubyte @A
 extsub $ff77 = stash(ubyte data @A, ubyte bank @X, ubyte index @Y)  clobbers(X)     ;  note: The the zero page address containing the base address is passed in stavec ($03B2)
-extsub $ff7d = PRIMM()
+extsub $ff7d = PRIMM()  ; print immediate string  (the 0-terminated string immediately following the JSR PRIMM instruction)
 
 ; high level graphics & fonts
 extsub $ff20 = GRAPH_init(uword vectors @R0)  clobbers(A,X,Y)
@@ -598,6 +620,10 @@ const ubyte  EXTAPI_scnsiz = $0D
 const ubyte  EXTAPI_kbd_leds = $0E
 const ubyte  EXTAPI_memory_decompress_from_func = $0F
 const ubyte  EXTAPI_default_palette = $10
+const ubyte  EXTAPI_has_machine_property = $11
+const ubyte  EXTAPI_kbdbuf_get = $12
+const ubyte  EXTAPI_kbdbuf_clear = $13
+const ubyte  EXTAPI_blink_enable = $14
 
 ; extapi16 call numbers
 const ubyte  EXTAPI16_test = $00
@@ -654,6 +680,30 @@ sub mouse_present() -> bool {
     return cx16.r0L != $fc       ; $fc = BAT_FAIL
 }
 
+asmsub get_charset() -> ubyte @A {
+    ;Get charset mode.  result: 0=unknown, 1=ISO, 2=PETSCII upper case/gfx, 3=PETSCII lowercase.
+    %asm {{
+
+KERNAL_MODE = $0372         ; internal kernal variable, risky to read it, but it's ben stable for many releases.
+        lda  KERNAL_MODE
+        beq  _end
+
+        bit  #$40                ;ISO mode flag
+        beq  +                   ;usually KERNAL_MODE 1 or 6 (| $40)
+        lda  #1
+        bra  _end
+
++       bit  #1                  ;PETSCII upper case/graphics
+        bne  +                   ;usually KERNAL_MODE 2 or 4
+        lda  #2
+        bra  _end
+
++       lda  #3                   ;PETSCII upper/lower case
+_end:
+        rts
+    }}
+}
+
 ; shims for the kernal routines called via the extapi call:
 
 asmsub mouse_set_pos(uword xpos @R0, uword ypos @R1) clobbers(X) {
@@ -736,31 +786,42 @@ asmsub set_default_palette() {
     }}
 }
 
-asmsub get_charset() -> ubyte @A {
-    ;Get charset mode.  result: 0=unknown, 1=ISO, 2=PETSCII upper case/gfx, 3=PETSCII lowercase.
+asmsub has_machine_property(ubyte property @X) clobbers(A,X) -> bool @Pc {
+    ; requires rom v49+
     %asm {{
-
-KERNAL_MODE = $0372         ; internal kernal variable, risky to read it, but it's ben stable for many releases.
-        lda  KERNAL_MODE
-        beq  _end
-
-        bit  #$40                ;ISO mode flag
-        beq  +                   ;usually KERNAL_MODE 1 or 6 (| $40)
-        lda  #1
-        bra  _end
-
-+       bit  #1                  ;PETSCII upper case/graphics
-        bne  +                   ;usually KERNAL_MODE 2 or 4
-        lda  #2
-        bra  _end
-
-+       lda  #3                   ;PETSCII upper/lower case
-_end:
-        rts
+        lda  #EXTAPI_has_machine_property
+        jmp  cx16.extapi
+    }}
+}
+
+asmsub kbdbuf_get() clobbers(X,Y) -> ubyte @A {
+    ; requires rom v49+
+    %asm {{
+        lda  #EXTAPI_kbdbuf_get
+        jmp  cx16.extapi
+    }}
+}
+
+asmsub kbdbuf_clear() clobbers(A) {
+    ; requires rom v49+
+    %asm {{
+        lda  #EXTAPI_kbdbuf_clear
+        jmp  cx16.extapi
+    }}
+}
+
+asmsub blink_enable(bool enable @X) clobbers(A,X) {
+    ; requires rom v49+
+    %asm {{
+        txa
+        and  #1
+        eor  #1
+        tax
+        lda  #EXTAPI_blink_enable
+        jmp  cx16.extapi
     }}
 }
 
-; TODO : implement shims for the remaining extapi calls.
 
 
 ; ---- end of kernal routines ----
@@ -1454,22 +1515,6 @@ sub search_x16edit() -> ubyte {
     return 255
 }
 
-    asmsub cpu_is_65816() -> bool @A {
-        ; -- Returns true when you have a 65816 cpu, false when it's a 6502.
-        %asm {{
-			php
-			clv
-			.byte $e2, $ea  ; SEP #$ea, should be interpreted as 2 NOPs by 6502. 65c816 will set the Overflow flag.
-			bvc +
-			lda #1
-			plp
-			rts
-+			lda #0
-			plp
-			rts
-        }}
-    }
-
     sub set_program_args(str args_ptr, ubyte args_size) {
         ; -- Set the inter-program arguments.
         ; standardized way to pass arguments between programs is in ram bank 0, address $bf00-$bfff.
@@ -1576,6 +1621,7 @@ sys {
 asmsub  set_irq(uword handler @AY) clobbers(A)  {
     ; Sets the handler for the VSYNC interrupt, and enable that interrupt.
 	%asm {{
+	    php
         sei
         sta  _vector
         sty  _vector+1
@@ -1585,7 +1631,7 @@ asmsub  set_irq(uword handler @AY) clobbers(A)  {
         sta  cbm.CINV+1
         lda  #1
         tsb  cx16.VERA_IEN      ; enable the vsync irq
-        cli
+        plp
         rts
 
 _irq_handler
@@ -1615,6 +1661,7 @@ _vector	.word ?
 
