Merge 1d89ef8613 into 5cd1ecb98a

It needs some more thought before committing to it.

.github/workflows/ci.yml

@@ -27,12 +27,12 @@ jobs:
 
     steps:
     - name: Install dependencies
-      run: sudo apt-get install -y libxkbcommon-dev
+      run: sudo apt-get install -y libxkbcommon-dev libasound2-dev
       if: matrix.os == 'ubuntu-latest'
     - name: Checkout
-      uses: actions/checkout@v2
+      uses: actions/checkout@v3
     - name: Cache build dirs
-      uses: actions/cache@v2
+      uses: actions/cache@v3
       with:
         path: |
           ~/.cargo/bin/
@@ -44,7 +44,7 @@ jobs:
     - name: Build
       run: cargo build --release --verbose
     - name: Upload artifact
-      uses: actions/upload-artifact@v2
+      uses: actions/upload-artifact@v3
       with:
         name: uw8-${{ matrix.build }}
         path: target/release/${{ matrix.exe }}

.github/workflows/main.yml

@@ -8,12 +8,12 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: checkout
-        uses: actions/checkout@v2
+        uses: actions/checkout@v3
       - name: build_and_deploy
-        uses: shalzz/zola-deploy-action@v0.14.1
+        uses: shalzz/zola-deploy-action@70a101a14bbdeed13e7a42a9ed06b35c9e9e826e
         env:
           # Target branch
           PAGES_BRANCH: gh-pages
           BUILD_DIR: site
           # Provide personal access token
-          TOKEN: $GITHUB_ACTOR:${{ secrets.GITHUB_TOKEN }}
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Cargo.toml

@@ -1,26 +1,31 @@
 [package]
 name = "uw8"
-version = "0.1.2"
+version = "0.3.0"
 edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [features]
 default = ["native", "browser"]
-native = ["wasmtime"]
+native = ["wasmtime", "uw8-window", "cpal", "rubato" ]
 browser = ["warp", "tokio", "tokio-stream", "webbrowser"]
 
 [dependencies]
-wasmtime = { version = "0.30", optional = true }
+wasmtime = { version = "19.0.1", optional = true }
 anyhow = "1"
-minifb = { version = "0.20", default-features = false, features = ["x11"] }
-notify = "4"
-pico-args = "0.4"
-curlywas = { git = "https://github.com/exoticorn/curlywas.git", rev = "89638565" }
+env_logger = "0.11.3"
+log = "0.4"
+uw8-window = { path = "uw8-window", optional = true }
+notify-debouncer-mini = { version = "0.4.1", default-features = false }
+pico-args = "0.5"
+curlywas = { git = "https://github.com/exoticorn/curlywas.git", rev = "0e7ea50" }
 wat = "1"
 uw8-tool = { path = "uw8-tool" }
 same-file = "1"
-warp = { version = "0.3.2", optional = true }
-tokio = { version = "1.17.0", features = ["sync", "rt"], optional = true }
-tokio-stream = { version = "0.1.8", features = ["sync"], optional = true }
-webbrowser = { version = "0.6.0", optional = true }
+warp = { version = "0.3.6", optional = true }
+tokio = { version = "1.37.0", features = ["sync", "rt"], optional = true }
+tokio-stream = { version = "0.1.15", features = ["sync"], optional = true }
+webbrowser = { version = "0.8.13", optional = true }
+ansi_term = "0.12.1"
+cpal = { version = "0.15.3", optional = true }
+rubato = { version = "0.15.0", optional = true }

README.md

@@ -15,9 +15,9 @@ See [here](https://exoticorn.github.io/microw8/) for more information and docs.
 
 ## Downloads
 
-* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-linux.tgz)
-* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-macos.tgz)
-* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-windows.zip)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-windows.zip)
 
 The download includes
 
@@ -43,6 +43,27 @@ Options:
 -l LEVEL, --level LEVEL : Compression level (0-9). Higher compression levels are really slow.
 -o FILE, --output FILE  : Write the loaded and optionally packed cart back to disk.
 
+when using the native runtime:
+
+-m, --no-audio          : Disable audio, also reduces cpu load a bit
+--no-gpu                : Force old cpu-only window code
+--filter FILTER         : Select an upscale filter at startup
+--fullscreen            : Start in fullscreen mode
+
+Note that the cpu-only window does not support fullscreen nor upscale filters.
+
+Unless --no-gpu is given, uw8 will first try to open a gpu accelerated window, falling back to the old cpu-only window if that fails.
+Therefore you should rarely need to manually pass --no-gpu. If you prefer the old pixel doubling look to the now default crt filter,
+you can just pass "--filter nearest" or "--filter 1".
+
+The upscale filter options are:
+1, nearest              : Anti-aliased nearest filter
+2, fast_crt             : Very simple, cheap crt filter, not very good below a window size of 960x720
+3, ss_crt               : Super sampled crt filter, a little more demanding on the GPU but scales well to smaller window sizes
+4, chromatic_crt        : Variant of fast_crt with a slight offset of the three color dots of a pixel, still pretty cheap
+5, auto_crt (default)   : ss_crt below 960x720, chromatic_crt otherwise
+
+You can switch the upscale filter at any time using the keys 1-5. You can toggle fullscreen with F.
 
 uw8 pack [<options>] <infile> <outfile>
 

examples/curlywas/cracklebass.cwa

@@ -0,0 +1,55 @@
+// port of cracklebass by pestis (originally on TIC-80)
+
+include "../include/microw8-api.cwa"
+
+const MUSIC_DATA = 0x20000;
+
+export fn upd() {
+    let inline t = 32!32 * 6 / 100;
+    let inline p = t / 1024;
+
+    let channel:i32;
+
+    loop channels {
+        let inline e = t * channel?MUSIC_DATA / 8;
+        let lazy pattern = (8 * channel + p)?(MUSIC_DATA + 56);
+        let lazy n = !!pattern * (8 * pattern + e / 16 % 8)?MUSIC_DATA;
+        let inline prev_ctrl = (channel * 6)?80;
+        (channel * 6)?80 = if n {
+            let inline base_note = 12 + 12 * channel?(MUSIC_DATA + 4) + n;
+            let inline pitch_drop = e % 16 * channel?(MUSIC_DATA + 94);
+            let inline key_pattern = p?(MUSIC_DATA + 8*4 + 56);
+            let inline key = select(key_pattern, (8 * key_pattern + t / 128 % 8)?MUSIC_DATA, 1);
+            (channel * 6)?83 = base_note - pitch_drop / 4 + key;
+            prev_ctrl & 0xfc | (e / 8 & 2) | 1
+        } else {
+            prev_ctrl & 0xfe
+        };
+
+        branch_if (channel := channel + 1) < 4: channels;
+    }
+}
+
+data 80 {
+    i8(
+        0x44, 0, 0, 0, 0x50, 0x40,
+        0x4, 0x50, 0, 0, 0x80, 0x80,
+        0x40, 0x80, 0, 0, 0x40, 0x40,
+        0, 0, 0, 0, 0x50, 0x50
+    )
+}
+
+data MUSIC_DATA {
+    i8(
+ 16, 2, 8, 8, 1, 2, 2, 3, 1, 0,
+  1,13,16, 0, 1, 8, 1, 0, 1,13,
+ 16, 1, 1, 8, 1, 0, 8,13,13, 0,
+ 16,13, 1, 0, 1, 0, 1, 0, 1, 1,
+  1, 0, 0, 0, 1, 0,13, 1, 1, 1,
+  6, 8, 1, 1, 6, 8, 1, 1, 2, 1,
+  2, 1, 2, 0, 0, 0, 0, 3, 3, 3,
+  5, 0, 0, 2, 1, 2, 1, 2, 1, 2,
+  0, 4, 4, 0, 4, 4, 4, 4, 0, 0,
+  0, 0, 6, 6, 0, 0, 0, 8
+    )
+}

examples/curlywas/scaled_text.cwa

@@ -0,0 +1,14 @@
+include "../include/microw8-api.cwa"
+
+export fn upd() {
+	printString(USER_MEM);
+}
+
+data USER_MEM {
+	// clear screen, switch to graphics text mode, text scale 4
+	i8(12, 5, 30, 4)
+	// text color, position, print two lines
+	i8(15, 86, 31, 8, 80) "Hello," i8(8, 8, 8, 8, 8, 10) "MicroW8!"
+	// print same two lines with different color and slight offset 
+	i8(15, 47, 31, 10, 82) "Hello," i8(8, 8, 8, 8, 8, 10) "MicroW8!" i8(0)
+}

examples/curlywas/simple_music.cwa

@@ -0,0 +1,38 @@
+include "../include/microw8-api.cwa"
+
+global mut frame = 0;
+
+export fn upd() {
+    if frame % 16 == 0 {
+        let ch: i32;
+        loop channels {
+            playNote(ch, (ch * 32 + (frame / 16) % 32)?0x20000);
+            branch_if ch := (ch + 1) % 4: channels;
+        }
+    }
+    frame = frame + 1;
+}
+
+data 0x20000 {
+    i8(
+        0x4e, 0x0, 0x0, 0x4c, 0x49, 0x0, 0x45, 0x47,
+        0x49, 0x47, 0x45, 0x44, 0x42, 0x0, 0x3d, 0x41,
+        0x44, 0x0, 0x0, 0x47, 0x49, 0x47, 0x45, 0x41,
+        0x44, 0x0, 0x0, 0x0, 0x42, 0x0, 0x0, 0x0,
+
+        0x25, 0, 0x49, 0x25, 0x25, 0, 0x49, 0x38,
+        0x25, 0, 0x49, 0x25, 0x25, 0, 0x49, 0x38,
+        0x25, 0, 0x49, 0x25, 0x25, 0, 0x49, 0x38,
+        0x25, 0, 0x49, 0x25, 0x25, 0, 0x49, 0x38,
+
+        0x2a, 0x0, 0x0, 0x0, 0x2d, 0x0, 0x0, 0x0,
+        0x2c, 0x0, 0x28, 0x0, 0x2a, 0x0, 0x0, 0x0,
+        0x25, 0x0, 0x0, 0x0, 0x29, 0x0, 0x0, 0x0,
+        0x2c, 0x0, 0x2d, 0x0, 0x2a, 0x0, 0x25, 0x0,
+
+        0x0, 0x0, 0x31, 0x0, 0x34, 0x0, 0x0, 0x36,
+        0x38, 0x39, 0x38, 0x34, 0x36, 0x0, 0x0, 0x0,
+        0x0, 0x3d, 0x3b, 0x39, 0x38, 0x0, 0x0, 0x0,
+        0x0, 0x39, 0x38, 0x39, 0x38, 0x0, 0x36, 0x0
+    )
+}

examples/curlywas/skipahead.cwa

@@ -8,7 +8,7 @@ global mut f: f32 = 2.0;
 
 export fn upd() {
     let y: i32;
-    let inline zero = 0.0;
+    let inline zero = 0_f;
 
     let lazy control_speed = 0.03125;
     s = s + 0.1875 - (f + control_speed) * isButtonPressed(4 <| cls(4)) as f32;
@@ -30,6 +30,8 @@ export fn upd() {
 
         if y == 180 & py > zero {
             if x > w | x < zero {
+                0?80 = 0xc3;
+                3?80 = 32;
                 return;
             }
             py = zero;
@@ -43,6 +45,9 @@ export fn upd() {
     circle(160 as f32, 160 as f32 + py, 22 as f32, -28);
     circle((160 - 6) as f32, (160 - 6) as f32 + py, 6 as f32, -26);
 
+    0?86 = py < zero;
+    3?86 = 32 - py as i32;
+
     px = px + (isButtonPressed(3) - isButtonPressed(2)) as f32 * control_speed;
     py = py + s;
     pz = pz + 1;

examples/curlywas/sprites.cwa

@@ -0,0 +1,21 @@
+include "../include/microw8-api.cwa"
+
+const SPRITE = 0x20000;
+
+export fn upd() {
+	cls(0);
+	let t = time() / 2_f;
+	let i: i32;
+	loop spriteLoop {
+		let inline x = sin(t * -1.3 + i as f32 * (3.141 / 30_f)) * 180_f + 160_f;
+		let inline y = sin(t * 1.7 + i as f32 * (3.141 / 40_f)) * 140_f + 120_f;
+		blitSprite(SPRITE, 16, x as i32, y as i32, (i & 3) * 0x200 + 0x100);
+		branch_if (i +:= 1) < 100: spriteLoop;
+	}
+}
+
+start fn start() {
+	printChar('OO');
+	circle(8_f, 8_f, 6_f, 75);
+	grabSprite(SPRITE, 16, 0, 0, 0);
+}

examples/curlywas/steadyon.cwa

@@ -0,0 +1,38 @@
+// Steady On Tim, It's Only A Budget Game
+// by Gasman / Hooy-Program
+// ported to MicroW8 by exoticorn/icebird
+
+include "../include/microw8-api.cwa"
+
+fn melody(t: i32, T: i32) -> i32 {
+    let inline riff_pos = abs(((T&31) - 16) as f32) as i32;
+    let lazy shift = ((1-((T>>5)&3))%2-1) as f32 / 6 as f32;
+
+    let inline note_count = 5 - (T >= 512);
+    let inline octave = (riff_pos/5) as f32;
+    let inline riff_note = 5514 >> (riff_pos % note_count * 4) & 15;
+    let inline melody_freq = pow(2 as f32, shift + octave - (riff_note as f32 / 12 as f32));
+    let inline melody = (t as f32 * melody_freq) as i32 & 128;
+    
+    let inline arp_note = ((0x85>>((t>>12)%3*4)) & 15) - 1;
+    let inline arp_freq = pow(2 as f32, shift + (arp_note as f32 / 12 as f32));
+    let inline arp_vol = (T >= 256) * (12-T%12);
+    let inline arpeggio = ((t as f32 * arp_freq) as i32 & 128) * arp_vol / 12;
+    
+    melody + arpeggio
+}
+
+export fn snd(t: i32) -> f32 {
+    let lazy T = t/10000;
+    
+    let inline mel_arp = melody(t, T)/3 + melody(t, T-3)/5;
+
+    let inline bass_vol = (T >= 128) & (197 >> (T % 8));
+    let inline bass_freq = pow(2 as f32, (((T & 4) * ((T & 7) - 1)) as f32 / 24 as f32 - 5 as f32));
+    let inline bass = ((t as f32 * bass_freq) as i32 & 63) * bass_vol;
+
+    let inline snare_ish = (random() & 31) * (8 - (T + 4) % 8) / 8;
+
+    let inline sample = mel_arp + bass + snare_ish;
+    sample as f32 / 255 as f32
+}

examples/curlywas/tim_ges.cwa

@@ -0,0 +1,64 @@
+// Steady On Tim, It's Only A Budget Game
+// original bytebeat by Gasman / Hooy-Program
+// ported to MicroW8/GES by exoticorn/icebird
+
+import "env.memory" memory(4);
+
+fn melody(ch: i32, t: i32, T: i32) {
+    let lazy riff_pos = abs(((T&31) - 16) as f32) as i32;
+    let lazy shift = ((1-((T>>5)&3))%2-1) * 2;
+
+    let inline note_count = 5 - (T >= 512);
+    let inline octave = (riff_pos/5) * 12;
+    let inline riff_note = 5514 >> (riff_pos % note_count * 4) & 15;
+    let inline melody_note = shift + octave - riff_note;
+
+    ch?1 = 230 - riff_pos * 14;
+    ch?3 = melody_note + 64;
+
+    let inline arp_note = shift + ((0x85>>((t/2)%3*4)) & 15) - 1;
+    80?3 = arp_note + 64;
+}
+
+export fn upd() {
+    let lazy t = 32!32 / (1000/60);
+    let lazy T = t / 7;
+    melody(98, t, T - 3);
+    melody(92, t, T);
+
+    80?0 = ((T >= 256) & (T/12+(T-3)/12)) * 2 | 0x48; // arp trigger
+
+    if T >= 128 {
+        let inline bass_step = T % 8;
+        86?3 = if bass_step / 2 == 2 {
+            86?0 = 0xd6;
+            81
+        } else {
+            86?0 = ((197 >> bass_step) & 1) | 0x48;
+            ((T & 4) * ((T & 7) - 1)) / 2 + 28
+        };
+    }
+}
+
+data 80 {
+    i8(
+        0, 0x90, 0, 0, 0, 0x90,
+        0, 0x4c, 0, 0, 0, 0x4c,
+        0x19, 0, 0, 0, 0, 0x4c,
+        0x19, 0, 0, 0, 0, 0x4c,
+        0xfa, 0x84,
+        0xc1, 0xc1, 0, 107, 0, 0x4c
+    )
+}
+
+/*
+include "../../platform/src/ges.cwa"
+
+import "env.pow" fn pow(f32, f32) -> f32;
+import "env.exp" fn exp(f32) -> f32;
+import "env.sin" fn sin(f32) -> f32;
+
+export fn snd(t: i32) -> f32 {
+    gesSnd(t)
+}
+*/

examples/include/microw8-api.cwa

@@ -30,9 +30,13 @@ import "env.printInt" fn printInt(i32);
 import "env.setTextColor" fn setTextColor(i32);
 import "env.setBackgroundColor" fn setBackgroundColor(i32);
 import "env.setCursorPosition" fn setCursorPosition(i32, i32);
-import "env.rectangle_outline" fn rectangle_outline(f32, f32, f32, f32, i32);
-import "env.circle_outline" fn circle_outline(f32, f32, f32, i32);
+import "env.rectangleOutline" fn rectangleOutline(f32, f32, f32, f32, i32);
+import "env.circleOutline" fn circleOutline(f32, f32, f32, i32);
 import "env.exp" fn exp(f32) -> f32;
+import "env.playNote" fn playNote(i32, i32);
+import "env.sndGes" fn sndGes(i32) -> f32;
+import "env.blitSprite" fn blitSprite(i32, i32, i32, i32, i32);
+import "env.grabSprite" fn grabSprite(i32, i32, i32, i32, i32);
 
 const TIME_MS = 0x40;
 const GAMEPAD = 0x44;

examples/include/microw8-api.wat

@@ -30,9 +30,13 @@
 (import "env" "setTextColor" (func $setTextColor (param i32)))
 (import "env" "setBackgroundColor" (func $setBackgroundColor (param i32)))
 (import "env" "setCursorPosition" (func $setCursorPosition (param i32) (param i32)))
-(import "env" "rectangle_outline" (func $rectangle_outline (param f32) (param f32) (param f32) (param f32) (param i32)))
-(import "env" "circle_outline" (func $circle_outline (param f32) (param f32) (param f32) (param i32)))
+(import "env" "rectangleOutline" (func $rectangleOutline (param f32) (param f32) (param f32) (param f32) (param i32)))
+(import "env" "circleOutline" (func $circleOutline (param f32) (param f32) (param f32) (param i32)))
 (import "env" "exp" (func $exp (param f32) (result f32)))
+(import "env" "playNote" (func $playNote (param i32) (param i32)))
+(import "env" "sndGes" (func $sndGes (param i32) (result f32)))
+(import "env" "blitSprite" (func $blitSprite (param i32) (param i32) (param i32) (param i32) (param i32)))
+(import "env" "grabSprite" (func $grabSprite (param i32) (param i32) (param i32) (param i32) (param i32)))
 
 ;; to use defines, include this file with a preprocessor
 ;; like gpp (https://logological.org/gpp).

examples/wat/tunnel_opt.wat

@@ -0,0 +1,74 @@
+(module
+  (import "env" "atan2" (func $atan2 (param f32 f32) (result f32)))
+  (import "env" "time" (func $time (result f32)))
+  (import "env" "memory" (memory 4))
+  (func (export "upd")
+    (local $y i32)
+    (local $i i32) 
+    (local $x i32)
+
+    (loop $pixels
+      i32.const 1
+      local.get $i
+
+      local.get $i
+
+      i32.const 36928
+      f32.convert_i32_s
+      local.get $i
+      i32.const 320
+      i32.rem_s
+      i32.const 160
+      i32.sub
+      local.tee $x
+      local.get $x
+      i32.mul
+      local.get $i
+      i32.const 320
+      i32.div_s
+      i32.const 120
+      i32.sub
+      local.tee $y
+      local.get $y
+      i32.mul
+      i32.add
+      f32.convert_i32_s
+      f32.sqrt
+      f32.div
+      call $time
+      i32.const 163
+      f32.convert_i32_s
+      f32.mul
+      f32.add
+      i32.trunc_sat_f32_s
+
+      local.get $x
+      f32.convert_i32_s
+      local.get $y
+      f32.convert_i32_s
+      call $atan2
+      i32.const 163
+      f32.convert_i32_s
+      f32.mul
+      call $time
+      i32.const 64
+      f32.convert_i32_s
+      f32.mul
+      f32.add
+      i32.trunc_f32_s
+
+      i32.xor
+      i32.const 4
+      i32.shr_s
+      i32.const 15
+      i32.and
+      i32.store8 offset=120
+
+      i32.add
+      local.tee $i
+      i32.const 76800
+      i32.rem_s
+      br_if $pixels
+  )
+ )
+)

platform/Cargo.lock

@@ -19,47 +19,54 @@ dependencies = [
 
 [[package]]
 name = "ahash"
-version = "0.7.6"
+version = "0.8.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fcb51a0695d8f838b1ee009b3fbf66bda078cd64590202a864a8f3e8c4315c47"
+checksum = "e89da841a80418a9b391ebaea17f5c112ffaaa96f621d2c285b5174da76b9011"
 dependencies = [
- "getrandom",
+ "cfg-if",
  "once_cell",
  "version_check",
+ "zerocopy",
 ]
 
 [[package]]
 name = "anyhow"
-version = "1.0.52"
+version = "1.0.81"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "84450d0b4a8bd1ba4144ce8ce718fbc5d071358b1e5384bace6536b3d1f2d5b3"
+checksum = "0952808a6c2afd1aa8947271f3a60f1a6763c7b912d210184c5149b5cf147247"
 
 [[package]]
 name = "ariadne"
-version = "0.1.3"
+version = "0.1.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7080ae01b2f0c312065d4914cd0f0de045eb8832e9415b355106a6cff3073cb4"
+checksum = "f1cb2a2046bea8ce5e875551f5772024882de0b540c7f93dfc5d6cf1ca8b030c"
 dependencies = [
  "yansi",
 ]
 
 [[package]]
 name = "autocfg"
-version = "1.0.1"
+version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cdb031dd78e28731d87d56cc8ffef4a8f36ca26c38fe2de700543e627f8a464a"
+checksum = "f1fdabc7756949593fe60f30ec81974b613357de856987752631dea1e3394c80"
+
+[[package]]
+name = "bitflags"
+version = "2.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf4b9d6a944f767f8e5e0db018570623c85f3d925ac718db4e06d0187adb21c1"
 
 [[package]]
 name = "bytemuck"
-version = "1.7.3"
+version = "1.15.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "439989e6b8c38d1b6570a384ef1e49c8848128f5a97f3914baef02920842712f"
+checksum = "5d6d68c57235a3a081186990eca2867354726650f42f7516ca50c28d6281fd15"
 
 [[package]]
 name = "cc"
-version = "1.0.72"
+version = "1.0.92"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "22a9137b95ea06864e018375b72adfb7db6e6f68cfc8df5a04d00288050485ee"
+checksum = "2678b2e3449475e95b0aa6f9b506a28e61b3dc8996592b983695e8ebb58a8b41"
 
 [[package]]
 name = "cdivsufsort"
@@ -88,54 +95,47 @@ dependencies = [
 
 [[package]]
 name = "const-random"
-version = "0.1.13"
+version = "0.1.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f590d95d011aa80b063ffe3253422ed5aa462af4e9867d43ce8337562bac77c4"
+checksum = "87e00182fe74b066627d63b85fd550ac2998d4b0bd86bfed477a0ae4c7c71359"
 dependencies = [
  "const-random-macro",
- "proc-macro-hack",
 ]
 
 [[package]]
 name = "const-random-macro"
-version = "0.1.13"
+version = "0.1.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "615f6e27d000a2bffbc7f2f6a8669179378fa27ee4d0a509e985dfc0a7defb40"
+checksum = "f9d839f2a20b0aee515dc581a6172f2321f96cab76c1a38a4c584a194955390e"
 dependencies = [
  "getrandom",
- "lazy_static",
- "proc-macro-hack",
+ "once_cell",
  "tiny-keccak",
 ]
 
 [[package]]
 name = "crc32fast"
-version = "1.3.0"
+version = "1.4.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "738c290dfaea84fc1ca15ad9c168d083b05a714e1efddd8edaab678dc28d2836"
+checksum = "b3855a8a784b474f333699ef2bbca9db2c4a1f6d9088a90a2d25b1eb53111eaa"
 dependencies = [
  "cfg-if",
 ]
 
 [[package]]
 name = "crossbeam-channel"
-version = "0.5.1"
+version = "0.5.12"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "06ed27e177f16d65f0f0c22a213e17c696ace5dd64b14258b52f9417ccb52db4"
+checksum = "ab3db02a9c5b5121e1e42fbdb1aeb65f5e02624cc58c43f2884c6ccac0b82f95"
 dependencies = [
- "cfg-if",
  "crossbeam-utils",
 ]
 
 [[package]]
 name = "crossbeam-utils"
-version = "0.8.5"
+version = "0.8.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d82cfc11ce7f2c3faef78d8a684447b40d503d9681acebed6cb728d45940c4db"
-dependencies = [
- "cfg-if",
- "lazy_static",
-]
+checksum = "248e3bacc7dc6baa3b21e405ee045c3047101a49145e7e9eca583ab4c2ca5345"
 
 [[package]]
 name = "crunchy"
@@ -146,42 +146,52 @@ checksum = "7a81dae078cea95a014a339291cec439d2f232ebe854a9d672b796c6afafa9b7"
 [[package]]
 name = "curlywas"
 version = "0.1.0"
-source = "git+https://github.com/exoticorn/curlywas.git?rev=89638565#896385654ab2c089200920b6dea4abec641c88d6"
+source = "git+https://github.com/exoticorn/curlywas.git?rev=0e7ea50#0e7ea508cd0e76836283ae68a44c9097df83c8ac"
 dependencies = [
  "anyhow",
  "ariadne",
  "chumsky",
- "pico-args",
+ "pico-args 0.4.2",
  "wasm-encoder 0.10.0",
  "wasmparser 0.83.0",
 ]
 
 [[package]]
-name = "fallible_collections"
-version = "0.4.4"
+name = "equivalent"
+version = "1.0.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "52db5973b6a19247baf19b30f41c23a1bfffc2e9ce0a5db2f60e3cd5dc8895f7"
+checksum = "5443807d6dff69373d433ab9ef5378ad8df50ca6298caf15de6e52e24aaf54d5"
+
+[[package]]
+name = "fallible-iterator"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
+
+[[package]]
+name = "fallible_collections"
+version = "0.4.9"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a88c69768c0a15262df21899142bc6df9b9b823546d4b4b9a7bc2d6c448ec6fd"
 dependencies = [
- "hashbrown",
+ "hashbrown 0.13.2",
 ]
 
 [[package]]
 name = "flate2"
-version = "1.0.22"
+version = "1.0.28"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1e6988e897c1c9c485f43b47a529cef42fde0547f9d8d41a7062518f1d8fc53f"
+checksum = "46303f565772937ffe1d394a4fac6f411c6013172fadde9dcdb1e147a086940e"
 dependencies = [
- "cfg-if",
  "crc32fast",
- "libc",
  "miniz_oxide",
 ]
 
 [[package]]
 name = "getrandom"
-version = "0.2.3"
+version = "0.2.14"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7fcd999463524c52659517fe2cea98493cfe485d10565e7b0fb07dbba7ad2753"
+checksum = "94b22e06ecb0110981051723910cbf0b5f5e09a2062dd7663334ee79a9d1286c"
 dependencies = [
  "cfg-if",
  "libc",
@@ -189,14 +199,37 @@ dependencies = [
 ]
 
 [[package]]
-name = "hashbrown"
-version = "0.11.2"
+name = "gimli"
+version = "0.26.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ab5ef0d4909ef3724cc8cce6ccc8572c5c817592e9285f5464f8e86f8bd3726e"
+checksum = "22030e2c5a68ec659fde1e949a745124b48e6fa8b045b7ed5bd1fe4ccc5c4e5d"
 dependencies = [
- "ahash 0.7.6",
+ "fallible-iterator",
+ "indexmap 1.9.3",
+ "stable_deref_trait",
 ]
 
+[[package]]
+name = "hashbrown"
+version = "0.12.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
+
+[[package]]
+name = "hashbrown"
+version = "0.13.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "43a3c133739dddd0d2990f9a4bdf8eb4b21ef50e4851ca85ab661199821d510e"
+dependencies = [
+ "ahash 0.8.11",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.14.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "290f1a1d9242c78d09ce40a5e87e7554ee637af1351968159f4952f028f75604"
+
 [[package]]
 name = "heck"
 version = "0.3.3"
@@ -213,10 +246,24 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "25a2bc672d1148e28034f176e01fffebb08b35768468cc954630da77a1449005"
 
 [[package]]
-name = "lazy_static"
-version = "1.4.0"
+name = "indexmap"
+version = "1.9.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
+checksum = "bd070e393353796e801d209ad339e89596eb4c8d430d18ede6a1cced8fafbd99"
+dependencies = [
+ "autocfg",
+ "hashbrown 0.12.3",
+]
+
+[[package]]
+name = "indexmap"
+version = "2.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "168fb715dda47215e360912c096649d23d58bf392ac62f73919e831745e40f26"
+dependencies = [
+ "equivalent",
+ "hashbrown 0.14.3",
+]
 
 [[package]]
 name = "leb128"
@@ -225,17 +272,24 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
 
 [[package]]
-name = "libc"
-version = "0.2.112"
+name = "lexopt"
+version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1b03d17f364a3a042d5e5d46b053bbbf82c92c9430c592dd4c064dc6ee997125"
+checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
+
+[[package]]
+name = "libc"
+version = "0.2.153"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9c198f91728a82281a64e1f4f9eeb25d82cb32a5de251c6bd1b5154d63a8e7bd"
 
 [[package]]
 name = "lodepng"
-version = "3.4.7"
+version = "3.10.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "24844d5c0b922ddd52fb5bf0964a4c7f8e799a946ec01bb463771eb04fc1a323"
+checksum = "a42d298694b14401847de29abd44adf278b42e989e516deac7b72018400002d8"
 dependencies = [
+ "crc32fast",
  "fallible_collections",
  "flate2",
  "libc",
@@ -244,47 +298,42 @@ dependencies = [
 
 [[package]]
 name = "log"
-version = "0.4.14"
+version = "0.4.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "51b9bbe6c47d51fc3e1a9b945965946b4c44142ab8792c50835a980d362c2710"
-dependencies = [
- "cfg-if",
-]
+checksum = "90ed8c1e510134f979dbc4f070f87d4313098b704861a105fe34231c70a3901c"
 
 [[package]]
 name = "miniz_oxide"
-version = "0.4.4"
+version = "0.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a92518e98c078586bc6c934028adcca4c92a53d6a958196de835170a01d84e4b"
+checksum = "9d811f3e15f28568be3407c8e7fdb6514c1cda3cb30683f15b6a1a1dc4ea14a7"
 dependencies = [
  "adler",
- "autocfg",
 ]
 
 [[package]]
 name = "num-traits"
-version = "0.2.14"
+version = "0.2.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9a64b1ec5cda2586e284722486d802acf1f7dbdc623e2bfc57e65ca1cd099290"
+checksum = "da0df0e5185db44f69b44f26786fe401b6c293d1907744beaa7fa62b2e5a517a"
 dependencies = [
  "autocfg",
 ]
 
 [[package]]
 name = "once_cell"
-version = "1.9.0"
+version = "1.19.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "da32515d9f6e6e489d7bc9d84c71b060db7247dc035bbe44eac88cf87486d8d5"
+checksum = "3fdb12b2476b595f9358c5161aa467c2438859caa136dec86c26fdd2efe17b92"
 
 [[package]]
 name = "pbr"
-version = "1.0.4"
+version = "1.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ff5751d87f7c00ae6403eb1fcbba229b9c76c9a30de8c1cf87182177b168cea2"
+checksum = "ed5827dfa0d69b6c92493d6c38e633bbaa5937c153d0d7c28bf12313f8c6d514"
 dependencies = [
  "crossbeam-channel",
  "libc",
- "time",
  "winapi",
 ]
 