 asmsub  restore_irq() clobbers(A) {
 	%asm {{
+	    php
 	    sei
 	    lda  _orig_irqvec
 	    sta  cbm.CINV
@@ -1624,7 +1671,7 @@ asmsub  restore_irq() clobbers(A) {
 	    and  #%11110000     ; disable all Vera IRQs but the vsync
 	    ora  #%00000001
 	    sta  cx16.VERA_IEN
-	    cli
+	    plp
 	    rts
         .section BSS_NOCLEAR
 _orig_irqvec    .word  ?
@@ -1636,9 +1683,10 @@ _orig_irqvec    .word  ?
 asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
     ; Sets the handler for the LINE interrupt, and enable (only) that interrupt.
 	%asm {{
+	        php
             sei
-            sta  _vector
-            sty  _vector+1
+            sta  user_vector
+            sty  user_vector+1
             lda  cx16.r0
             ldy  cx16.r0+1
             lda  cx16.VERA_IEN
@@ -1652,7 +1700,7 @@ asmsub  set_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
             sta  cbm.CINV
             lda  #>_raster_irq_handler
             sta  cbm.CINV+1
-            cli
+            plp
             rts
 
 _raster_irq_handler
@@ -1668,15 +1716,32 @@ _raster_irq_handler
             pla
             rti
 _run_custom
-		    jmp  (_vector)
+		    jmp  (user_vector)
 		    .section BSS
-_vector	    .word ?
+user_vector	    .word ?
 		    .send BSS
 
 		    ; !notreached!
         }}
 }
 
+asmsub update_rasterirq(uword handler @AY, uword rasterpos @R0) clobbers(A) {
+    ; -- just update the IRQ handler and raster line position for the raster IRQ
+    ;    this is much more efficient than calling set_rasterirq() again every time.
+    ;    (but you have to call that one initially at least once to setup the prog8 handler itself)
+    %asm {{
+        php
+        sei
+        sta  sys.set_rasterirq.user_vector
+        sty  sys.set_rasterirq.user_vector+1
+        lda  cx16.r0L
+        ldy  cx16.r0H
+        jsr  set_rasterline
+        plp
+        rts
+    }}
+}
+
 asmsub  set_rasterline(uword line @AY) {
     %asm {{
         php
@@ -1732,13 +1797,15 @@ _loop       lda  P8ZP_SCRATCH_W1
             bne  +
             rts
 
-+           sei
++           php
+            sei
             jsr  cbm.RDTIM
-            cli
+            plp
             sta  P8ZP_SCRATCH_B1
--           sei
+-           php
+            sei
             jsr  cbm.RDTIM
-            cli
+            plp
             cmp  P8ZP_SCRATCH_B1
             beq  -
 
@@ -1759,6 +1826,26 @@ _loop       lda  P8ZP_SCRATCH_W1
         }}
     }
 
+    asmsub waitrasterline(uword line @AY) {
+        ; -- CPU busy wait until the given raster line is reached
+        %asm {{
+            cpy  #0
+            bne  _larger
+-           cmp  cx16.VERA_SCANLINE_L
+            bne  -
+            bit  cx16.VERA_IEN
+            bvs  -
+            rts
+_larger
+            cmp  cx16.VERA_SCANLINE_L
+            bne  _larger
+            bit  cx16.VERA_IEN
+            bvc  _larger
+            rts
+        }}
+    }
+
+
     asmsub internal_stringcopy(str source @R0, str target @AY) clobbers (A,Y) {
         ; Called when the compiler wants to assign a string value to another string.
         %asm {{
@@ -2093,6 +2180,47 @@ save_SCRATCH_ZPWORD2	.word  ?
         }}
     }
 
+    inline asmsub pushl(long value @R0R1_32) {
+        %asm {{
+            lda  cx16.r0
+            pha
+            lda  cx16.r0+1
+            pha
+            lda  cx16.r0+2
+            pha
+            lda  cx16.r0+3
+            pha
+        }}
+    }
+
+    inline asmsub popl() -> long @R0R1_32 {
+        %asm {{
+            pla
+            sta  cx16.r0+3
+            pla
+            sta  cx16.r0+2
+            pla
+            sta  cx16.r0+1
+            pla
+            sta  cx16.r0
+        }}
+    }
+
+    asmsub cpu_is_65816() -> bool @A {
+        ; -- Returns true when you have a 65816 cpu, false when it's a 6502.
+        %asm {{
+			php
+			clv
+			.byte $e2, $ea  ; SEP #$ea, should be interpreted as 2 NOPs by 6502. 65c816 will set the Overflow flag.
+			bvc +
+			lda #1
+			plp
+			rts
++			lda #0
+			plp
+			rts
+        }}
+    }
 }
 
 p8_sys_startup {

compiler/res/prog8lib/cx16/textio.p8

@@ -615,6 +615,13 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub size() clobbers(A) -> ubyte @X, ubyte @Y {
+    ; -- returns the text screen width in X and height in Y (number of columns and rows)
+    %asm {{
+        jmp  cbm.SCREEN
+    }}
+}
+
 asmsub waitkey() -> ubyte @A {
     %asm {{
 -       jsr cbm.GETIN

compiler/res/prog8lib/cx16/verafx.p8

@@ -116,26 +116,29 @@ verafx {
 
 
     asmsub mult16(uword value1 @R0, uword value2 @R1) clobbers(X) -> uword @AY {
-        ; Returns the 16 bits unsigned result of R0*R1 in AY.
+        ; Returns the lower 16 bits unsigned result of R0*R1 in AY
         ; Note: only the lower 16 bits!   (the upper 16 bits are not valid for unsigned word multiplications, only for signed)
         ; Verafx doesn't support unsigned values like this for full 32 bit result.
         ; Note: clobbers VRAM $1f9bc - $1f9bf (inclusive)
         %asm {{
-            lda  cx16.r0
-            sta  P8ZP_SCRATCH_W1
-            lda  cx16.r0+1
-            sta  P8ZP_SCRATCH_W1+1
-            jsr  verafx.muls
-            ldx  P8ZP_SCRATCH_W1
-            stx  cx16.r0
-            ldx  P8ZP_SCRATCH_W1+1
-            stx  cx16.r0+1
+            jmp  muls16
+        }}
+    }
+
+    asmsub muls16(word value1 @R0, word value2 @R1) clobbers(X) -> word @AY {
+        ; Returns just the lower 16 bits signed result of the multiplication in cx16.AY.
+        ; Note: clobbers R0, R1, and VRAM $1f9bc - $1f9bf (inclusive)
+        %asm {{
+            jsr  muls
+            lda  cx16.r0L
+            ldy  cx16.r0H
             rts
         }}
     }
 
-    asmsub muls(word value1 @R0, word value2 @R1) clobbers(X) -> word @AY, word @R0 {
-        ; Returns the 32 bits signed result in AY and R0  (lower word, upper word).
+
+    asmsub muls(word value1 @R0, word value2 @R1) clobbers(X) -> long @R0R1_32 {
+        ; Returns the 32 bits signed result in R0:R1  (lower word, upper word).
         ; Vera Fx multiplication support only works on signed values!
         ; Note: clobbers VRAM $1f9bc - $1f9bf (inclusive)
         %asm {{
@@ -171,12 +174,14 @@ verafx {
             stz  cx16.VERA_DATA0      ; multiply and write out result
             lda  #%00010001           ; $01 with Increment 1
             sta  cx16.VERA_ADDR_H     ; so we can read out the result
-            lda  cx16.VERA_DATA0      ; store the lower 16 bits of the result in AY
+            lda  cx16.VERA_DATA0      ; store the lower 16 bits of the result in R0
             ldy  cx16.VERA_DATA0
-            ldx  cx16.VERA_DATA0      ; store the upper 16 bits of the result in R0
-            stx  cx16.r0s
-            ldx  cx16.VERA_DATA0
-            stx  cx16.r0s+1
+            sta  cx16.r0L
+            sty  cx16.r0H
+            lda  cx16.VERA_DATA0      ; store the upper 16 bits of the result in R1
+            ldy  cx16.VERA_DATA0      ; store the upper 16 bits of the result in R1
+            sta  cx16.r1L
+            sty  cx16.r1H
             stz  cx16.VERA_FX_CTRL    ; Cache write disable
             stz  cx16.VERA_FX_MULT    ; $9F2C  reset multiply bit
             stz  cx16.VERA_CTRL       ; reset DCSEL

compiler/res/prog8lib/math.p8

@@ -220,6 +220,12 @@ _sinecosR8	.char  trunc(127.0 * sin(range(180+45) * rad(360.0/180.0)))
         }}
     }
 
+    sub mul32(uword a, uword b) -> long {
+        ; return 32 bits result of a*b
+        cx16.r2 = a*b
+        return mklong2(mul16_last_upper(), cx16.r2)
+    }
+
 sub direction_sc(byte x1, byte y1, byte x2, byte y2) -> ubyte {
     ; From a pair of signed coordinates around the origin, calculate discrete direction between 0 and 23 into A.
     cx16.r0L = 3        ; quadrant
@@ -548,14 +554,13 @@ log2_tab
     }
 
     sub crc16(^^ubyte data, uword length) -> uword {
-        ; calculates the CRC16 (XMODEM) checksum of the buffer.
+        ; calculates the CRC16 (XMODEM) checksum of the buffer. Clobbers R14 and R15.
         ; There are also "streaming" crc16_start/update/end routines below, that allow you to calculate crc16 for data that doesn't fit in a single memory block.
+        cx16.r14 = data
         crc16_start()
-        cx16.r13 = data
-        cx16.r14 = data+length
-        while cx16.r13!=cx16.r14 {
-            crc16_update(@(cx16.r13))
-            cx16.r13++
+        repeat length {
+            crc16_update(@(cx16.r14))
+            cx16.r14++
         }
         return crc16_end()
     }
@@ -602,19 +607,17 @@ log2_tab
         return cx16.r15
     }
 
-    sub crc32(^^ubyte data, uword length) {
+    sub crc32(^^ubyte data, uword length) -> long {
         ; Calculates the CRC-32 (ISO-HDLC/PKZIP) checksum of the buffer.
-        ; because prog8 doesn't have 32 bits integers, we have to split up the calculation over 2 words.
-        ; result stored in cx16.r14 (low word) and cx16.r15 (high word)
+        ; Clobbers R13 through R15,  returns the 32 bits result as long
         ; There are also "streaming" crc32_start/update/end routines below, that allow you to calculate crc32 for data that doesn't fit in a single memory block.
+        cx16.r13 = data
         crc32_start()
-        cx16.r12 = data
-        cx16.r13 = data+length
-        while cx16.r12!=cx16.r13 {
-            crc32_update(@(cx16.r12))
-            cx16.r12++
+        repeat length {
+            crc32_update(@(cx16.r13))
+            cx16.r13++
         }
-        crc32_end()
+        return crc32_end()
     }
 
     sub crc32_start() {
@@ -666,13 +669,12 @@ log2_tab
 ;        }
     }
 
-    sub crc32_end() {
-        ; finalize the "streaming" crc32
-        ; result stored in cx16.r14 (low word) and cx16.r15 (high word)
-        void crc32_end_result()
+    sub crc32_end() -> long {
+        ; finalize the "streaming" crc32 and returns the result
+        return crc32_end_result()
     }
 
-    asmsub crc32_end_result() -> uword @R15, uword @R14 {
+    asmsub crc32_end_result() -> long @R14R15_32 {
         ; finalize the "streaming" crc32
         ; returns the result value in cx16.r15 (high word) and r14 (low word)
         %asm {{

compiler/res/prog8lib/pet32/syslib.p8

@@ -81,6 +81,29 @@ asmsub RDTIM16() clobbers(X) -> uword @AY {
     }}
 }
 
+asmsub SETTIML(long jiffies @R0R1_32) {
+    ; -- just like SETTIM, but with a single 32 bit (lower 24 bits used) argument.
+    %asm {{
+        lda  cx16.r0
+        ldx  cx16.r0+1
+        ldy  cx16.r0+2
+        jmp  SETTIM
+    }}
+}
+
+asmsub RDTIML() clobbers(X) -> long @R0R1_32 {
+    ; --  like RDTIM() and returning the timer value as a 32 bit (lower 24 bits used) value.
+    %asm {{
+        jsr  RDTIM
+        sta  cx16.r0
+        stx  cx16.r0+1
+        sty  cx16.r0+2
+        lda  #0
+        sta  cx16.r0+3
+        rts
+    }}
+}
+
 asmsub kbdbuf_clear() {
     ; -- convenience helper routine to clear the keyboard buffer
     %asm {{
@@ -509,6 +532,37 @@ save_SCRATCH_ZPWORD2	.word  ?
         }}
     }
 