@@ -294,6 +343,12 @@ version = "0.4.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "db8bcd96cb740d03149cbad5518db9fd87126a10ab519c011893b1754134c468"
 
+[[package]]
+name = "pico-args"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5be167a7af36ee22fe3115051bc51f6e6c7054c9348e28deb4f49bd6f705a315"
+
 [[package]]
 name = "platform"
 version = "0.1.0"
@@ -304,35 +359,29 @@ dependencies = [
  "uw8-tool",
 ]
 
-[[package]]
-name = "proc-macro-hack"
-version = "0.5.19"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "dbf0c48bc1d91375ae5c3cd81e3722dff1abcf81a30960240640d223f59fe0e5"
-
 [[package]]
 name = "proc-macro2"
-version = "1.0.36"
+version = "1.0.79"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c7342d5883fbccae1cc37a2353b09c87c9b0f3afd73f5fb9bba687a1f733b029"
+checksum = "e835ff2298f5721608eb1a980ecaee1aef2c132bf95ecc026a11b7bf3c01c02e"
 dependencies = [
- "unicode-xid",
+ "unicode-ident",
 ]
 
 [[package]]
 name = "quote"
-version = "1.0.15"
+version = "1.0.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "864d3e96a899863136fc6e99f3d7cae289dafe43bf2c5ac19b70df7210c0a145"
+checksum = "291ec9ab5efd934aaf503a6466c5d5251535d108ee747472c3977cc5acc868ef"
 dependencies = [
  "proc-macro2",
 ]
 
 [[package]]
 name = "rgb"
-version = "0.8.31"
+version = "0.8.37"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9a374af9a0e5fdcdd98c1c7b64f05004f9ea2555b6c75f211daa81268a3c50f1"
+checksum = "05aaa8004b64fd573fc9d002f4e632d51ad4f026c2b5ba95fcb6c2f32c2c47d8"
 dependencies = [
  "bytemuck",
 ]
@@ -347,24 +396,57 @@ dependencies = [
 ]
 
 [[package]]
-name = "syn"
-version = "1.0.86"
+name = "semver"
+version = "1.0.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8a65b3f4ffa0092e9887669db0eae07941f023991ab58ea44da8fe8e2d511c6b"
+checksum = "92d43fe69e652f3df9bdc2b85b2854a0825b86e4fb76bc44d945137d053639ca"
+
+[[package]]
+name = "stable_deref_trait"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a8f112729512f8e442d81f95a8a7ddf2b7c6b8a1a6f509a95864142b30cab2d3"
+
+[[package]]
+name = "syn"
+version = "1.0.109"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237"
 dependencies = [
  "proc-macro2",
  "quote",
- "unicode-xid",
+ "unicode-ident",
 ]
 
 [[package]]
-name = "time"
-version = "0.1.43"
+name = "syn"
+version = "2.0.58"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ca8a50ef2360fbd1eeb0ecd46795a87a19024eb4b53c5dc916ca1fd95fe62438"
+checksum = "44cfb93f38070beee36b3fef7d4f5a16f27751d94b187b666a5cc5e9b0d30687"
 dependencies = [
- "libc",
- "winapi",
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
+]
+
+[[package]]
+name = "thiserror"
+version = "1.0.58"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "03468839009160513471e86a034bb2c5c0e4baae3b43f79ffc55c4a5427b3297"
+dependencies = [
+ "thiserror-impl",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.58"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c61f3ba182994efc43764a46c018c347bc492c79f024e705f46567b418f6d4f7"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.58",
 ]
 
 [[package]]
@@ -377,26 +459,26 @@ dependencies = [
 ]
 
 [[package]]
-name = "unicode-segmentation"
-version = "1.9.0"
+name = "unicode-ident"
+version = "1.0.12"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7e8820f5d777f6224dc4be3632222971ac30164d4a258d595640799554ebfd99"
+checksum = "3354b9ac3fae1ff6755cb6db53683adb661634f67557942dea4facebec0fee4b"
 
 [[package]]
-name = "unicode-xid"
-version = "0.2.2"
+name = "unicode-segmentation"
+version = "1.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3"
+checksum = "d4c87d22b6e3f4a18d4d40ef354e97c90fcb14dd91d7dc0aa9d8a1172ebf7202"
 
 [[package]]
 name = "upkr"
-version = "0.1.0"
-source = "git+https://github.com/exoticorn/upkr.git?rev=2e7983fc#2e7983fc650788d98da2eecef2d16f63e849e4a0"
+version = "0.2.1"
+source = "git+https://github.com/exoticorn/upkr.git?rev=080db40d0088bbee2bdf3c5c75288ac7853d6b7a#080db40d0088bbee2bdf3c5c75288ac7853d6b7a"
 dependencies = [
  "anyhow",
  "cdivsufsort",
- "pbr",
- "pico-args",
+ "lexopt",
+ "thiserror",
 ]
 
 [[package]]
@@ -405,31 +487,33 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "pbr",
- "pico-args",
+ "pico-args 0.5.0",
  "upkr",
  "walrus",
- "wasm-encoder 0.8.0",
- "wasmparser 0.81.0",
+ "wasm-encoder 0.201.0",
+ "wasmparser 0.201.0",
 ]
 
 [[package]]
 name = "version_check"
-version = "0.9.3"
+version = "0.9.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5fecdca9a5291cc2b8dcf7dc02453fee791a280f3743cb0905f8822ae463b3fe"
+checksum = "49874b5167b65d7193b8aba1567f5c7d93d001cafc34600cee003eda787e483f"
 
 [[package]]
 name = "walrus"
-version = "0.19.0"
+version = "0.20.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4eb08e48cde54c05f363d984bb54ce374f49e242def9468d2e1b6c2372d291f8"
+checksum = "2c03529cd0c4400a2449f640d2f27cd1b48c3065226d15e26d98e4429ab0adb7"
 dependencies = [
  "anyhow",
+ "gimli",
  "id-arena",
  "leb128",
  "log",
  "walrus-macro",
- "wasmparser 0.77.0",
+ "wasm-encoder 0.29.0",
+ "wasmparser 0.80.2",
 ]
 
 [[package]]
@@ -441,23 +525,14 @@ dependencies = [
  "heck",
  "proc-macro2",
  "quote",
- "syn",
+ "syn 1.0.109",
 ]
 
 [[package]]
 name = "wasi"
-version = "0.10.2+wasi-snapshot-preview1"
+version = "0.11.0+wasi-snapshot-preview1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6"
-
-[[package]]
-name = "wasm-encoder"
-version = "0.8.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db0c351632e46cc06a58a696a6c11e4cf90cad4b9f8f07a0b59128d616c29bb0"
-dependencies = [
- "leb128",
-]
+checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423"
 
 [[package]]
 name = "wasm-encoder"
@@ -469,16 +544,28 @@ dependencies = [
 ]
 
 [[package]]
-name = "wasmparser"
-version = "0.77.0"
+name = "wasm-encoder"
+version = "0.29.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b35c86d22e720a07d954ebbed772d01180501afe7d03d464f413bb5f8914a8d6"
+checksum = "18c41dbd92eaebf3612a39be316540b8377c871cb9bde6b064af962984912881"
+dependencies = [
+ "leb128",
+]
+
+[[package]]
+name = "wasm-encoder"
+version = "0.201.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b9c7d2731df60006819b013f64ccc2019691deccf6e11a1804bc850cd6748f1a"
+dependencies = [
+ "leb128",
+]
 
 [[package]]
 name = "wasmparser"
-version = "0.81.0"
+version = "0.80.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98930446519f63d00a836efdc22f67766ceae8dbcc1571379f2bcabc6b2b9abc"
+checksum = "449167e2832691a1bff24cde28d2804e90e09586a448c8e76984792c44334a6b"
 
 [[package]]
 name = "wasmparser"
@@ -486,6 +573,17 @@ version = "0.83.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "718ed7c55c2add6548cca3ddd6383d738cd73b892df400e96b9aa876f0141d7a"
 
+[[package]]
+name = "wasmparser"
+version = "0.201.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "84e5df6dba6c0d7fafc63a450f1738451ed7a0b52295d83e868218fa286bf708"
+dependencies = [
+ "bitflags",
+ "indexmap 2.2.6",
+ "semver",
+]
+
 [[package]]
 name = "winapi"
 version = "0.3.9"
@@ -510,6 +608,26 @@ checksum = "712e227841d057c1ee1cd2fb22fa7e5a5461ae8e48fa2ca79ec42cfc1931183f"
 
 [[package]]
 name = "yansi"
-version = "0.5.0"
+version = "0.5.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9fc79f4a1e39857fc00c3f662cbf2651c771f00e9c15fe2abc341806bd46bd71"
+checksum = "09041cd90cf85f7f8b2df60c646f853b7f535ce68f85244eb6731cf89fa498ec"
+
+[[package]]
+name = "zerocopy"
+version = "0.7.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "74d4d3961e53fa4c9a25a8637fc2bfaf2595b3d3ae34875568a5cf64787716be"
+dependencies = [
+ "zerocopy-derive",
+]
+
+[[package]]
+name = "zerocopy-derive"
+version = "0.7.32"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ce1b18ccd8e73a9321186f97e46f9f04b778851177567b1975109d26a08d2a6"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.58",
+]

platform/Cargo.toml

@@ -6,7 +6,7 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-curlywas = { git="https://github.com/exoticorn/curlywas.git", rev="89638565" }
+curlywas = { git="https://github.com/exoticorn/curlywas.git", rev="0e7ea50" }
 uw8-tool = { path="../uw8-tool" }
 anyhow = "1"
-lodepng = "3.4"
+lodepng = "3.7.2"

platform/src/ges.cwa

@@ -0,0 +1,221 @@
+const GesChannelState.Trigger = 0;
+const GesChannelState.EnvState = 1;
+const GesChannelState.EnvVol = 2;
+const GesChannelState.Phase = 4;
+const GesChannelState.Size = 8;
+
+const GesState.Filter = GesChannelState.Size * 4;
+const GesState.Size = GesState.Filter + 8*4;
+
+const GesStateOffset = 32;
+const GesBufferOffset = 32 + GesState.Size;
+
+export fn sndGes(t: i32) -> f32 {
+    let baseAddr = 0!0x12c78;
+    if !(t & 127) {
+        let i: i32;
+        loop clearLoop {
+            (baseAddr + i)!GesBufferOffset = 0;
+            branch_if (i := i + 4) < 128*4: clearLoop;
+        }
+
+        let ch: i32;
+        loop channelLoop {
+            let lazy channelState = baseAddr + GesStateOffset + ch * GesChannelState.Size;
+            let lazy channelReg = baseAddr + ch * 6;
+            let envState = channelState?GesChannelState.EnvState;
+            let envVol = i32.load16_u(channelState, GesChannelState.EnvVol);
+
+            let lazy oldTrigger = channelState?GesChannelState.Trigger;
+            let lazy ctrl = channelReg?0;
+            if (oldTrigger ^ ctrl) & (ctrl | 2) & 3 {
+                envState = 1;
+                envVol = 0;
+            }
+            channelState?GesChannelState.Trigger = ctrl;
+
+            if envState {
+                let lazy attack = channelReg?4 & 15;
+                envVol = envVol + 12 * pow(1.675, (15 - attack) as f32) as i32;
+                if envVol >= 65535 {
+                    envVol = 65535;
+                    envState = 0;
+                }
+            } else {
+                let inline decay = (channelReg - (ctrl & 1))?5 >> 4;
+                let inline dec = 8 * pow(1.5625, (15 - decay) as f32) as i32;
+                envVol = envVol - ((dec * (envVol + 8192)) >> 16);
+                let inline sustain = (channelReg?5 & 15) << 12;
+                let lazy targetVol = (ctrl & 1) * sustain;
+                if envVol < targetVol {
+                    envVol = targetVol;
+                }
+            }
+            channelState?GesChannelState.EnvState = envState;
+
+            i32.store16(envVol, channelState, GesChannelState.EnvVol);
+
+            let inline note = i32.load16_u(channelReg, 2);
+            let lazy freq = 440 as f32 * pow(2.0, (note - 69*256) as f32 / (12*256) as f32);
+            let phaseInc = (freq * (65536.0 / 44100.0)) as i32;
+
+            let phase = channelState!GesChannelState.Phase;
+
+            let inline pulseWidth = channelReg?1;
+            let invPhaseInc = 1 as f32 / phaseInc as f32;
+
+            i = 0;
+            let wave = ctrl >> 6;
+            if wave < 2 {
+                if wave {
+                    let pulsePhase1 = pulseWidth << 23;
+                    let pulsePhase2 = (511 - pulseWidth) << 23;
+                    loop sawLoop {
+                        let p = (phase ^ 32768) << 16;
+                        let saw = (p >> 16) - polyBlep(phase, invPhaseInc, -32767);
+                        let saw2 = select(p #>= pulsePhase1 & p #< pulsePhase2, -saw, saw);
+                        let saw2 = saw2 -
+                            polyBlep((p - pulsePhase1) >> 16, invPhaseInc, -saw) -
+                            polyBlep((p - pulsePhase2) >> 16, invPhaseInc, saw);
+                        (baseAddr + i)!(GesBufferOffset + 128*4) = saw2;
+                        phase = phase + phaseInc;
+                        branch_if (i := i + 4) < 64*4: sawLoop;
+                    }
+                }
+                else
+                {
+                    let pulsePhase = 32768 + pulseWidth * 128;
+                    loop rectLoop {
+                        (baseAddr + i)!(GesBufferOffset + 128*4) = select((phase & 65535) < pulsePhase, -32768, 32767) -
+                            polyBlep(phase, invPhaseInc, -32767) -
+                            polyBlep(phase - pulsePhase, invPhaseInc, 32767);
+                        phase = phase + phaseInc;
+                        branch_if (i := i + 4) < 64*4: rectLoop;
+                    }
+                }
+            } else {
+                if wave == 2 {
+                    let scale = pulseWidth + 256;
+                    loop triLoop {
+                        let s = phase << 16;
+                        s = (s ^ (s >> 31));
+                        s = (s >> 8) * scale;
+                        s = (s ^ (s >> 31));
+                        (baseAddr + i)!(GesBufferOffset + 128*4) = (s >> 15) - 32768;
+                        phase = phase + phaseInc;
+                        branch_if (i := i + 4) < 64*4: triLoop;
+                    }
+                } else {
+                    loop noiseLoop {
+                        let s = phase >> 12;
+                        let inline pulse = ((phase >> 8) & 255) >= pulseWidth;
+                        s = s * 0x6746ba73;
+                        s = s ^ (s >> 15) * pulse;
+                        (baseAddr + i)!(GesBufferOffset + 128*4) = (s * 0x835776c7) >> 16;
+                        phase = phase + phaseInc;
+                        branch_if (i := i + 4) < 64*4: noiseLoop;
+                    }
+                }
+            }
+
+            channelState!GesChannelState.Phase = phase;
+
+            if ctrl & 32 {
+                let lazy modSrc = (ch - 1) & 3;
+                let inline channelState = baseAddr + GesStateOffset + modSrc * GesChannelState.Size;
+                let inline channelReg = baseAddr + modSrc * 6;
+
+                let inline note = i32.load16_u(channelReg, 2);
+                let inline freq = 440 as f32 * pow(2.0, (note - 69*256) as f32 / (12*256) as f32);
+                let phaseInc = (freq * (65536.0 / 44100.0)) as i32;
+
+                let phase = channelState!GesChannelState.Phase;
+                if modSrc < ch {
+                    phase = phase - (phaseInc << 6);
+                }
+
+                i = 0;
+                loop ringLoop {
+                    let s = phase << 16;
+                    s = (s ^ (s >> 31));
+                    (baseAddr + i)!(GesBufferOffset + 128*4) = ((baseAddr + i)!(GesBufferOffset + 128*4) * ((s >> 15) - 32768)) >> 15;
+                    phase = phase + phaseInc;
+                    branch_if (i := i + 4) < 64*4: ringLoop;
+                }
+            }
+
+            let channelVol = ((baseAddr + (ch >> 1))?24 >> ((ch & 1) * 4)) & 15;
+            envVol = envVol * channelVol / 15;
+
+            let leftVol = (select(ctrl & 16, 0x3d5b, 0x6a79) >> (ch * 4)) & 15;
+            let rightVol = 16 - leftVol;
+
+            let lazy filter = (ctrl >> 2) & 3;
+
+            i = 0;
+            if filter < 2 {
+                if filter {
+                    let f = (4096 as f32 - min(4096 as f32, 4096 as f32 * exp(freq * (-8.0 * 3.141 / 44100.0)))) as i32;
+                    let low = (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter);
+                    loop filterLoop {
+                        let in = ((baseAddr + i)!(GesBufferOffset + 128*4) * envVol) >> 18;
+                        low = low + (((in - low) * f) >> 12);
+                        (baseAddr + i * 2)!GesBufferOffset = (baseAddr + i * 2)!GesBufferOffset + ((low * leftVol) >> 4);
+                        (baseAddr + i * 2)!(GesBufferOffset + 4) = (baseAddr + i * 2)!(GesBufferOffset + 4) + ((low * rightVol) >> 4);
+                        branch_if (i := i + 4) < 64*4: filterLoop;
+                        (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter) = low;
+                        (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter + 4) = 0;
+                    }
+                } else {
+                    loop mixLoop {
+                        let sample = ((baseAddr + i)!(GesBufferOffset + 128*4) * envVol) >> 18;
+                        (baseAddr + i * 2)!GesBufferOffset = (baseAddr + i * 2)!GesBufferOffset + ((sample * leftVol) >> 4);
+                        (baseAddr + i * 2)!(GesBufferOffset + 4) = (baseAddr + i * 2)!(GesBufferOffset + 4) + ((sample * rightVol) >> 4);
+                        branch_if (i := i + 4) < 64*4: mixLoop;
+                        (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter) = sample;
+                        (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter + 4) = 0;
+                    }
+                }
+            } else {
+                filter = filter - 2;
+                let ctrl = (baseAddr + filter)?26;
+                let note = i32.load16_u(baseAddr + filter * 2, 28);
+                let inline freq = 440 as f32 * pow(2.0, (note - 69*256) as f32 / (12*256) as f32);
+                let F = (8192 as f32 * sin(min(0.25, freq / 44100 as f32) * 3.1415)) as i32;
+                let Q = 8192 - (ctrl >> 4) * (7000/15);
+                let Qlimit = (8192*4096/F - F/2) * 3 / 4;
+                if Q > Qlimit {
+                    Q = Qlimit;
+                }
+                let low_out = ctrl & 1;
+                let high_out = (ctrl >> 1) & 1;
+                let band_out = (ctrl >> 2) & 1;
+                let low = (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter);
+                let band = (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter + 4);
+                loop filterLoop {
+                    let in = ((baseAddr + i)!(GesBufferOffset + 128*4) * envVol) >> 18;
+
+                    let high = in - low - ((band * Q) >> 12);
+                    band = band + ((F * high) >> 12);
+                    low = low + ((F * band) >> 12);
+
+                    let sample = low * low_out + high * high_out + band * band_out;
+                    (baseAddr + i * 2)!GesBufferOffset = (baseAddr + i * 2)!GesBufferOffset + ((sample * leftVol) >> 4);
+                    (baseAddr + i * 2)!(GesBufferOffset + 4) = (baseAddr + i * 2)!(GesBufferOffset + 4) + ((sample * rightVol) >> 4);
+                    branch_if (i := i + 4) < 64*4: filterLoop;
+                    (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter) = low;
+                    (baseAddr + ch * 8)!(GesStateOffset + GesState.Filter + 4) = band;
+                }
+            }
+
+            branch_if (ch := ch + 1) < 4: channelLoop;
+        }
+    }
+    ((baseAddr + (t & 127) * 4)!GesBufferOffset) as f32 / 32768 as f32
+}
+
+fn polyBlep(transientPhase: i32, invPhaseInc: f32, magnitude: i32) -> i32 {
+    let lazy t = ((transientPhase << 16) >> 16) as f32 * invPhaseInc;
+    let lazy x = max(0 as f32, 1 as f32 - abs(t));
+    (f32.copysign(x * x, t) * magnitude as f32) as i32
+}