+    inline asmsub pushl(long value @R0R1_32) {
+        %asm {{
+            lda  cx16.r0
+            pha
+            lda  cx16.r0+1
+            pha
+            lda  cx16.r0+2
+            pha
+            lda  cx16.r0+3
+            pha
+        }}
+    }
+
+    inline asmsub popl() -> long @R0R1_32 {
+        %asm {{
+            pla
+            sta  cx16.r0+3
+            pla
+            sta  cx16.r0+2
+            pla
+            sta  cx16.r0+1
+            pla
+            sta  cx16.r0
+        }}
+    }
+
+    sub cpu_is_65816() -> bool {
+        ; Returns true when you have a 65816 cpu, false when it's a 6502.
+        return false
+    }
+
 }
 
 cx16 {
@@ -686,11 +740,6 @@ cx16 {
         }}
     }
 
-    sub cpu_is_65816() -> bool {
-        ; Returns true when you have a 65816 cpu, false when it's a 6502.
-        return false
-    }
-
 }
 
 p8_sys_startup {

compiler/res/prog8lib/pet32/textio.p8

@@ -182,8 +182,8 @@ sub  setclr  (ubyte col, ubyte row, ubyte color)  {
 }
 
 
-sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor_ignored)  {
-	; ---- set char at the given position on the screen. charcolor is ignored on PET
+sub  setcc  (ubyte col, ubyte row, ubyte character, ubyte charcolor)  {
+	; ---- set char at the given position on the screen. NOTE: charcolor is ignored on PET
 	%asm {{
   		lda  row
 		asl  a
@@ -236,6 +236,17 @@ asmsub height() clobbers(X, Y) -> ubyte @A {
     }}
 }
 
+asmsub size() clobbers(A) -> ubyte @X, ubyte @Y {
+    ; -- returns the text screen width in X and height in Y (number of columns and rows)
+    %asm {{
+        jsr  width
+        tax
+        jsr  height
+        tay
+        rts
+    }}
+}
+
 asmsub waitkey() -> ubyte @A {
     %asm {{
 -       jsr cbm.GETIN

compiler/res/prog8lib/prog8_funcs.asm

@@ -29,6 +29,36 @@ abs_w_into_AY	.proc
 +		rts
 		.pend
 
+abs_l_into_R14R15	.proc
+	; -- R14:R15 = abs(R14:R15)
+    lda  cx16.r15H
+    bmi  +   ; Negative if high bit of highest byte is set
+    rts
++
+    ; Invert all four bytes
+    lda  cx16.r14L
+    eor  #$FF
+    sta  cx16.r14L
+    lda  cx16.r14H
+    eor  #$FF
+    sta  cx16.r14H
+    lda  cx16.r15L
+    eor  #$FF
+    sta  cx16.r15L
+    lda  cx16.r15H
+    eor  #$FF
+    sta  cx16.r15H
+    ; Add 1 to whole 32-bit value
+    inc  cx16.r14L
+    bne  +
+    inc  cx16.r14H
+    bne  +
+    inc  cx16.r15L
+    bne  +
+    inc  cx16.r15H
++   rts
+		.pend
+
 func_sign_b_into_A	.proc
 		cmp  #0
 		beq  _zero
@@ -59,17 +89,39 @@ _possibly_zero	cmp  #0
 
 func_sign_w_into_A	.proc
 		cpy  #0
-		beq  _possibly_zero
-		bmi  _neg
-_pos		lda  #1
-		rts
-_neg		lda  #-1
-		rts
-_possibly_zero	cmp  #0
-		bne  _pos
+		bmi  _negative
+		bne  _positive
+		cmp  #0
+		beq  _zero
+_positive	lda  #1
+_zero		rts
+_negative	lda  #-1
 		rts
 		.pend
 
+func_sign_l_r14r15_into_A      .proc
+    lda  cx16.r14+3              ; msb
+    bmi  _negative
+    bne  _positive
+    lda  cx16.r14+2
+    bne  _positive
+    lda  cx16.r14+1
+    bne  _positive
+    lda  cx16.r14
+    beq  _zero
+    lda  #1
+_zero
+    rts
+_negative
+    lda  #-1
+    rts
+_positive
+    lda  #1
+    rts
+
+.pend
+
+
 
 func_sqrt16_into_A	.proc
 		; integer square root
@@ -405,6 +457,7 @@ func_peek   .proc
 	; -- read the byte value on the address in AY, into A
 	sta  P8ZP_SCRATCH_W1
 	sty  P8ZP_SCRATCH_W1+1
+from_scratchW1
 	ldy  #0
 	lda  (P8ZP_SCRATCH_W1),y
 	rts
@@ -414,6 +467,7 @@ func_peekw   .proc
 	; -- read the word value on the address in AY, into AY
 	sta  P8ZP_SCRATCH_W1
 	sty  P8ZP_SCRATCH_W1+1
+from_scratchW1
 	ldy  #0
 	lda  (P8ZP_SCRATCH_W1),y
 	pha
@@ -424,6 +478,26 @@ func_peekw   .proc
 	rts
 	.pend
 
+func_peekl   .proc
+	; -- read the ;pmg value on the address in AY, into R14:R15
+	sta  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+from_scratchW1
+	ldy  #0
+	lda  (P8ZP_SCRATCH_W1),y
+	sta  cx16.r14
+	iny
+	lda  (P8ZP_SCRATCH_W1),y
+	sta  cx16.r14+1
+	iny
+	lda  (P8ZP_SCRATCH_W1),y
+	sta  cx16.r14+2
+	iny
+	lda  (P8ZP_SCRATCH_W1),y
+	sta  cx16.r14+3
+	rts
+	.pend
+
 
 func_pokew   .proc
 	; -- store the word value in AY in the address in P8ZP_SCRATCH_W1
@@ -436,6 +510,35 @@ func_pokew   .proc
 	rts
 	.pend
 