platform/src/ges_only.cwa

@@ -0,0 +1,6 @@
+import "env.memory" memory(1);
+import "env.sin" fn sin(f32) -> f32;
+import "env.pow" fn pow(f32, f32) -> f32;
+import "env.exp" fn exp(f32) -> f32;
+
+include "ges.cwa"

platform/src/main.rs

@@ -11,16 +11,16 @@ fn main() -> Result<()> {
     convert_font()?;
 
     println!("Compiling loader module");
-    let loader = curlywas::compile_file("src/loader.cwa", curlywas::Options::default())?;
-    File::create("bin/loader.wasm")?.write_all(&loader.wasm)?;
+    let loader = curlywas::compile_file("src/loader.cwa", curlywas::Options::default()).0?;
+    File::create("bin/loader.wasm")?.write_all(&loader)?;
 
-    println!("Loader (including base module): {} bytes", loader.wasm.len());
+    println!("Loader (including base module): {} bytes", loader.len());
 
     println!("Compiling platform module");
-    let platform = curlywas::compile_file("src/platform.cwa", curlywas::Options::default())?;
+    let platform = curlywas::compile_file("src/platform.cwa", curlywas::Options::default()).0?;
     println!("Compressing platform module");
     let platform = uw8_tool::pack(
-        &platform.wasm,
+        &platform,
         &uw8_tool::PackConfig::default().with_compression_level(4),
     )?;
     File::create("bin/platform.uw8")?.write_all(&platform)?;

platform/src/platform.cwa

@@ -1,6 +1,10 @@
 import "env.memory" memory(4);
 
+import "env.sin" fn sin(f32) -> f32;
 import "env.cos" fn cos(f32) -> f32;
+import "env.pow" fn pow(f32, f32) -> f32;
+import "env.exp" fn exp(f32) -> f32;
+import "env.logChar" fn logChar(i32);
 
 export fn time() -> f32 {
     (0!64) as f32 / 1000 as f32
@@ -29,11 +33,9 @@ export fn random() -> i32 {
 }
 
 export fn random64() -> i64 {
-    let state: i64;
-    randomState = (state := (
-                          state := randomState ^ (randomState #>> 12i64)
-                      ) ^ (state << 25i64)
-                  ) ^ (state #>> 27i64);
+    let lazy state = randomState ^ (randomState #>> 12i64);
+    let lazy state = state ^ (state << 25i64);
+    randomState = state ^ (state #>> 27i64);
     randomState * 0x2545f4914f6cdd1di64
 }
 
@@ -59,12 +61,9 @@ export fn cls(col: i32) {
     let i: i32;
     textCursorX = 0;
     textCursorY = 0;
-    graphicsText = 0;
-    col = (col & 255) * 0x1010101;
-    loop pixels {
-        i!120 = col;
-        branch_if (i := i + 4) < 320*240: pixels;
-    }
+    outputChannel = 0;
+    textScale = 1;
+    memory.fill(120, col, 320*240);
 }
 
 export fn setPixel(x: i32, y: i32, col: i32) {
@@ -88,12 +87,70 @@ fn clamp(v: i32, min: i32, max: i32) -> i32 {
 export fn hline(x1: i32, x2: i32, y: i32, col: i32) {
     x1 = clamp(x1, 0, 320);
     x2 = clamp(x2, 0, 320);
-    if x1 < x2 & y #< 240 {
+    if  y #>= 240 {
+        return;
+    }
+    let word_start = (x1 + 3) & -4;
+    let word_end = x2 & -4;
+    if word_end > word_start {
+        col = (col & 255) * 0x1010101;
+        let ptr = y * 320 + x1;
+        let end = ptr + word_start - x1;
+        if ptr + 2 <= end {
+            ptr?120 = col;
+            ptr?121 = col;
+            ptr += 2;
+        }
+        if ptr < end {
+            ptr?120 = col;
+            ptr += 1;
+        }
+        end += word_end - word_start;
+        loop words {
+            if ptr + 16 <= end {
+                ptr!120 = col;
+                ptr!124 = col;
+                ptr!128 = col;
+                ptr!132 = col;
+                ptr += 16;
+                branch words;
+            }
+            if ptr + 8 <= end {
+                ptr!120 = col;
+                ptr!124 = col;
+                ptr += 8;
+            }
+            if ptr < end {
+                ptr!120 = col;
+                ptr += 4;
+            }
+        }
+        end += x2 - word_end;
+        if ptr + 2 <= end {
+            ptr?120 = col;
+            ptr?121 = col;
+            ptr += 2;
+        }
+        if ptr < end {
+            ptr?120 = col;
+        }
+    } else {
         let ptr = y * 320 + x1;
         let end = ptr + x2 - x1;
-        loop pixels {
+        if ptr + 4 <= end {
+            ptr?120 = col;
+            ptr?121 = col;
+            ptr?122 = col;
+            ptr?123 = col;
+            ptr += 4;
+        }
+        if ptr + 2 <= end {
+            ptr?120 = col;
+            ptr?121 = col;
+            ptr += 2;
+        }
+        if ptr < end {
             ptr?120 = col;
-            branch_if (ptr := ptr + 1) < end: pixels;
         }
     }
 }
@@ -115,7 +172,7 @@ export fn rectangle(x: f32, y: f32, w: f32, h: f32, col: i32) {
     }
 }
 
-export fn rectangle_outline(x: f32, y: f32, w: f32, h: f32, col: i32) {
+export fn rectangleOutline(x: f32, y: f32, w: f32, h: f32, col: i32) {
     let xl = nearest(x) as i32;
     let xr = nearest(x + w) as i32;
     let yt = nearest(y) as i32;
@@ -156,7 +213,7 @@ export fn circle(cx: f32, cy: f32, radius: f32, col: i32) {
     }
 }
 
-export fn circle_outline(cx: f32, cy: f32, radius: f32, col: i32) {
+export fn circleOutline(cx: f32, cy: f32, radius: f32, col: i32) {
     let prev_w: f32;
     let y = clamp(nearest(cy - radius) as i32, -1, 241);
     let maxY = clamp(nearest(cy + radius) as i32, -1, 241);
@@ -296,6 +353,92 @@ export fn line(x1: f32, y1: f32, x2: f32, y2: f32, col: i32) {
     setPixel(i32.trunc_sat_f32_s(x1 + f * dx), i32.trunc_sat_f32_s(y1 + f * dy), col);
 }
 
+export fn blitSprite(sprite: i32, size: i32, x: i32, y: i32, control: i32) {
+    let lazy width = size & 65535;
+    let lazy height = select(size >> 16, size >> 16, width);
+
+    let lazy x0 = select(x < 0, -x, 0);
+    let lazy x1 = select(x + width > 320, 320 - x, width);
+    let lazy y0 = select(y < 0, -y, 0);
+    let lazy y1 = select(y + height > 240, 240 - y, height);
+
+    let lazy numRows = y1 - y0;
+    let lazy numCols = x1 - x0;
+    if numRows <= 0 | numCols <= 0 {
+        return;
+    }
+
+    let trans = (control & 511) - 256;
+    let lazy flip_x = 1 - ((control >> 8) & 2);
+    let lazy flip_y = 1 - ((control >> 9) & 2);
+    if flip_x < 0 {
+        sprite += width - 1;
+    }
+    if flip_y < 0 {
+        sprite += (height - 1) * width;
+    }
+
+    let spriteRow = sprite + x0 * flip_x + y0 * flip_y * width;
+    let screenRow = x + x0 + (y + y0) * 320;
+
+    loop yloop {
+        let lx = 0;
+        loop xloop {
+            let lazy col = (spriteRow + lx * flip_x)?0;
+            if col != trans {
+                (screenRow + lx)?120 = col;
+            }
+            branch_if (lx +:= 1) < numCols: xloop;
+        }
+        spriteRow += width * flip_y;
+        screenRow += 320;
+        branch_if numRows -:= 1: yloop;
+    }
+}
+
+export fn grabSprite(sprite: i32, size: i32, x: i32, y: i32, control: i32) {
+    let lazy width = size & 65535;
+    let lazy height = select(size >> 16, size >> 16, width);
+
+    let lazy x0 = select(x < 0, -x, 0);
+    let lazy x1 = select(x + width > 320, 320 - x, width);
+    let lazy y0 = select(y < 0, -y, 0);
+    let lazy y1 = select(y + height > 240, 240 - y, height);
+
+    let lazy numRows = y1 - y0;
+    let lazy numCols = x1 - x0;
+    if numRows <= 0 | numCols <= 0 {
+        return;
+    }
+
+    let trans = (control & 511) - 256;
+    let lazy flip_x = 1 - ((control >> 8) & 2);
+    let lazy flip_y = 1 - ((control >> 9) & 2);
+    if flip_x < 0 {
+        sprite += width - 1;
+    }
+    if flip_y < 0 {
+        sprite += (height - 1) * width;
+    }
+
+    let spriteRow = sprite + x0 * flip_x + y0 * flip_y * width;
+    let screenRow = x + x0 + (y + y0) * 320;
+
+    loop yloop {
+        let lx = 0;
+        loop xloop {
+            let lazy col = (screenRow + lx)?120;
+            if col != trans {
+                (spriteRow + lx * flip_x)?0 = col;
+            }
+            branch_if (lx +:= 1) < numCols: xloop;
+        }
+        spriteRow += width * flip_y;
+        screenRow += 320;
+        branch_if numRows -:= 1: yloop;
+    }
+}
+
 //////////
 // TEXT //
 //////////
@@ -304,7 +447,8 @@ global mut textCursorX = 0;
 global mut textCursorY = 0;
 global mut textColor = 15;
 global mut bgColor = 0;
-global mut graphicsText = 0;
+global mut outputChannel = 0;
+global mut textScale = 1;
 
 export fn printChar(char: i32) {
     loop chars {
@@ -316,6 +460,13 @@ export fn printChar(char: i32) {
 global mut controlCodeLength = 0;
 
 fn printSingleChar(char: i32) {
+    let charSize = 8 * textScale;
+
+    if outputChannel >= 2 & (char < 4 | char > 6) {
+        logChar(char);
+        return;
+    }
+
     controlCodeLength?0x12d20 = char;
     controlCodeLength = controlCodeLength + 1;
     char = 0x12d20?0;
@@ -329,54 +480,61 @@ fn printSingleChar(char: i32) {
         return;
     }
 
-    if char == 4 | char == 5 {
-        graphicsText = char == 5;
-	    textCursorX = 0;
-        textCursorY = 0;
+    if char >= 4 & char <= 6 {
+        outputChannel = char - 4;
+        if !outputChannel {
+            textCursorX = 0;
+            textCursorY = 0;
+        }
+        return;
+    }
+
+    if char == 7 {
+        80?0 = 80?0 ^ 2;
         return;
     }
 
     if char == 8 {
-        textCursorX = textCursorX - 8;
-        if !graphicsText & textCursorX < 0 {
-            textCursorX = 320-8;
+        textCursorX = textCursorX - charSize;
+        if !outputChannel & textCursorX < 0 {
+            textCursorX = 320-charSize;
             printSingleChar(11);
         }
         return;
     }
 
     if char == 9 {
-        if !graphicsText & textCursorX >= 320 {
+        if !outputChannel & textCursorX >= 320 {
             printChar(0xd0a);
         }
-        textCursorX = textCursorX + 8;
+        textCursorX = textCursorX + charSize;
         return;
     }
 
     if char == 10 {
-        textCursorY = textCursorY + 8;
-        if !graphicsText & textCursorY >= 240 {
-            textCursorY = 240 - 8;
+        textCursorY = textCursorY + charSize;
+        if !outputChannel & textCursorY >= 240 {
+            textCursorY = 240 - charSize;
             let i: i32;
             loop scroll_copy {
-                i!120 = i!(120 + 320 * 8);
-                branch_if (i := i + 4) < 320 * (240 - 8): scroll_copy;
+                i!120 = (i + 320 * charSize)!120;
+                branch_if (i := i + 4) < 320 * (240 - charSize): scroll_copy;
             }
-            rectangle(0 as f32, (240 - 8) as f32, 320 as f32, 8 as f32, bgColor);
+            rectangle(0 as f32, (240 - charSize) as f32, 320 as f32, charSize as f32, bgColor);
         }
         return;
     }
 
     if char == 11 {
-        textCursorY = textCursorY - 8;
-        if !graphicsText & textCursorY < 0 {
+        textCursorY = textCursorY - charSize;
+        if !outputChannel & textCursorY < 0 {
             textCursorY = 0;
-            let i = 320 * (240 - 8);
+            let i = 320 * (240 - charSize);
             loop scroll_copy {
-                i!(116 + 320 * 8) = i!116;
+                (i + 320 * charSize)!116 = i!116;
                 branch_if (i := i - 4): scroll_copy;
             }
-            rectangle(0 as f32, 0 as f32, 320 as f32, 8 as f32, bgColor);
+            rectangle(0 as f32, 0 as f32, 320 as f32, charSize as f32, bgColor);
         }
         return;
     }
@@ -408,9 +566,15 @@ fn printSingleChar(char: i32) {
         return;
     }
 
+    if char == 30 {
+        let scale = 0x12d20?1;
+        textScale = select(scale > 0 & scale <= 16, scale, 1);
+        return;
+    }
+
     if char == 31 {
-        textCursorX = 0x12d20?1 * (8 - graphicsText * 6);
-        textCursorY = 0x12d20?2 * (8 - graphicsText * 7);
+        textCursorX = 0x12d20?1 * (8 - outputChannel * 6);
+        textCursorY = 0x12d20?2 * (8 - outputChannel * 7);
         return;
     }
 
@@ -430,35 +594,37 @@ data(0x12d00) {
     1, 1, 1, 1, // 16-19,
     1, 1, 1, 1, // 20-23,
     1, 1, 1, 1, // 24-27,
-    1, 1, 1, 3 // 28-31
+    1, 1, 2, 3 // 28-31
   )
 }
 
 fn drawChar(char: i32) {
-    if !graphicsText & textCursorX >= 320 {
+    if !outputChannel & textCursorX >= 320 {
         printChar(0xd0a);
     }
 
+    let charSize = 8 * textScale;
+
     let y: i32;
     loop rows {
-        let bits = (char * 8 + y)?0x13400;
+        let bits = (char * 8 + y / textScale)?0x13400;
         let x = 0;
-        if graphicsText {
+        if outputChannel {
             loop pixels {
-                if (bits := bits << 1) & 256 {
+                if (bits << (x / textScale)) & 128 {
                     setPixel(textCursorX + x, textCursorY + y, textColor);
                 }
-                branch_if (x := x + 1) < 8: pixels;
+                branch_if (x := x + 1) < charSize: pixels;
             }
         } else {
             loop pixels {
-                setPixel(textCursorX + x, textCursorY + y, select((bits := bits << 1) & 256, textColor, bgColor));
-                branch_if (x := x + 1) < 8: pixels;
+                setPixel(textCursorX + x, textCursorY + y, select((bits << (x / textScale)) & 128, textColor, bgColor));
+                branch_if (x := x + 1) < charSize: pixels;
             }
         }
-        branch_if (y := y + 1) < 8: rows;
+        branch_if (y := y + 1) < charSize: rows;
     }
-    textCursorX = textCursorX + 8;
+    textCursorX = textCursorX + charSize;
 }
 
 export fn printString(ptr: i32) {
@@ -495,11 +661,37 @@ export fn setBackgroundColor(col: i32) {
 }
 
 export fn setCursorPosition(x: i32, y: i32) {
-    let lazy scale = select(graphicsText, 1, 8);
+    let lazy scale = select(outputChannel, 1, 8);
     textCursorX = x * scale;
     textCursorY = y * scale;
 }
 
+///////////
+// SOUND //
+///////////
+
+include "ges.cwa"
+
+export fn playNote(channel: i32, note: i32) {
+    (channel * 6)?80 = (channel * 6)?80 & 0xfe ^ if note {
+        (channel * 6)?83 = note & 127;
+        2 | !(note >> 7)
+    } else {
+        0
+    };
+}
+
+data 80 {
+    i8(
+        0x80, 0xc0, 0, 81, 0xa0, 0x50,
+        0xc4, 0, 0, 69, 0x60, 0x40,
+        0x44, 0xb0, 0, 69, 0x90, 0x43,
+        0x4, 0xf0, 0, 69, 0xa4, 0x44,
+        0xff, 0xff,
+        1, 1, 0, 100, 0, 100
+    )
+}
+
 ///////////
 // SETUP //
 ///////////
@@ -540,10 +732,19 @@ start fn setup() {
         branch_if (i := i - 1) >= 0: expand_sweetie;
     }
 
+    memory.fill(0, 0, 64);
+    memory.fill(112, 0, 8);
+    memory.fill(0x14000, 0, 0x2c000);
+
+
     cls(0);
     randomSeed(random());
 }
 
+data 0x12c78 {
+    i32(80)
+}
+
 data 0x13000+192*4 {
     i32(
         0x2c1c1a,

site/content/_index.md

@@ -14,12 +14,14 @@ The initial motivation behind MicroW8 was to explore whether there was a way to
 * Memory: 256KB
 * Gamepad input (D-Pad + 4 Buttons)
 
+For detailed [documentation see here](docs).
+
 ## Examples
-* [Fireworks](v0.1.2#AgwvgP+M59snqjl4CMKw5sqm1Zw9yJCbSviMjeLUdHus2a3yl/a99+uiBeqZgP/2jqSjrLjRk73COMM6OSLpsxK8ugT1kuk/q4hQUqqPpGozHoa0laulzGGcahzdfdJsYaK1sIdeIYS9M5PnJx/Wk9H+PvWEPy2Zvv7I6IW7Fg==) (127 bytes): Some fireworks to welcome 2022.
-* [Skip Ahead](v0.1.2#AgyfpZ80wkW28kiUZ9VIK4v+RPnVxqjK1dz2BcDoNyQPsS2g4OgEzkTe6jyoAfFOmqKrS8SM2aRljBal9mjNn8i4fP9eBK+RehQKxxGtJa9FqftvqEnh3ez1YaYxqj7jgTdzJ/WAYVmKMovBT1myrX3FamqKSOgMsNedLhVTLAhQup3sNcYEjGNo8b0HZ5+AgMgCwYRGCe//XQOMAaAAzqDILgmpEZ/43RKHcQpHEQwbURfNQJpadJe2sz3q5FlQnTGXQ9oSMokidhlC+aR/IpNHieuBGLhFZ2GfnwVQ0geBbQpTPA==) (229 bytes): A port of my [TIC-80 256byte game](http://tic80.com/play?cart=1735) from LoveByte'21
-* [OhNoAnotherTunnel](v0.1.2#AgPP1oEFvPzY/rBZwTumtYn37zeMFgpir1Bkn91jsNcp26VzoUpkAOOJTtnzVBfW+/dGnnIdbq/irBUJztY5wuua80DORTYZndgdwZHcSk15ajc4nyO0g1A6kGWyW56oZk0iPYJA9WtUmoj0Plvy1CGwIZrMe57X7QZcdqc3u6zjTA41Tpiqi9vnO3xbhi8o594Vx0XPXwVzpYq1ZCTYenfAGaXKkDmAFJqiVIsiCg==) (175 bytes): A port of my [entry](http://tic80.com/play?cart=1871) in the Outline'21 bytebattle final
-* [Technotunnel](v0.1.2#AhPXpq894LaUhp5+HQf39f39/Jc8g5zUrBSc0uyKh36ivskczhY84h55zL8gWpkdvKuRQI+KIt80isKzh8jkM8nILcx0RUvyk8yjE8TgNsgkcORVI0RY5k3qE4ySjaycxa2DVZH61UWZuLsCouuwT7I80TbmmetQSbMywJ/avrrCZIAH0UzQfvOiCJNG48NI0FFY1vjB7a7dcp8Uqg==) (157 bytes): A port of my [entry](https://tic80.com/play?cart=1873) in the Outline'21 bytebattle quater final
-* [Font & Palette](v0.1.2#At/p39+IBnj6ry1TRe7jzVy2A4tXgBvmoW2itzoyF2aM28pGy5QDiKxqrk8l9sbWZLtnAb+jgOfU+9QhpuyCAkhN6gPOU481IUL/df96vNe3h288Dqwhd3sfFpothIVFsMwRK72kW2hiR7zWsaXyy5pNmjR6BJk4piWx9ApT1ZwoUajhk6/zij6itq/FD1U3jj/J3MOwqZ2ef8Bv6ZPQlJIYVf62icGa69wS6SI1qBpIFiF14F8PcztRVbKIxLpT4ArCS6nz6FPnyUkqATGSBNPJ): Just a simple viewer for the default font and palette.
+* [Skip Ahead](v0.2.0#AgVfq24KI2Ok2o8qVtPYj27fSuGnfeSKgbOkIOsaEQMov8TDYQ6UjdjwkZrYcM1i9alo4/+Bhm1PRFEa0YHJlJAk/PGoc2K41rejv9ZSqJqIHNjr7cappqhOR2jT+jk+0b0+U6hO+geRCTP2aufWs7L+f/Z27NFY8LKlqPSv+C6Rd6+ohoKi6sYl5Kcrlf1cyTinV7jTTnmbcXWVDBA5rRKxAGMUTDS8rHxqSztRITOaQVP1pSdYgi/BDdOJOxSOIkeaId84S+Ycls5na7EgwSfVIpgqF+tcfkUecb8t2mQrXA7pyKrh/wzHn5N6Oe5aOgmzY2YpTIct) (249 bytes): A port of my [TIC-80 256byte game](http://tic80.com/play?cart=1735) from LoveByte'21, now with sound
+* [Fireworks](v0.2.0#AgwvgP+M59snqjl4CMKw5sqm1Zw9yJCbSviMjeLUdHus2a3yl/a99+uiBeqZgP/2jqSjrLjRk73COMM6OSLpsxK8ugT1kuk/q4hQUqqPpGozHoa0laulzGGcahzdfdJsYaK1sIdeIYS9M5PnJx/Wk9H+PvWEPy2Zvv7I6IW7Fg==) (127 bytes): Some fireworks to welcome 2022.
+* [OhNoAnotherTunnel](v0.2.0#AgPP1oEFvPzY/rBZwTumtYn37zeMFgpir1Bkn91jsNcp26VzoUpkAOOJTtnzVBfW+/dGnnIdbq/irBUJztY5wuua80DORTYZndgdwZHcSk15ajc4nyO0g1A6kGWyW56oZk0iPYJA9WtUmoj0Plvy1CGwIZrMe57X7QZcdqc3u6zjTA41Tpiqi9vnO3xbhi8o594Vx0XPXwVzpYq1ZCTYenfAGaXKkDmAFJqiVIsiCg==) (175 bytes): A port of my [entry](http://tic80.com/play?cart=1871) in the Outline'21 bytebattle final
+* [Technotunnel](v0.2.0#AhPXpq894LaUhp5+HQf39f39/Jc8g5zUrBSc0uyKh36ivskczhY84h55zL8gWpkdvKuRQI+KIt80isKzh8jkM8nILcx0RUvyk8yjE8TgNsgkcORVI0RY5k3qE4ySjaycxa2DVZH61UWZuLsCouuwT7I80TbmmetQSbMywJ/avrrCZIAH0UzQfvOiCJNG48NI0FFY1vjB7a7dcp8Uqg==) (157 bytes): A port of my [entry](https://tic80.com/play?cart=1873) in the Outline'21 bytebattle quater final
+* [Font & Palette](v0.2.0#At/p39+IBnj6ry1TRe7jzVy2A4tXgBvmoW2itzoyF2aM28pGy5QDiKxqrk8l9sbWZLtnAb+jgOfU+9QhpuyCAkhN6gPOU481IUL/df96vNe3h288Dqwhd3sfFpothIVFsMwRK72kW2hiR7zWsaXyy5pNmjR6BJk4piWx9ApT1ZwoUajhk6/zij6itq/FD1U3jj/J3MOwqZ2ef8Bv6ZPQlJIYVf62icGa69wS6SI1qBpIFiF14F8PcztRVbKIxLpT4ArCS6nz6FPnyUkqATGSBNPJ): Just a simple viewer for the default font and palette.
 
 Examplers for older versions:
 
@@ -29,41 +31,20 @@ Examplers for older versions:
 
 ## Versions
 
-### v0.1.2
+### v0.3.0
 
-* [Web runtime](v0.1.2)
-* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-linux.tgz)
-* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-macos.tgz)
-* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-windows.zip)
+* [Web runtime](v0.3.0)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-windows.zip)
 
 Changes:
 
-* add option to `uw8 run` to run the cart in the browser using the web runtime
-* CurlyWas: implement `include` support
-* CurlyWas: implement support for constants
-* fix crash when trying to draw zero sized line
+* add blitSprite and grabSprite API calls
+* add support for integer scaling up to 16x for printing text
+* fix incompatibility with sound devices only offering 16bit audio formats
+* add support for br_table instruction in packed carts
 
-### v0.1.1
+### Older versions
 
-* [Web runtime](v0.1.1)
-* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-linux.tgz)
-* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-macos.tgz)
-* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-windows.zip)
-
-Changes:
-
-* implement more robust file watcher
-* add basic video recording on F10 in web runtime
-* add screenshot on F9
-* add watchdog to interrupt hanging update in native runtime
-* add devkit mode to web runtime
-* add unpack and compile commands to uw8
-* add support for table/element section in pack command
-* disable wayland support (caused missing window decorations in gnome)
-
-### v0.1.0
-
-* [Web runtime](v0.1.0)
-* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-linux.tgz)
-* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-macos.tgz)
-* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-windows.zip)
+[Find older versions here.](versions)

site/content/docs.md

@@ -5,22 +5,28 @@ description = "Docs"
 
 # Overview
 
-MicroW8 loads WebAssembly modules with a maximum size of 256kb. You module needs to export
+MicroW8 loads WebAssembly modules with a maximum size of 256kb. Your module needs to export
 a function `fn upd()` which will be called once per frame.
 After calling `upd` MicroW8 will display the 320x240 8bpp framebuffer located
 at offset 120 in memory with the 32bpp palette located at 0x13000.
 
 The memory has to be imported as `env` `memory` and has a maximum size of 256kb (4 pages).
 