+func_pokew_scratchW2   .proc
+	; -- store the word value in AY in the address in P8ZP_SCRATCH_W2
+	sty  P8ZP_SCRATCH_REG
+	ldy  #0
+	sta  (P8ZP_SCRATCH_W2),y
+	iny
+	lda  P8ZP_SCRATCH_REG
+	sta  (P8ZP_SCRATCH_W2),y
+	rts
+	.pend
+
+func_pokel   .proc
+	; -- store the long value in R14:R15 in the address in AY
+	sta  P8ZP_SCRATCH_W1
+	sty  P8ZP_SCRATCH_W1+1
+	ldy  #0
+	lda  cx16.r14
+	sta  (P8ZP_SCRATCH_W1),y
+	iny
+	lda  cx16.r14+1
+	sta  (P8ZP_SCRATCH_W1),y
+	iny
+	lda  cx16.r14+2
+	sta  (P8ZP_SCRATCH_W1),y
+	iny
+	lda  cx16.r14+3
+	sta  (P8ZP_SCRATCH_W1),y
+	rts
+	.pend
 
 func_clamp_byte .proc
 	; signed value in A, result in A

compiler/res/prog8lib/prog8_lib.asm

@@ -266,6 +266,25 @@ strcpy		.proc
 		rts
 		.pend
 
+strncpy		.proc
+		; copy a string (must be 0-terminated) from A/Y to (P8ZP_SCRATCH_W1)
+		; with maximum length to copy in X.
+		; returns the length of the string that was copied in Y.
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #$ff
+-		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		sta  (P8ZP_SCRATCH_W1),y
+		beq  +
+		dex
+		bne  -
+		iny
+		lda  #0
+		sta  (P8ZP_SCRATCH_W1),y
++		rts
+		.pend
+
 strcmp_expression	.proc
 		; -- compare strings, result in A
 		lda  _arg_s2
@@ -449,3 +468,195 @@ _fullpage       lda  (P8ZP_SCRATCH_W1),y
                 bne  _fullpage
                 rts
 		.pend
+
+
+long_not_equals     .proc
+		; checks if the 32 bits long value pointed to by AY is NOT equal to the one pointed to by P8ZP_SCRATCH_W1
+		; returns A=1 if NOT equals otherwise A=0
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #3
+-		lda  (P8ZP_SCRATCH_W1),y
+		cmp  (P8ZP_SCRATCH_W2),y
+		bne  _notequal
+		dey
+		bpl  -
+		lda  #0
+		rts
+_notequal       lda  #1
+		rts
+		.pend
+
+
+long_add_inplace        .proc
+		; long pointed to by AY += long pointed to by P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		clc
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W2),y
+		adc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		adc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		adc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		adc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		rts
+		.pend
+
+long_sub_inplace    .proc
+		; long pointed to by AY -= long pointed to by P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		sec
+		lda  (P8ZP_SCRATCH_W2),y
+		sbc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		sbc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		sbc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		sbc  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		rts
+		.pend
+
+long_or_inplace       .proc
+		; long pointed to by AY OR= long pointed to by P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W2),y
+		ora  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		ora  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		ora  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		ora  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		rts
+		.pend
+
+long_and_inplace       .proc
+		; long pointed to by AY AND= long pointed to by P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W2),y
+		and  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		and  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		and  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		and  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		rts
+		.pend
+
+long_xor_inplace       .proc
+		; long pointed to by AY XOR= long pointed to by P8ZP_SCRATCH_W1
+		sta  P8ZP_SCRATCH_W2
+		sty  P8ZP_SCRATCH_W2+1
+		ldy  #0
+		lda  (P8ZP_SCRATCH_W2),y
+		eor  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		eor  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		eor  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		iny
+		lda  (P8ZP_SCRATCH_W2),y
+		eor  (P8ZP_SCRATCH_W1),y
+		sta  (P8ZP_SCRATCH_W2),y
+		rts
+		.pend
+
+long_shiftleftX_inplace         .proc
+		; bit shift left X bits the long value pointed to by AY
+		cpx  #0
+		beq  _end
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+-		ldy  #0
+		lda  (P8ZP_SCRATCH_W1),y
+		asl  a
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		rol  a
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		rol  a
+		sta  (P8ZP_SCRATCH_W1),y
+		iny
+		lda  (P8ZP_SCRATCH_W1),y
+		rol  a
+		sta  (P8ZP_SCRATCH_W1),y
+		dex
+		bne  -
+_end		rts
+		.pend
+
+long_shiftrightX_inplace        .proc
+		; bit shift right X bits the long value pointed to by AY
+		cpx  #0
+		beq  _end
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+-		ldy  #3
+		lda  (P8ZP_SCRATCH_W1),y
+		asl  a  ; save sign bit
+		lda  (P8ZP_SCRATCH_W1),y
+		ror  a
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		ror  a
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		ror  a
+		sta  (P8ZP_SCRATCH_W1),y
+		dey
+		lda  (P8ZP_SCRATCH_W1),y
+		ror  a
+		sta  (P8ZP_SCRATCH_W1),y
+		dex
+		bne  -
+_end		rts
+		.pend

compiler/res/prog8lib/shared_cbm_diskio.p8

@@ -550,7 +550,7 @@ done:
         return list_filename
 
 io_error:
-        list_filename = "io error"
+        void strings.copy("io error", list_filename)
         goto done
     }
 

compiler/res/prog8lib/shared_cbm_textio_functions.p8

@@ -109,6 +109,12 @@ txt {
         }}
     }
 
+    sub print_ulhex(long value, bool prefix) {
+        if prefix
+            cbm.CHROUT('$')
+        print(conv.str_ulhex(value))
+    }
+
     asmsub  print_uw0  (uword value @ AY) clobbers(A,X,Y)  {
         ; ---- print the uword in A/Y in decimal form, with left padding 0s (5 positions total)
         %asm {{
@@ -165,6 +171,11 @@ _allzero    lda  #'0'
         }}
     }
 
+    sub print_l(long value) {
+        ; prints a 32 bit value to the screen
+        print(conv.str_l(value))
+    }
+
     asmsub  input_chars  (^^ubyte buffer @ AY) clobbers(A) -> ubyte @ Y  {
         ; ---- Input a string (max. 80 chars) from the keyboard, in PETSCII encoding.
         ;      Returns length in Y. (string is terminated with a 0 byte as well)

compiler/res/prog8lib/shared_compression.p8

@@ -1,4 +1,4 @@
-; **experimental** data compression/decompression routines, API subject to change!!
+; data compression/decompression routines
 ; This file contains the shared routines that work on all targets.
 