+If the module exports a function called `start`, it will be called once after the module is
+loaded.
+
 # Memory map
 
 ```
 00000-00040: user memory
 00040-00044: time since module start in ms
 00044-0004c: gamepad state
-0004c-00078: reserved
+0004c-00050: reserved
+00050-00070: sound data (synced to sound thread)
+00070-00078: reserved
 00078-12c78: frame buffer
-12c78-13000: reserved
+12c78-12c7c: sound registers/work area base address (for sndGes function)
+12c7c-13000: reserved
 13000-13400: palette
 13400-13c00: font
 13c00-14000: reserved
@@ -140,13 +146,13 @@ Fills the circle at `cx, cy` and with `radius` with the given color index.
 
 (Sets all pixels where the pixel center lies inside the circle.)
 
-### fn rectangle_outline(x: f32, y: f32, w: f32, h: f32, color: i32)
+### fn rectangleOutline(x: f32, y: f32, w: f32, h: f32, color: i32)
 
 Draws a one pixel outline on the inside of the given rectangle.
 
 (Draws the outermost pixels that are still inside the rectangle area.)
 
-### fn circle_outline(cx: f32, cy: f32, radius: f32, color: i32)
+### fn circleOutline(cx: f32, cy: f32, radius: f32, color: i32)
 
 Draws a one pixel outline on the inside of the given circle.
 
@@ -156,6 +162,21 @@ Draws a one pixel outline on the inside of the given circle.
 
 Draws a line from `x1,y1` to `x2,y2` in the given color index.
 
+### fn blitSprite(spriteData: i32, size: i32, x: i32, y: i32, control: i32)
+
+Copies the pixel data at `spriteData` onto the screen at `x`, `y`. The size of the sprite is passed as `width | (height << 16)`.
+If the height is given as 0, the sprite is is treated as square (width x width).
+
+The control parameter controls masking and flipping of the sprite:
+* bits 0-7: color mask index
+* bit 8: switch on masked blit (pixel with color mask index are treated as transparent)
+* bit 9: flip sprite x
+* bit 10: flip sprite y
+
+### fn grabSprite(spriteData: i32, size: i32, x: i32, y: i32, control: i32)
+
+Copies the pixel data on the screen at `x`, `y` to `spriteData`. Parameters are exactly the same as `blitSprite`.
+
 ## Input
 
 MicroW8 provides input from a gamepad with one D-Pad and 4 buttons, or a keyboard emulation thereof.
@@ -217,34 +238,58 @@ When printing characters only the foreground pixels are set, the background is "
 
 Moving/printing past any border does not cause any special operation, the cursor just goes off-screen.
 
+### Text scale
+
+An integer text scale factor in the range 1x-16x can be set with control char 30. An attempt to
+set a scale outside that range will reset the scale to 1x.
+
+After startup and `cls` the scale is initialized to 1x.
+
 ### Control chars
 
 Characters 0-31 are control characters and don't print by default. They take the next 0-2 following characters as parameters.
 Avoid the reserved control chars, they are currently NOPs but their behavior can change in later MicroW8 versions.
 
-| Code  | Parameters | Operation                            |
-| ----- | ---------- | ------------------------------------ |
-| 0     | -          | Nop                                  |
-| 1     | char       | Print char (including control chars) |
-| 2-3   | -          | Reserved                             |
-| 4     | -          | Switch to normal mode                |
-| 5     | -          | Switch to graphics mode              |
-| 6-7   | -          | Reserved                             |
-| 8     | -          | Move cursor left                     |
-| 9     | -          | Move cursor right                    |
-| 10    | -          | Move cursor down                     |
-| 11    | -          | Move cursor up                       |
-| 12    | -          | do `cls(background_color)`           |
-| 13    | -          | Move cursor to the left border       |
-| 14    | color      | Set the background color             |
-| 15    | color      | Set the text color                   |
-| 16-23 | -          | Reserved                             |
-| 24    | -          | Swap text/background colors          |
-| 25-30 | -          | Reserved                             |
-| 31    | x, y       | Set cursor position (*)              |
+| Code  | Parameters | Operation                                  |
+| ----- | ---------- | ------------------------------------------ |
+| 0     | -          | Nop                                        |
+| 1     | char       | Print char (including control chars)       |
+| 2-3   | -          | Reserved                                   |
+| 4     | -          | Switch to normal mode, reset cursor to 0,0 |
+| 5     | -          | Switch to graphics mode                    |
+| 6     | -          | Switch output to (debug) console           |
+| 7     | -          | Bell / trigger sound channel 0             |
+| 8     | -          | Move cursor left                           |
+| 9     | -          | Move cursor right                          |
+| 10    | -          | Move cursor down                           |
+| 11    | -          | Move cursor up                             |
+| 12    | -          | do `cls(background_color)`                 |
+| 13    | -          | Move cursor to the left border             |
+| 14    | color      | Set the background color                   |
+| 15    | color      | Set the text color                         |
+| 16-23 | -          | Reserved                                   |
+| 24    | -          | Swap text/background colors                |
+| 25-29 | -          | Reserved                                   |
+| 30    | scale      | Set text scale (1-16)                      |
+| 31    | x, y       | Set cursor position (*)                    |
 
 (*) In graphics mode, the x coordinate is doubled when using control char 31 to be able to cover the whole screen with one byte.
 
+#### Debug output
+
+Control code 6 switches all text output (except codes 4 and 5 to switch output back to the screen) to the console. Where exactly this ends
+up (if at all) is an implementation detail of the runtimes. The native dev-runtime writes the debug output to `stdout`, the web runtime to
+the debug console using `console.log`. Both implementations buffer the output until they encounter a newline character (10) in the output stream.
+
+There may be future runtimes that ignore the debug output completely.
+
+In CurlyWas, a simple way to log some value might look like this:
+```
+printChar('\06V: '); // switch to console out, print some prefix
+printInt(some_value);
+printChar('\n\4'); // newline and switch back to screen
+```
+
 ### fn printChar(char: i32)
 
 Prints the character in the lower 8 bits of `char`. If the upper 24 bits are non-zero, right-shifts `char` by 8 bits and loops back to the beginning.
@@ -269,6 +314,130 @@ Sets the background color.
 
 Sets the cursor position. In normal mode `x` and `y` are multiplied by 8 to get the pixel position, in graphics mode they are used as is.
 
+## Sound
+
+### Low level operation
+
+MicroW8 actually runs two instances of your module. On the first instance, it calls `upd` and displays the framebuffer found in its memory. On the
+second instance, it calls `snd` instead with an incrementing sample index and expects that function to return sound samples for the left and right
+channel at 44100 Hz. If your module does not export a `snd` function, it calls the api function `sndGes` instead.
+
+As the only means of communication, 32 bytes starting at address 0x00050 are copied from main to sound memory after `upd` returns.
+
+By default, the `sndGes` function generates sound based on the 32 bytes at 0x00050. This means that in the default configuration those 32 bytes act
+as sound registers. See the `sndGes` function for the meaning of those registers.
+
+### export fn snd(sampleIndex: i32) -> f32
+
+If the module exports a `snd` function, it is called 88200 times per second to provide PCM sample data for playback (44.1kHz stereo).
+The `sampleIndex` will start at 0 and increments by 1 for each call. On even indices the function is expected to return a sample value for
+the left channel, on odd indices for the right channel.
+
+### fn playNote(channel: i32, note: i32)
+
+Triggers a note (1-127) on the given channel (0-3). Notes are semitones with 69 being A4 (same as MIDI). A note value of 0 stops the
+sound playing on that channel. A note value 128-255 will trigger note-128 and immediately stop it (playing attack+release parts of envelope).
+
+This function assumes the default setup, with the `sndGes` registers located at 0x00050.
+
+### fn sndGes(sampleIndex: i32) -> f32
+
+This implements a sound chip, generating sound based on 32 bytes of sound registers.
+
+The spec of this sound chip are:
+
+- 4 channels with individual volume control (0-15)
+- rect, saw, tri, noise wave forms selectable per channel
+- each wave form supports some kind of pulse width modulation
+- each channel has an optional automatic low pass filter, or can be sent to one of two manually controllable filters
+- each channel can select between a narrow and a wide stereo positioning. The two stereo positions of each channel are fixed.
+- optional ring modulation
+
+This function requires 1024 bytes of working memory, the first 32 bytes of which are interpreted as the sound registers.
+The base address of its working memory can be configured by writing the address to 0x12c78. It defaults to 0x00050.
+
+Here is a short description of the 32 sound registers.
+
+```
+00 - CTRL0
+06 - CTRL1
+0c - CTRL2
+12 - CTRL3
+  | 7  6 |   5  |   4  |  3  2  |    1    |    0    |
+  | wave | ring | wide | filter | trigger | note on |
+
+  note on: stay in decay/sustain part of envelope
+  trigger: the attack part of the envlope is triggered when either this changes
+           or note on is changed from 0 to 1.
+  filter : 0 - no filter
+           1 - fixed 6db 1-pole filter with cutoff two octaves above note
+           2 - programmable filter 0
+           3 - programmable filter 1
+  wide   : use wide stereo panning
+  ring   : ring modulate with triangle wave at frequency of previous channel
+  wave   : 0 - rectangle
+           1 - saw
+           2 - triangle
+           3 - noise
+
+01 - PULS0
+07 - PULS1
+0d - PULS2
+13 - PULS3
+  Pulse width 0-255, with 0 being the plain version of each wave form.
+  rectangle - 50%-100% pulse width
+  saw       - inverts 0%-100% of the saw wave form around the center
+  triangle  - morphs into an octave up triangle wave
+  noise     - blends into a decimated saw wave (just try it out)
+
+02 - FINE0
+08 - FINE1
+0e - FINE2
+14 - FINE3
+  Fractional note
+
+03 - NOTE0
+09 - NOTE1
+0f - NOTE2
+15 - NOTE3
+  Note, 69 = A4
+
+04 - ENVA0
+0a - ENVA1
+10 - ENVA2
+16 - ENVA3
+  | 7 6 5 4 | 3 2 1 0 |
+  | decay   | attack  |
+
+05 - ENVB0
+0b - ENVB1
+11 - ENVB2
+17 - ENVB3
+  | 7 6 5 4 | 3 2 1 0 |
+  | release | sustain |
+
+18 - VO01
+  | 7 6 5 4  | 3 2 1 0  |
+  | volume 1 | volume 0 |
+
+19 - VO23
+  | 7 6 5 4  | 3 2 1 0  |
+  | volume 3 | volume 2 |
+
+1a - FCTR0
+1b - FCTR1
+  | 7 6 5 4   | 3 | 2    | 1    | 0   |
+  | resonance | 0 | band | high | low |
+
+1c - FFIN0
+1e - FFIN1
+  cutoff frequency - fractional note
+
+1d - FNOT0
+1f - FNOT1
+  cutoff frequency - note
+```
+
 # The `uw8` tool
 
 The `uw8` tool included in the MicroW8 download includes a number of useful tools for developing MicroW8 carts. For small productions written in
@@ -293,6 +462,30 @@ and execution of the cart is stopped. Defaults to 30 (0.5s)
 * `-l LEVEL`, `--level LEVEL`: Compression level (0-9). Higher compression levels are really slow.
 * `-o FILE`, `--output FILE`: Write the loaded and optionally packed cart back to disk.
 
+when using the native runtime:
+
+* `-m`, `--no-audio`: Disable audio, also reduces cpu load a bit
+* `--no-gpu`:  Force old cpu-only window code
+* `--filter FILTER`:  Select an upscale filter at startup
+* `--fullscreen`:  Start in fullscreen mode
+
+Note that the cpu-only window does not support fullscreen nor upscale filters.
+
+Unless --no-gpu is given, uw8 will first try to open a gpu accelerated window, falling back to the old cpu-only window if that fails.
+Therefore you should rarely need to manually pass --no-gpu. If you prefer the old pixel doubling look to the now default crt filter,
+you can just pass `--filter nearest` or `--filter 1`.
+
+The upscale filter options are:
+```
+1, nearest              : Anti-aliased nearest filter
+2, fast_crt             : Very simple, cheap crt filter, not very good below a window size of 960x720
+3, ss_crt               : Super sampled crt filter, a little more demanding on the GPU but scales well to smaller window sizes
+4, chromatic_crt        : Variant of fast_crt with a slight offset of the three color dots of a pixel, still pretty cheap
+5, auto_crt (default)   : ss_crt below 960x720, chromatic_crt otherwise
+```
+
+You can switch the upscale filter at any time using the keys 1-5. You can toggle fullscreen with F.
+
 ## `uw8 pack`
 
 Usage:

site/content/versions.md

@@ -0,0 +1,146 @@
++++
+description = "Versions"
++++
+
+### v0.3.0
+
+* [Web runtime](../v0.3.0)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.3.0/microw8-0.3.0-windows.zip)
+
+Changes:
+
+* add blitSprite and grabSprite API calls
+* add support for integer scaling up to 16x for printing text
+* fix incompatibility with sound devices only offering 16bit audio formats
+* add support for br_table instruction in packed carts
+
+### v0.2.2
+
+* [Web runtime](../v0.2.2)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.2/microw8-0.2.2-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.2/microw8-0.2.2-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.2/microw8-0.2.2-windows.zip)
+
+Changes:
+
+* call `start` function after loading cart if the cart exports one
+* fix `sndGes` having the wrong name and not being included in the auto imports
+* fix control codes 4-6 (change text output mode) being invoked when used as parameters in other control sequences
+* only open browser window once a cart was compiled sucessfully when running with `-b`
+
+### v0.2.1
+
+* [Web runtime](../v0.2.1)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.1/microw8-0.2.1-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.1/microw8-0.2.1-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.1/microw8-0.2.1-windows.zip)
+
+Changes:
+
+* new gpu accelerated renderer with (optional) crt filter
+* optimized `hline` function, a big speed-up when drawing large filled circles or rectangles
+* print fractional size of packed `uw8` cart
+
+### v0.2.0
+
+* [Web runtime](../v0.2.0)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.0/microw8-0.2.0-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.0/microw8-0.2.0-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.0/microw8-0.2.0-windows.zip)
+
+Changes:
+
+* [add sound support!](docs#sound)
+* add support to redirect text output to the console for debugging using control code 6
+* update curlywas:
+  * add support for `else if`
+  * add support for escape sequences in strings
+  * add support for char literals
+  * add support for binop-assignment, eg. `+=`, `^=`, `<<=` etc. (also support for the tee operator: `+:=`)
+  * "integer constant cast to float" literal syntax in CurlyWas (ex. `1_f` is equivalent to `1 as f32`)
+
+### v0.2.0-rc3
+
+* [Web runtime](../v0.2.0-rc3)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc3/microw8-0.2.0-rc3-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc3/microw8-0.2.0-rc3-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc3/microw8-0.2.0-rc3-windows.zip)
+
+Changes:
+
+* improve timing stability some more. essentially now guaranteeing that "frame = time_ms * 6 / 100" returns
+  consecutive frame numbers, provided the module can be run at 60 fps
+* add support to redirect text output to the console for debugging using control code 6
+* update curlywas:
+  * add support for `else if`
+  * add support for escape sequences in strings
+  * add support for char literals
+  * add support for binop-assignment, eg. `+=`, `^=`, `<<=` etc. (also support for the tee operator: `+:=`)
+
+### v0.2.0-rc2
+
+* [Web runtime](../v0.2.0-rc2)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc2/microw8-0.2.0-rc2-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc2/microw8-0.2.0-rc2-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc2/microw8-0.2.0-rc2-windows.zip)
+
+Changes:
+
+* fix timing issues of sound playback, especially on systems with large sound buffers
+
+### v0.2.0-rc1
+
+* [Web runtime](../v0.2.0-rc1)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc1/microw8-0.2.0-rc1-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc1/microw8-0.2.0-rc1-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc1/microw8-0.2.0-rc1-windows.zip)
+
+Changes:
+
+* [add sound support](docs#sound)
+* "integer constant cast to float" literal syntax in CurlyWas (ex. `1_f` is equivalent to `1 as f32`)
+
+Known issues:
+
+* timing accuracy/update frequency of sound support currently depends on sound buffer size
+
+### v0.1.2
+
+* [Web runtime](../v0.1.2)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.2/microw8-0.1.2-windows.zip)
+
+Changes:
+
+* add option to `uw8 run` to run the cart in the browser using the web runtime
+*../ CurlyWas: implement `include` support
+* CurlyWas: implement support for constants
+* fix crash when trying to draw zero sized line
+
+### v0.1.1
+
+* [Web runtime](../v0.1.1)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.1/microw8-0.1.1-windows.zip)
+
+Changes:
+
+* implement more robust file watcher
+* add basic video recording on F10 in web runtime
+*../ add screenshot on F9
+* add watchdog to interrupt hanging update in native runtime
+* add devkit mode to web runtime
+*../ add unpack and compile commands to uw8
+* add support for table/element section in pack command
+* disable wayland support (caused missing window decorations in gnome)
+
+### v0.1.0
+
+* [Web runtime](../v0.1.0)
+* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-linux.tgz)
+* [MacOS](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-macos.tgz)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.1.0/microw8-0.1.0-windows.zip)

site/templates/index.html

@@ -4,7 +4,7 @@
         <section>
             <h1 class="text-center heading-text">A WebAssembly based fantasy console</h1>
         </section>
-        <a href="v0.1.1">
+        <a href="v0.3.0">
             <img class="demonstration-gif" style="width:640px;height:480px;image-rendering:pixelated" src="img/technotunnel.png"></img>
         </a>
     </div>

src/filewatcher.rs

@@ -1,9 +1,13 @@
 use anyhow::{anyhow, bail, Result};
-use notify::{DebouncedEvent, RecommendedWatcher, Watcher};
+use notify_debouncer_mini::{
+    new_debouncer,
+    notify::{self, RecommendedWatcher},
+    DebouncedEvent, DebouncedEventKind, Debouncer,
+};
 use std::{collections::BTreeSet, path::PathBuf, sync::mpsc, time::Duration};
 
 pub struct FileWatcher {
-    watcher: RecommendedWatcher,
+    debouncer: Debouncer<RecommendedWatcher>,
     watched_files: BTreeSet<PathBuf>,
     directories: BTreeSet<PathBuf>,
     rx: mpsc::Receiver<DebouncedEvent>,
@@ -12,9 +16,18 @@ pub struct FileWatcher {
 impl FileWatcher {
     pub fn new() -> Result<FileWatcher> {
         let (tx, rx) = mpsc::channel();
-        let watcher = notify::watcher(tx, Duration::from_millis(100))?;
+        let debouncer = new_debouncer(Duration::from_millis(100), move |res| match res {
+            Ok(events) => {
+                for event in events {
+                    let _ = tx.send(event);
+                }
+            }
+            Err(err) => {
+                eprintln!("Error watching for file changes: {err}");
+            }
+        })?;
         Ok(FileWatcher {
-            watcher,
+            debouncer,
             watched_files: BTreeSet::new(),
             directories: BTreeSet::new(),
             rx,
@@ -26,7 +39,8 @@ impl FileWatcher {
         let parent = path.parent().ok_or_else(|| anyhow!("File has no parent"))?;
 
         if !self.directories.contains(parent) {
-            self.watcher
+            self.debouncer
+                .watcher()
                 .watch(parent, notify::RecursiveMode::NonRecursive)?;
             self.directories.insert(parent.to_path_buf());
         }
@@ -36,16 +50,18 @@ impl FileWatcher {
     }
 
     pub fn poll_changed_file(&self) -> Result<Option<PathBuf>> {
-        let event = self.rx.try_recv();
-        match event {
-            Ok(DebouncedEvent::Create(path) | DebouncedEvent::Write(path)) => {
-                let handle = same_file::Handle::from_path(&path)?;
-                for file in &self.watched_files {
-                    if handle == same_file::Handle::from_path(file)? {
-                        return Ok(Some(path));
+        match self.rx.try_recv() {
+            Ok(event) => match event.kind {
+                DebouncedEventKind::Any => {
+                    let handle = same_file::Handle::from_path(&event.path)?;
+                    for file in &self.watched_files {
+                        if handle == same_file::Handle::from_path(file)? {
+                            return Ok(Some(event.path));
+                        }
                     }
                 }
-            }
+                _ => (),
+            },
             Err(mpsc::TryRecvError::Disconnected) => bail!("File watcher disconnected"),
             _ => (),
         }

src/lib.rs

@@ -14,9 +14,6 @@ use anyhow::Result;
 
 pub trait Runtime {
     fn is_open(&self) -> bool;
-    fn set_timeout(&mut self, _timeout: u32) {
-        eprintln!("Warning: runtime doesn't support timeout");
-    }
     fn load(&mut self, module_data: &[u8]) -> Result<()>;
     fn run_frame(&mut self) -> Result<()>;
 }

src/main.rs

@@ -13,8 +13,13 @@ use uw8::RunWebServer;
 use uw8::Runtime;
 
 fn main() -> Result<()> {
+    env_logger::Builder::from_env(env_logger::Env::default()).init();
     let mut args = Arguments::from_env();
 
+    // try to enable ansi support in win10 cmd shell
+    #[cfg(target_os = "windows")]
+    let _ = ansi_term::enable_ansi_support();
+
     match args.subcommand()?.as_deref() {
         Some("version") => {
             println!("{}", env!("CARGO_PKG_VERSION"));
@@ -48,6 +53,7 @@ fn main() -> Result<()> {
 #[cfg(any(feature = "native", feature = "browser"))]
 fn run(mut args: Arguments) -> Result<()> {
     let watch_mode = args.contains(["-w", "--watch"]);
+    #[allow(unused)]
     let timeout: Option<u32> = args.opt_value_from_str(["-t", "--timeout"])?;
 
     let mut config = Config::default();
@@ -75,6 +81,18 @@ fn run(mut args: Arguments) -> Result<()> {
     #[cfg(not(feature = "native"))]
     let run_browser = args.contains(["-b", "--browser"]) || true;
 
+    #[allow(unused)]
+    let disable_audio = args.contains(["-m", "--no-audio"]);
+
+    #[cfg(feature = "native")]
+    let window_config = {
+        let mut config = uw8_window::WindowConfig::default();
+        if !run_browser {
+            config.parse_arguments(&mut args);
+        }
+        config
+    };
+
     let filename = args.free_from_os_str::<PathBuf, bool>(|s| Ok(s.into()))?;
 
     let mut watcher = uw8::FileWatcher::new()?;
@@ -85,7 +103,13 @@ fn run(mut args: Arguments) -> Result<()> {
         #[cfg(not(feature = "native"))]
         unimplemented!();
         #[cfg(feature = "native")]
-        Box::new(MicroW8::new()?)
+        {
+            let mut microw8 = MicroW8::new(timeout, window_config)?;
+            if disable_audio {
+                microw8.disable_audio();
+            }
+            Box::new(microw8)
+        }
     } else {
         #[cfg(not(feature = "browser"))]
         unimplemented!();
@@ -93,27 +117,20 @@ fn run(mut args: Arguments) -> Result<()> {
         Box::new(RunWebServer::new())
     };
 
-    if let Some(timeout) = timeout {
-        runtime.set_timeout(timeout);
-    }
-
     let mut first_run = true;
 
     while runtime.is_open() {
         if first_run || watcher.poll_changed_file()?.is_some() {
-            match start_cart(&filename, &mut *runtime, &config) {
-                Ok(dependencies) => {
-                    if watch_mode {
-                        for dep in dependencies {
-                            watcher.add_file(dep)?;
-                        }
-                    }
+            let (result, dependencies) = start_cart(&filename, &mut *runtime, &config);
+            if watch_mode {
+                for dep in dependencies {
+                    watcher.add_file(dep)?;
                 }
-                Err(err) => {
-                    eprintln!("Load error: {}", err);
-                    if !watch_mode {
-                        exit(1);
-                    }
+            }
+            if let Err(err) = result {
+                eprintln!("Load error: {}", err);
+                if !watch_mode {
+                    exit(1);
                 }
             }
             first_run = false;
@@ -136,37 +153,49 @@ struct Config {
     output_path: Option<PathBuf>,
 }
 
-fn load_cart(filename: &Path, config: &Config) -> Result<(Vec<u8>, Vec<PathBuf>)> {
+fn load_cart(filename: &Path, config: &Config) -> (Result<Vec<u8>>, Vec<PathBuf>) {
     let mut dependencies = Vec::new();
-    let mut cart = match SourceType::of_file(filename)? {
-        SourceType::Binary => {
-            let mut cart = vec![];
-            File::open(filename)?.read_to_end(&mut cart)?;
-            dependencies.push(filename.to_path_buf());
-            cart
-        }
-        SourceType::Wat => {
-            let cart = wat::parse_file(filename)?;
-            dependencies.push(filename.to_path_buf());
-            cart
-        }
-        SourceType::CurlyWas => {
-            let module = curlywas::compile_file(filename, curlywas::Options::default())?;
-            dependencies = module.dependencies;
-            module.wasm
-        }
-    };
+    fn inner(filename: &Path, config: &Config, dependencies: &mut Vec<PathBuf>) -> Result<Vec<u8>> {
+        let mut cart = match SourceType::of_file(filename)? {
+            SourceType::Binary => {
+                let mut cart = vec![];
+                File::open(filename)?.read_to_end(&mut cart)?;
+                cart
+            }
+            SourceType::Wat => {
+                let cart = wat::parse_file(filename)?;
+                cart
+            }
+            SourceType::CurlyWas => {
+                let (module, deps) = curlywas::compile_file(filename, curlywas::Options::default());
+                *dependencies = deps;
+                module?
+            }
+        };
 
-    if let Some(ref pack_config) = config.pack {
-        cart = uw8_tool::pack(&cart, pack_config)?;
-        println!("packed size: {} bytes", cart.len());
+        if let Some(ref pack_config) = config.pack {
+            cart = uw8_tool::pack(&cart, pack_config)?;
+            println!(
+                "\npacked size: {} bytes ({:.2})",
+                cart.len(),
+                uw8_tool::compressed_size(&cart)
+            );
+        }
+
+        if let Some(ref path) = config.output_path {
+            File::create(path)?.write_all(&cart)?;
+        }
+
+        Ok(cart)
     }
 
-    if let Some(ref path) = config.output_path {
-        File::create(path)?.write_all(&cart)?;
+    let result = inner(filename, config, &mut dependencies);
+
+    if dependencies.is_empty() {
+        dependencies.push(filename.to_path_buf());
     }
 
-    Ok((cart, dependencies))
+    (result, dependencies)
 }
 
 enum SourceType {
@@ -205,14 +234,22 @@ impl SourceType {
 }
 
 #[cfg(any(feature = "native", feature = "browser"))]
-fn start_cart(filename: &Path, runtime: &mut dyn Runtime, config: &Config) -> Result<Vec<PathBuf>> {
-    let cart = load_cart(filename, config)?;
+fn start_cart(
+    filename: &Path,
+    runtime: &mut dyn Runtime,
+    config: &Config,
+) -> (Result<()>, Vec<PathBuf>) {
+    let (cart, dependencies) = load_cart(filename, config);
+    let cart = match cart {
+        Ok(cart) => cart,
+        Err(err) => return (Err(err), dependencies),
+    };
 
-    if let Err(err) = runtime.load(&cart.0) {
+    if let Err(err) = runtime.load(&cart) {
         eprintln!("Load error: {}", err);
-        Err(err)
+        (Err(err), dependencies)
     } else {
-        Ok(cart.1)
+        (Ok(()), dependencies)
     }
 }
 
@@ -231,13 +268,14 @@ fn pack(mut args: Arguments) -> Result<()> {
 
     let out_file = args.free_from_os_str::<PathBuf, bool>(|s| Ok(s.into()))?;
 
-    let (cart, _) = load_cart(
+    let cart = load_cart(
         &in_file,
         &Config {
             pack: Some(pack_config),
             output_path: None,
         },
-    )?;
+    )
+    .0?;
 
     File::create(out_file)?.write_all(&cart)?;
 
@@ -260,8 +298,8 @@ fn compile(mut args: Arguments) -> Result<()> {
     let in_file = args.free_from_os_str::<PathBuf, bool>(|s| Ok(s.into()))?;
     let out_file = args.free_from_os_str::<PathBuf, bool>(|s| Ok(s.into()))?;
 
-    let module = curlywas::compile_file(in_file, options)?;
-    File::create(out_file)?.write_all(&module.wasm)?;
+    let module = curlywas::compile_file(in_file, options).0?;
+    File::create(out_file)?.write_all(&module)?;
 
     Ok(())
 }

src/run_native.rs

@@ -1,41 +1,34 @@
-use std::sync::{Arc, Mutex};
+use std::sync::{mpsc, Arc, Mutex};
 use std::time::Duration;
 use std::{thread, time::Instant};
 
-use anyhow::Result;
-use minifb::{Key, Window, WindowOptions};
+use anyhow::{anyhow, Result};
+use cpal::traits::*;
+use rubato::Resampler;
+use uw8_window::{Window, WindowConfig};
 use wasmtime::{
-    Engine, GlobalType, Memory, MemoryType, Module, Mutability, Store, TypedFunc, ValType,
+    Engine, Func, GlobalType, Memory, MemoryType, Module, Mutability, Store, TypedFunc, ValType,
 };
 
-static GAMEPAD_KEYS: &[Key] = &[
-    Key::Up,
-    Key::Down,
-    Key::Left,
-    Key::Right,
-    Key::Z,
-    Key::X,
-    Key::A,
-    Key::S,
-];
-
 pub struct MicroW8 {
+    window: Window,
+    stream: Option<cpal::Stream>,
     engine: Engine,
     loader_module: Module,
-    window: Window,
-    window_buffer: Vec<u32>,
-    instance: Option<UW8Instance>,
+    disable_audio: bool,
+    module_data: Option<Vec<u8>>,
     timeout: u32,
+    instance: Option<UW8Instance>,
 }
 
 struct UW8Instance {
     store: Store<()>,
     memory: Memory,
     end_frame: TypedFunc<(), ()>,
-    update: TypedFunc<(), ()>,
+    update: Option<TypedFunc<(), ()>>,
     start_time: Instant,
-    module: Vec<u8>,
     watchdog: Arc<Mutex<UW8WatchDog>>,
+    sound_tx: Option<mpsc::SyncSender<RegisterUpdate>>,
 }
 
 impl Drop for UW8Instance {
@@ -47,66 +40,60 @@ impl Drop for UW8Instance {
 }
 
 struct UW8WatchDog {
-    interupt: wasmtime::InterruptHandle,
-    timeout: u32,
+    engine: Engine,
     stop: bool,
 }
 
 impl MicroW8 {
-    pub fn new() -> Result<MicroW8> {
-        let engine = wasmtime::Engine::new(wasmtime::Config::new().interruptable(true))?;
+    pub fn new(timeout: Option<u32>, window_config: WindowConfig) -> Result<MicroW8> {
+        let mut config = wasmtime::Config::new();
+        config.cranelift_opt_level(wasmtime::OptLevel::Speed);
+        if timeout.is_some() {
+            config.epoch_interruption(true);
+        }
+        let engine = wasmtime::Engine::new(&config)?;
 
         let loader_module =
             wasmtime::Module::new(&engine, include_bytes!("../platform/bin/loader.wasm"))?;
 
-        let options = WindowOptions {
-            scale: minifb::Scale::X2,
-            scale_mode: minifb::ScaleMode::AspectRatioStretch,
-            resize: true,
-            ..Default::default()
-        };
-        let mut window = Window::new("MicroW8", 320, 240, options)?;
-        window.limit_update_rate(Some(std::time::Duration::from_micros(16666)));
+        let window = Window::new(window_config)?;
 
         Ok(MicroW8 {
+            window,
+            stream: None,
             engine,
             loader_module,
-            window,
-            window_buffer: vec![0u32; 320 * 240],
+            disable_audio: false,
+            module_data: None,
+            timeout: timeout.unwrap_or(0),
             instance: None,
-            timeout: 30,
         })
     }
 
-    fn reset(&mut self) {
-        self.instance = None;
-        for v in &mut self.window_buffer {
-            *v = 0;
-        }
+    pub fn disable_audio(&mut self) {
+        self.disable_audio = true;
     }
 }
 
 impl super::Runtime for MicroW8 {
     fn is_open(&self) -> bool {
-        self.window.is_open() && !self.window.is_key_down(Key::Escape)
-    }
-
-    fn set_timeout(&mut self, timeout: u32) {
-        self.timeout = timeout;
+        self.window.is_open()
     }
 
     fn load(&mut self, module_data: &[u8]) -> Result<()> {
-        self.reset();
+        self.stream = None;
+        self.instance = None;
 
         let mut store = wasmtime::Store::new(&self.engine, ());
+        store.set_epoch_deadline(60);
 
         let memory = wasmtime::Memory::new(&mut store, MemoryType::new(4, Some(4)))?;
 
         let mut linker = wasmtime::Linker::new(&self.engine);
-        linker.define("env", "memory", memory)?;
+        linker.define(&store, "env", "memory", memory)?;
 
         let loader_instance = linker.instantiate(&mut store, &self.loader_module)?;
-        let load_uw8 = loader_instance.get_typed_func::<i32, i32, _>(&mut store, "load_uw8")?;
+        let load_uw8 = loader_instance.get_typed_func::<i32, i32>(&mut store, "load_uw8")?;
 
         let platform_data = include_bytes!("../platform/bin/platform.uw8");
         memory.data_mut(&mut store)[..platform_data.len()].copy_from_slice(platform_data);
@@ -119,46 +106,12 @@ impl super::Runtime for MicroW8 {
         let module_length = load_uw8.call(&mut store, module_data.len() as i32)? as u32 as usize;
         let module = wasmtime::Module::new(&self.engine, &memory.data(&store)[..module_length])?;
 
-        linker.func_wrap("env", "acos", |v: f32| v.acos())?;
-        linker.func_wrap("env", "asin", |v: f32| v.asin())?;
-        linker.func_wrap("env", "atan", |v: f32| v.atan())?;
-        linker.func_wrap("env", "atan2", |x: f32, y: f32| x.atan2(y))?;
-        linker.func_wrap("env", "cos", |v: f32| v.cos())?;
-        linker.func_wrap("env", "exp", |v: f32| v.exp())?;
-        linker.func_wrap("env", "log", |v: f32| v.ln())?;
-        linker.func_wrap("env", "sin", |v: f32| v.sin())?;
-        linker.func_wrap("env", "tan", |v: f32| v.tan())?;
-        linker.func_wrap("env", "pow", |a: f32, b: f32| a.powf(b))?;
-        for i in 9..64 {
-            linker.func_wrap("env", &format!("reserved{}", i), || {})?;
-        }
-        for i in 0..16 {
-            linker.define(
-                "env",
-                &format!("g_reserved{}", i),