 compression {

compiler/res/prog8lib/shared_floats_functions.p8

@@ -160,6 +160,9 @@ sub log2(float value) -> float {
         ldy  #>FL_LOG2_const
         jsr  MOVFM
         jmp  FDIVT
+
+FL_LOG2_const	.byte  $80, $31, $72, $17, $f8	; log(2)
+        ; !notreached!
     }}
 }
 

compiler/res/prog8lib/sorting.p8

@@ -138,31 +138,29 @@ _done
         }
     }
 
-    ; TODO convert to ^^uword once code size regression is fixed
-    sub shellsort_uw(uword @requirezp values, ubyte num_elements) {
+    sub shellsort_uw(^^uword @requirezp values, ubyte num_elements) {
         ; sorts the values array (no-split unsigned words).
         num_elements--
         ubyte gap
         for gap in [132, 57, 23, 10, 4, 1] {
             ubyte i
             for i in gap to num_elements {
-                uword @zp temp = peekw(values+i*$0002)
+                uword @zp temp = values[i]
                 ubyte @zp j = i
                 ubyte @zp k = j-gap
                 while j>=gap {
-                    uword @zp v = peekw(values+k*$0002)
+                    uword @zp v = values[k]
                     if v <= temp break
-                    pokew(values+j*$0002, v)
+                    values[j] = v
                     j = k
                     k -= gap
                 }
-                pokew(values+j*$0002, temp)
+                values[j] = temp
             }
         }
     }
 
-    ; TODO convert to ^^uword once code size regression is fixed
-    sub shellsort_by_ub(^^ubyte @requirezp ub_keys, uword @requirezp wordvalues, ubyte num_elements) {
+    sub shellsort_by_ub(^^ubyte @requirezp ub_keys, ^^uword @requirezp wordvalues, ubyte num_elements) {
         ; sorts the 'wordvalues' array (no-split array of words) according to the 'ub_keys' array (which also gets sorted ofcourse).
         num_elements--
         ubyte @zp gap
@@ -170,7 +168,7 @@ _done
             ubyte i
             for i in gap to num_elements {
                 ubyte @zp temp = ub_keys[i]
-                uword temp_wv = peekw(wordvalues + i*$0002)
+                uword temp_wv = wordvalues[i]
                 ubyte @zp j = i
                 ubyte @zp k = j-gap
                 repeat {
@@ -178,18 +176,17 @@ _done
                     if v <= temp break
                     if j < gap break
                     ub_keys[j] = v
-                    pokew(wordvalues + j*$0002, peekw(wordvalues + k*$0002))
+                    wordvalues[j] = wordvalues[k]
                     j = k
                     k -= gap
                 }
                 ub_keys[j] = temp
-                pokew(wordvalues + j*$0002, temp_wv)
+                wordvalues[j] = temp_wv
             }
         }
     }
 
-    ; TODO convert to ^^uword once code size regression is fixed
-    sub shellsort_by_uw(uword @requirezp uw_keys, uword @requirezp wordvalues, ubyte num_elements) {
+    sub shellsort_by_uw(^^uword @requirezp uw_keys, ^^uword @requirezp wordvalues, ubyte num_elements) {
         ; sorts the 'wordvalues' array according to the 'uw_keys' array (which also gets sorted ofcourse).
         ; both arrays should be no-split array of words. uw_keys are unsigned.
         num_elements--
@@ -197,20 +194,20 @@ _done
         for gap in [132, 57, 23, 10, 4, 1] {
             ubyte i
             for i in gap to num_elements {
-                uword @zp temp = peekw(uw_keys+i*$0002)
-                uword temp_wv = peekw(wordvalues + i*$0002)
+                uword @zp temp = uw_keys[i]
+                uword temp_wv = wordvalues[i]
                 ubyte @zp j = i
                 ubyte @zp k = j-gap
                 while j>=gap {
-                    uword @zp v = peekw(uw_keys+k*2)
+                    uword @zp v = uw_keys[k]
                     if v <= temp break
-                    pokew(uw_keys+j*2, v)
-                    pokew(wordvalues + j*$0002, peekw(wordvalues + k*$0002))
+                    uw_keys[j] = v
+                    wordvalues[j] = wordvalues[k]
                     j = k
                     k -= gap
                 }
-                pokew(uw_keys+j*2, temp)
-                pokew(wordvalues + j*$0002, temp_wv)
+                uw_keys[j] = temp
+                wordvalues[j] = temp_wv
             }
         }
     }

compiler/res/prog8lib/strings.p8

@@ -213,8 +213,6 @@ _found      tya
     asmsub copy(str source @R0, str 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
@@ -224,6 +222,18 @@ _found      tya
         }}
     }
 
+    asmsub ncopy(str source @R0, str target @AY, ubyte maxlength @X) clobbers(A, X) -> ubyte @Y {
+        ; Copy a string to another, overwriting that one.
+        ; Returns the length of the string that was copied.
+        %asm {{
+		sta  P8ZP_SCRATCH_W1
+		sty  P8ZP_SCRATCH_W1+1
+		lda  cx16.r0
+		ldy  cx16.r0+1
+		jmp  prog8_lib.strncpy
+        }}
+    }
+
     asmsub append(str target @R0, str suffix @R1) clobbers(Y) -> ubyte @A {
         ; Append the suffix string to the target. (make sure the buffer is large enough!)
         ; Returns the length of the resulting string.
@@ -249,6 +259,41 @@ _found      tya
         }}
     }
 
+    asmsub nappend(str target @R0, str suffix @R1, ubyte maxlength @X) clobbers(Y) -> ubyte @A {
+        ; Append the suffix string to the target. (make sure the buffer is large enough!)
+        ; Returns the length of the resulting string.
+        %asm {{
+            lda  cx16.r0
+            ldy  cx16.r0+1
+            jsr  length
+            sty  P8ZP_SCRATCH_B1
+            cpx  P8ZP_SCRATCH_B1
+            beq  _max_too_small
+            bmi  _max_too_small
+            txa
+            sec
+            sbc  P8ZP_SCRATCH_B1
+            sta  P8ZP_SCRATCH_B1
+            tya
+            clc
+            adc  cx16.r0
+            sta  P8ZP_SCRATCH_W1
+            lda  cx16.r0+1
+            adc  #0
+            sta  P8ZP_SCRATCH_W1+1
+            lda  cx16.r1
+            ldy  cx16.r1+1
+            ldx  P8ZP_SCRATCH_B1
+            jsr  prog8_lib.strncpy
+            tya
+            clc
+            adc  P8ZP_SCRATCH_B1
+            rts
+_max_too_small
+            rts
+        }}
+    }
+
     asmsub compare(str string1 @R0, str string2 @AY) clobbers(Y) -> byte @A {
         ; Compares two strings for sorting.
         ; Returns -1 (255), 0 or 1, meaning: string1 sorts before, equal or after string2.

compiler/res/prog8lib/virtual/bcd.p8

@@ -0,0 +1,3 @@
+bcd {
+    ; virtual machine does not support BCD
+}

compiler/res/prog8lib/virtual/conv.p8

@@ -102,7 +102,7 @@ sub  str_uwbin  (uword value) -> str {
 }
 
 sub  str_uwhex  (uword value) -> str {
-    ; ---- convert the uword in A/Y in hexadecimal string form (4 digits)
+    ; ---- convert the uword in hexadecimal string form (4 digits)
     ubyte bits = msb(value)
     string_out[0] = hex_digits[bits>>4]
     string_out[1] = hex_digits[bits&15]
@@ -113,6 +113,26 @@ sub  str_uwhex  (uword value) -> str {
     return string_out
 }
 
+sub  str_ulhex  (long value) -> str {
+    ; ---- convert the long in hexadecimal string form (8 digits)
+    uword upperw = msw(value)
+    uword lowerw = lsw(value)
+    ubyte bits = msb(upperw)
+    string_out[0] = hex_digits[bits>>4]
+    string_out[1] = hex_digits[bits&15]
+    bits = lsb(upperw)
+    string_out[2] = hex_digits[bits>>4]
+    string_out[3] = hex_digits[bits&15]
+    bits = msb(lowerw)
+    string_out[4] = hex_digits[bits>>4]
+    string_out[5] = hex_digits[bits&15]
+    bits = lsb(lowerw)
+    string_out[6] = hex_digits[bits>>4]
+    string_out[7] = hex_digits[bits&15]
+    string_out[8] = 0
+    return string_out
+}
+
 sub  str_uw0  (uword value) -> str {
     ; ---- convert the uword in A/Y in decimal string form, with left padding 0s (5 positions total)
     uword value2 = value/10
@@ -141,7 +161,7 @@ sub  str_uw  (uword value) -> str {
 }
 
 sub  str_w  (word value) -> str {
-    ; ---- convert the (signed) word in A/Y in decimal string form, without left padding 0's
+    ; ---- convert the (signed) word into decimal string form, without left padding 0's
     ^^ubyte out_ptr = &string_out
     if value<0 {
         @(out_ptr) = '-'
@@ -152,6 +172,16 @@ sub  str_w  (word value) -> str {
     return string_out
 }
 
+sub  str_l  (long value) -> str {
+    ; ---- convert the (signed) long into decimal string form, without left padding 0's
+    %ir {{
+        loadm.l r99200,conv.str_l.value
+        load.w r99000,conv.string_out
+        syscall 60 (r99200.l, r99000.w) : r99000.w
+        returnr.w r99000
+    }}
+}
+
 sub internal_str_uw(uword value, str out_ptr) {
     uword value2 = value/10
     ubyte digits = value-value2*10 as ubyte
@@ -249,6 +279,26 @@ sub  hex2uword(str string) -> uword {
     }
 }
 
+sub  hex2long(str string) -> long {
+    ; -- hexadecimal string (with or without '$') to long.
+    ;    stops parsing at the first character that's not a hex digit (except leading $)
+    long result
+    ubyte char
+    if @(string)=='$'
+        string++
+    repeat {
+        char = @(string)
+        if char==0
+            return result
+        result <<= 4
+        if char>='0' and char<='9'
+            result |= char-'0'
+        else
+            result |= char-'a'+10
+        string++
+    }
+}
+
 sub  bin2uword(str string) -> uword {
     ; -- binary string (with or without '%') to uword.
     ;    stops parsing at the first character that's not a 0 or 1. (except leading %)
@@ -267,12 +317,15 @@ sub  bin2uword(str string) -> uword {
     }
 }
 
-sub  any2uword(str string) -> uword {
+sub  any2uword(str string) -> uword, ubyte {
     ; -- convert any number string (any prefix allowed) to uword.
+    ;    returns the parsed word value, and the number of processed characters (including the prefix symbol)
+    ubyte length
+    while string[length]!=0 length++
     when string[0] {
-        '$' -> return hex2uword(string)
-        '%' -> return bin2uword(string)
-        else -> return str2uword(string)
+        '$' -> return hex2uword(string), length
+        '%' -> return bin2uword(string), length
+        else -> return str2uword(string), length
     }
 }
 

compiler/res/prog8lib/virtual/diskio.p8

@@ -206,7 +206,7 @@ diskio {
 
     sub save(str filenameptr, uword start_address, uword savesize) -> bool {
         %ir {{
-            load.b r99100,0
+            load.b r99100,#0
             loadm.w r99000,diskio.save.filenameptr
             loadm.w r99001,diskio.save.start_address
             loadm.w r99002,diskio.save.savesize
@@ -218,7 +218,7 @@ diskio {
     ; like save() but omits the 2 byte prg header.
     sub save_raw(str filenameptr, uword start_address, uword savesize) -> bool {
         %ir {{
-            load.b r99100,1
+            load.b r99100,#1
             loadm.w r99000,diskio.save_raw.filenameptr
             loadm.w r99001,diskio.save_raw.start_address
             loadm.w r99002,diskio.save_raw.savesize
@@ -284,7 +284,7 @@ diskio {
         ; get the load adress from a PRG file (usually $0801 but it can be different)
         if f_open(filename) {
             uword address
-            f_read(&address, 2)
+            void f_read(&address, 2)
             f_close()
             return address
         }

compiler/res/prog8lib/virtual/math.p8

@@ -304,6 +304,12 @@ math {
         }}
     }
 
+    sub mul32(uword a, uword b) -> long {
+        ; return 32 bits result of a*b
+        cx16.r2 = a*b
+        return mklong2(mul16_last_upper(), cx16.r2)
+    }
+
     sub diff(ubyte b1, ubyte b2) -> ubyte {
         if b1>b2
             return b1-b2
@@ -353,10 +359,9 @@ math {
         return cx16.r15
     }
 
-    sub crc32(^^ubyte data, uword length) {
+    sub crc32(^^ubyte data, uword length) -> long {
         ; Calculates the CRC-32 (ISO-HDLC/PKZIP) checksum of the buffer.
-        ; because prog8 doesn't have 32 bits integers, we have to split up the calculation over 2 words.
-        ; result stored in cx16.r14 (low word) and cx16.r15 (high word)
+        ; Clobbers R12 through R15, returns the result in R0:R1
         ; There are also "streaming" crc32_start/update/end routines below, that allow you to calculate crc32 for data that doesn't fit in a single memory block.
         crc32_start()
         cx16.r12 = data
@@ -365,7 +370,7 @@ math {
             crc32_update(@(cx16.r12))
             cx16.r12++
         }
-        crc32_end()
+        return crc32_end()
     }
 
     sub crc32_start() {
@@ -391,11 +396,11 @@ math {
         }
     }
 
-    sub crc32_end() {
-        ; finalize the "streaming" crc32
-        ; result stored in cx16.r14 (low word) and cx16.r15 (high word)
+    sub crc32_end() -> long {
+        ; finalize the "streaming" crc32 and return the result in R0:R1
         cx16.r15 ^= $ffff
         cx16.r14 ^= $ffff
+        return mklong2(cx16.r15, cx16.r14)
     }
 
     ; there's no crc32_end_result() here because IR cannot return multiple values yet

compiler/res/prog8lib/virtual/monogfx.p8

@@ -302,23 +302,11 @@ monogfx {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
-
         while radius>=yy {
             horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
             horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
-                last_y4 = new_y4
-            }
+            horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -327,10 +315,6 @@ monogfx {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-        horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub safe_disc(uword @zp xcenter, uword @zp ycenter, ubyte @zp radius, bool draw) {
@@ -340,23 +324,11 @@ monogfx {
             return
         ubyte @zp yy = 0
         word @zp decisionOver2 = (1 as word)-radius
-        uword last_y3 = ycenter+radius
-        uword last_y4 = ycenter-radius
-        uword new_y3, new_y4
-
         while radius>=yy {
             safe_horizontal_line(xcenter-radius, ycenter+yy, radius*$0002+1, draw)
             safe_horizontal_line(xcenter-radius, ycenter-yy, radius*$0002+1, draw)
-            new_y3 = ycenter+radius
-            if new_y3 != last_y3 {
-                safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-                last_y3 = new_y3
-            }
-            new_y4 = ycenter-radius
-            if new_y4 != last_y4 {
-                safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
-                last_y4 = new_y4
-            }
+            safe_horizontal_line(xcenter-yy, ycenter+radius, yy*$0002+1, draw)
+            safe_horizontal_line(xcenter-yy, ycenter-radius, yy*$0002+1, draw)
             yy++
             if decisionOver2>=0 {
                 radius--
@@ -365,10 +337,6 @@ monogfx {
             decisionOver2 += yy*$0002
             decisionOver2++
         }
-        ; draw the final two spans
-        yy--
-        safe_horizontal_line(xcenter-yy, last_y3, yy*$0002+1, draw)
-        safe_horizontal_line(xcenter-yy, last_y4, yy*$0002+1, draw)
     }
 
     sub plot(uword @zp xx, uword @zp yy, bool @zp draw) {

compiler/res/prog8lib/virtual/strings.p8

@@ -106,12 +106,23 @@ strings {
     sub copy(str source, str target) -> ubyte {
         ; 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.
         %ir {{
             loadm.w r99000,strings.copy.source
             loadm.w r99001,strings.copy.target
-            syscall 39 (r99000.w, r99001.w): r99100.b
+            load.b r99100,#255
+            syscall 39 (r99000.w, r99001.w, r99100.b): r99100.b
+            returnr.b r99100
+        }}
+    }
+
+    sub ncopy(str source, str target, ubyte maxlength) -> ubyte {
+        ; Copy a string to another, overwriting that one, but limited to the given length.
+        ; Returns the length of the string that was copied.
+        %ir {{
+            loadm.w r99000,strings.ncopy.source
+            loadm.w r99001,strings.ncopy.target
+            loadm.b r99100,strings.ncopy.maxlength
+            syscall 39 (r99000.w, r99001.w, r99100.b): r99100.b
             returnr.b r99100
         }}
     }
@@ -123,6 +134,16 @@ strings {
         return copy(suffix, target+cx16.r0L) + cx16.r0L
     }
 
+    sub nappend(str target, str suffix, ubyte maxlength) -> ubyte {
+        ; Append the suffix string to the target, up to the given maximum length of the combined string.
+        ; Returns the length of the resulting string.
+        cx16.r0L = length(target)
+        if maxlength<cx16.r0L
+            return cx16.r0L
+        maxlength -= cx16.r0L
+        return ncopy(suffix, target+cx16.r0L, maxlength) + cx16.r0L
+    }
+
     sub compare(str st1, str st2) -> byte {
         ; Compares two strings for sorting.
         ; Returns -1 (255), 0 or 1, meaning: string1 sorts before, equal or after string2.

compiler/res/prog8lib/virtual/syslib.p8

@@ -53,7 +53,8 @@ sys {
         %ir {{
             loadm.w r99000,sys.internal_stringcopy.source
             loadm.w r99001,sys.internal_stringcopy.tgt
-            syscall 39 (r99000.w, r99001.w): r99100.b
+            load.b r99100,#255
+            syscall 39 (r99000.w, r99001.w, r99100.b): r99100.b
         }}
     }
 
@@ -192,6 +193,14 @@ sys {
         }}
     }
 
+    sub pushl(long l) {
+        ; note: this *should* be inlined, however since the VM has separate program counter and value stacks, this also works
+        %ir {{
+            loadm.l r99200,sys.pushl.l
+            push.l r99200
+        }}
+    }
+
     sub push_returnaddress(uword w) {
         ; note: this actually doesn't do anything useful on the VM because the code execution doesn't use the simulated cpu stack
         %ir {{
@@ -222,6 +231,14 @@ sys {
         }}
     }
 
+    sub popl() -> long {
+        ; note: this *should* be inlined, however since the VM has separate program counter and value stacks, this also works
+        %ir {{
+            pop.l r99200
+            returnr.l r99200
+        }}
+    }
+
     sub read_flags() -> ubyte {
         ; "simulate" the 6502 status register a little bit
         if_neg {
@@ -244,6 +261,11 @@ sys {
 
         return cx16.r0L
     }
+
+    sub cpu_is_65816() -> bool {
+        ; Returns true when you have a 65816 cpu, false when it's a 6502.
+        return false
+    }
 }
 
 cx16 {