-                wasmtime::Global::new(
-                    &mut store,
-                    GlobalType::new(ValType::I32, Mutability::Const),
-                    0.into(),
-                )?,
-            )?;
-        }
+        add_native_functions(&mut linker, &mut store)?;
 
-        let platform_instance = linker.instantiate(&mut store, &platform_module)?;
-
-        for export in platform_instance.exports(&mut store) {
-            linker.define(
-                "env",
-                export.name(),
-                export
-                    .into_func()
-                    .expect("platform surely only exports functions"),
-            )?;
-        }
+        let platform_instance = instantiate_platform(&mut linker, &mut store, &platform_module)?;
 
         let watchdog = Arc::new(Mutex::new(UW8WatchDog {
-            interupt: store.interrupt_handle()?,
-            timeout: self.timeout,
+            engine: self.engine.clone(),
             stop: false,
         }));
 
@@ -166,16 +119,11 @@ impl super::Runtime for MicroW8 {
             let watchdog = watchdog.clone();
             thread::spawn(move || loop {
                 thread::sleep(Duration::from_millis(17));
-                if let Ok(mut watchdog) = watchdog.lock() {
+                if let Ok(watchdog) = watchdog.lock() {
                     if watchdog.stop {
                         break;
                     }
-                    if watchdog.timeout > 0 {
-                        watchdog.timeout -= 1;
-                        if watchdog.timeout == 0 {
-                            watchdog.interupt.interrupt();
-                        }
-                    }
+                    watchdog.engine.increment_epoch();
                 } else {
                     break;
                 }
@@ -183,11 +131,27 @@ impl super::Runtime for MicroW8 {
         }
 
         let instance = linker.instantiate(&mut store, &module)?;
-        if let Ok(mut watchdog) = watchdog.lock() {
-            watchdog.timeout = 0;
+        let end_frame = platform_instance.get_typed_func::<(), ()>(&mut store, "endFrame")?;
+        let update = instance.get_typed_func::<(), ()>(&mut store, "upd").ok();
+
+        if let Some(start) = instance.get_typed_func::<(), ()>(&mut store, "start").ok() {
+            start.call(&mut store, ())?;
         }
-        let end_frame = platform_instance.get_typed_func::<(), (), _>(&mut store, "endFrame")?;
-        let update = instance.get_typed_func::<(), (), _>(&mut store, "upd")?;
+
+        let (sound_tx, stream) = if self.disable_audio {
+            (None, None)
+        } else {
+            match init_sound(&self.engine, &platform_module, &module) {
+                Ok(sound) => {
+                    sound.stream.play()?;
+                    (Some(sound.tx), Some(sound.stream))
+                }
+                Err(err) => {
+                    eprintln!("Failed to init sound: {}", err);
+                    (None, None)
+                }
+            }
+        };
 
         self.instance = Some(UW8Instance {
             store,
@@ -195,66 +159,362 @@ impl super::Runtime for MicroW8 {
             end_frame,
             update,
             start_time: Instant::now(),
-            module: module_data.into(),
             watchdog,
+            sound_tx,
         });
-
+        self.stream = stream;
+        self.module_data = Some(module_data.into());
         Ok(())
     }
 
     fn run_frame(&mut self) -> Result<()> {
+        let input = self.window.begin_frame();
+
+        if input.reset {
+            if let Some(module_data) = self.module_data.take() {
+                self.load(&module_data)?;
+            }
+        }
+
+        let now = Instant::now();
         let mut result = Ok(());
         if let Some(mut instance) = self.instance.take() {
-            {
-                let time = instance.start_time.elapsed().as_millis() as i32;
-                let mut gamepad: u32 = 0;
-                for key in self.window.get_keys() {
-                    if let Some(index) = GAMEPAD_KEYS
-                        .iter()
-                        .enumerate()
-                        .find(|(_, &k)| k == key)
-                        .map(|(i, _)| i)
-                    {
-                        gamepad |= 1 << index;
-                    }
-                }
+            let time = (now - instance.start_time).as_millis() as i32;
+            let next_frame = {
+                let offset = ((time as u32 as i64 * 6) % 100 - 50) / 6;
+                let max = now + Duration::from_millis(17);
+                let next_center = now + Duration::from_millis((16 - offset) as u64);
+                next_center.min(max)
+            };
 
+            {
                 let mem = instance.memory.data_mut(&mut instance.store);
                 mem[64..68].copy_from_slice(&time.to_le_bytes());
-                mem[68..72].copy_from_slice(&gamepad.to_le_bytes());
+                mem[68..72].copy_from_slice(&input.gamepads);
             }
 
-            if let Ok(mut watchdog) = instance.watchdog.lock() {
-                watchdog.timeout = self.timeout;
-            }
-            result = instance.update.call(&mut instance.store, ());
-            if let Ok(mut watchdog) = instance.watchdog.lock() {
-                watchdog.timeout = 0;
+            instance.store.set_epoch_deadline(self.timeout as u64);
+            if let Some(ref update) = instance.update {
+                if let Err(err) = update.call(&mut instance.store, ()) {
+                    result = Err(err);
+                }
             }
             instance.end_frame.call(&mut instance.store, ())?;
 
             let memory = instance.memory.data(&instance.store);
-            let framebuffer = &memory[120..(120 + 320 * 240)];
-            let palette = &memory[0x13000..];
-            for (i, &color_index) in framebuffer.iter().enumerate() {
-                let offset = color_index as usize * 4;
-                self.window_buffer[i] = 0xff000000
-                    | ((palette[offset] as u32) << 16)
-                    | ((palette[offset + 1] as u32) << 8)
-                    | palette[offset + 2] as u32;
+
+            let mut sound_regs = [0u8; 32];
+            sound_regs.copy_from_slice(&memory[80..112]);
+            if let Some(ref sound_tx) = instance.sound_tx {
+                let _ = sound_tx.send(RegisterUpdate {
+                    time,
+                    data: sound_regs,
+                });
             }
 
-            if self.window.is_key_pressed(Key::R, minifb::KeyRepeat::No) {
-                self.load(&instance.module)?;
-            } else if result.is_ok() {
+            let framebuffer_mem = &memory[120..(120 + 320 * 240)];
+            let palette_mem = &memory[0x13000..];
+            self.window
+                .end_frame(framebuffer_mem, palette_mem, next_frame);
+
+            if result.is_ok() {
                 self.instance = Some(instance);
             }
         }
 
-        self.window
-            .update_with_buffer(&self.window_buffer, 320, 240)?;
-
         result?;
         Ok(())
     }
 }
+
+fn add_native_functions(
+    linker: &mut wasmtime::Linker<()>,
+    store: &mut wasmtime::Store<()>,
+) -> Result<()> {
+    linker.func_wrap("env", "acos", |v: f32| v.acos())?;
+    linker.func_wrap("env", "asin", |v: f32| v.asin())?;
+    linker.func_wrap("env", "atan", |v: f32| v.atan())?;
+    linker.func_wrap("env", "atan2", |x: f32, y: f32| x.atan2(y))?;
+    linker.func_wrap("env", "cos", |v: f32| v.cos())?;
+    linker.func_wrap("env", "exp", |v: f32| v.exp())?;
+    linker.func_wrap("env", "log", |v: f32| v.ln())?;
+    linker.func_wrap("env", "sin", |v: f32| v.sin())?;
+    linker.func_wrap("env", "tan", |v: f32| v.tan())?;
+    linker.func_wrap("env", "pow", |a: f32, b: f32| a.powf(b))?;
+    for i in 10..64 {
+        linker.func_wrap("env", &format!("reserved{}", i), || {})?;
+    }
+    let log_line = std::sync::Mutex::new(String::new());
+    linker.func_wrap("env", "logChar", move |c: i32| {
+        let mut log_line = log_line.lock().unwrap();
+        if c == 10 {
+            println!("{}", log_line);
+            log_line.clear();
+        } else {
+            log_line.push(c as u8 as char);
+        }
+    })?;
+    for i in 0..16 {
+        let global = wasmtime::Global::new(
+            &mut *store,
+            GlobalType::new(ValType::I32, Mutability::Const),
+            0.into(),
+        )?;
+        linker.define(&store, "env", &format!("g_reserved{}", i), global)?;
+    }
+
+    Ok(())
+}
+
+fn instantiate_platform(
+    linker: &mut wasmtime::Linker<()>,
+    store: &mut wasmtime::Store<()>,
+    platform_module: &wasmtime::Module,
+) -> Result<wasmtime::Instance> {
+    let platform_instance = linker.instantiate(&mut *store, &platform_module)?;
+
+    let exports: Vec<(String, Func)> = platform_instance
+        .exports(&mut *store)
+        .map(|e| {
+            (
+                e.name().to_owned(),
+                e.into_func()
+                    .expect("platform surely only exports functions"),
+            )
+        })
+        .collect();
+    for (name, func) in exports {
+        linker.define(&store, "env", &name, func)?;
+    }
+
+    Ok(platform_instance)
+}
+
+struct RegisterUpdate {
+    time: i32,
+    data: [u8; 32],
+}
+
+struct Uw8Sound {
+    stream: cpal::Stream,
+    tx: mpsc::SyncSender<RegisterUpdate>,
+}
+
+fn init_sound(
+    engine: &wasmtime::Engine,
+    platform_module: &wasmtime::Module,
+    module: &wasmtime::Module,
+) -> Result<Uw8Sound> {
+    let mut store = wasmtime::Store::new(engine, ());
+    store.set_epoch_deadline(60);
+
+    let memory = wasmtime::Memory::new(&mut store, MemoryType::new(4, Some(4)))?;
+
+    let mut linker = wasmtime::Linker::new(engine);
+    linker.define(&store, "env", "memory", memory)?;
+    add_native_functions(&mut linker, &mut store)?;
+
+    let platform_instance = instantiate_platform(&mut linker, &mut store, platform_module)?;
+    let instance = linker.instantiate(&mut store, module)?;
+
+    let snd = instance
+        .get_typed_func::<(i32,), f32>(&mut store, "snd")
+        .or_else(|_| platform_instance.get_typed_func::<(i32,), f32>(&mut store, "sndGes"))?;
+
+    let host = cpal::default_host();
+    let device = host
+        .default_output_device()
+        .ok_or_else(|| anyhow!("No audio output device available"))?;
+    let mut configs: Vec<_> = device
+        .supported_output_configs()?
+        .filter(|config| {
+            config.channels() == 2
+                && (config.sample_format() == cpal::SampleFormat::F32
+                    || config.sample_format() == cpal::SampleFormat::I16)
+        })
+        .collect();
+    configs.sort_by_key(|config| {
+        let rate = 44100
+            .max(config.min_sample_rate().0)
+            .min(config.max_sample_rate().0);
+        let prio = if rate >= 44100 {
+            rate - 44100
+        } else {
+            (44100 - rate) * 1000
+        };
+        prio + (config.sample_format() == cpal::SampleFormat::I16) as u32
+    });
+    let config = configs
+        .into_iter()
+        .next()
+        .ok_or_else(|| anyhow!("Could not find float or 16bit signed output config"))?;
+    let sample_rate = cpal::SampleRate(44100)
+        .max(config.min_sample_rate())
+        .min(config.max_sample_rate());
+    let config = config.with_sample_rate(sample_rate);
+    let buffer_size = match *config.buffer_size() {
+        cpal::SupportedBufferSize::Unknown => cpal::BufferSize::Default,
+        cpal::SupportedBufferSize::Range { min, max } => {
+            cpal::BufferSize::Fixed(256.max(min).min(max))
+        }
+    };
+    let sample_format = config.sample_format();
+    let config = cpal::StreamConfig {
+        buffer_size,
+        ..config.config()
+    };
+
+    let sample_rate = config.sample_rate.0 as usize;
+
+    let (tx, rx) = mpsc::sync_channel::<RegisterUpdate>(30);
+
+    struct Resampler {
+        resampler: rubato::FftFixedIn<f32>,
+        input_buffers: Vec<Vec<f32>>,
+        output_buffers: Vec<Vec<f32>>,
+        output_index: usize,
+    }
+    let mut resampler: Option<Resampler> = if sample_rate == 44100 {
+        None
+    } else {
+        let rs = rubato::FftFixedIn::new(44100, sample_rate, 128, 1, 2)?;
+        let input_buffers = rs.input_buffer_allocate(true);
+        let output_buffers = rs.output_buffer_allocate(true);
+        Some(Resampler {
+            resampler: rs,
+            input_buffers,
+            output_buffers,
+            output_index: usize::MAX,
+        })
+    };
+
+    let mut sample_index = 0;
+    let mut pending_updates: Vec<RegisterUpdate> = Vec::with_capacity(30);
+    let mut current_time = 0;
+
+    let mut callback = move |mut outer_buffer: &mut [f32], _: &_| {
+        let mut first_update = true;
+        while let Ok(update) = rx.try_recv() {
+            if first_update {
+                current_time += update.time.wrapping_sub(current_time) / 8;
+                first_update = false;
+            }
+            pending_updates.push(update);
+        }
+
+        while !outer_buffer.is_empty() {
+            store.set_epoch_deadline(30);
+            while pending_updates
+                .first()
+                .into_iter()
+                .any(|u| u.time.wrapping_sub(current_time) <= 0)
+            {
+                let update = pending_updates.remove(0);
+                memory.write(&mut store, 80, &update.data).unwrap();
+            }
+
+            let duration = if let Some(update) = pending_updates.first() {
+                ((update.time.wrapping_sub(current_time) as usize) * sample_rate + 999) / 1000
+            } else {
+                outer_buffer.len()
+            };
+            let step_size = (duration.max(64) * 2).min(outer_buffer.len());
+
+            let mut buffer = &mut outer_buffer[..step_size];
+
+            {
+                let mem = memory.data_mut(&mut store);
+                mem[64..68].copy_from_slice(&current_time.to_le_bytes());
+            }
+
+            fn clamp_sample(s: f32) -> f32 {
+                if s.is_nan() {
+                    0.0
+                } else {
+                    s.max(-1.0).min(1.0)
+                }
+            }
+
+            if let Some(ref mut resampler) = resampler {
+                while !buffer.is_empty() {
+                    let copy_size = resampler.output_buffers[0]
+                        .len()
+                        .saturating_sub(resampler.output_index)
+                        .min(buffer.len() / 2);
+                    if copy_size == 0 {
+                        resampler.input_buffers[0].clear();
+                        resampler.input_buffers[1].clear();
+                        for _ in 0..resampler.resampler.input_frames_next() {
+                            resampler.input_buffers[0].push(clamp_sample(
+                                snd.call(&mut store, (sample_index,)).unwrap_or(0.0),
+                            ));
+                            resampler.input_buffers[1].push(clamp_sample(
+                                snd.call(&mut store, (sample_index + 1,)).unwrap_or(0.0),
+                            ));
+                            sample_index = sample_index.wrapping_add(2);
+                        }
+
+                        resampler
+                            .resampler
+                            .process_into_buffer(
+                                &resampler.input_buffers,
+                                &mut resampler.output_buffers,
+                                None,
+                            )
+                            .unwrap();
+                        resampler.output_index = 0;
+                    } else {
+                        for i in 0..copy_size {
+                            buffer[i * 2] = resampler.output_buffers[0][resampler.output_index + i];
+                            buffer[i * 2 + 1] =
+                                resampler.output_buffers[1][resampler.output_index + i];
+                        }
+                        resampler.output_index += copy_size;
+                        buffer = &mut buffer[copy_size * 2..];
+                    }
+                }
+            } else {
+                for v in buffer {
+                    *v = clamp_sample(snd.call(&mut store, (sample_index,)).unwrap_or(0.0));
+                    sample_index = sample_index.wrapping_add(1);
+                }
+            }
+
+            outer_buffer = &mut outer_buffer[step_size..];
+            current_time = current_time.wrapping_add((step_size * 500 / sample_rate).max(1) as i32);
+        }
+    };
+    let stream = if sample_format == cpal::SampleFormat::F32 {
+        device.build_output_stream(
+            &config,
+            callback,
+            move |err| {
+                dbg!(err);
+            },
+            None,
+        )?
+    } else {
+        device.build_output_stream(
+            &config,
+            move |mut buffer: &mut [i16], info| {
+                let mut float_buffer = [0f32; 256];
+
+                while !buffer.is_empty() {
+                    let step_size = buffer.len().min(float_buffer.len());
+                    let step_buffer = &mut float_buffer[..step_size];
+                    callback(step_buffer, info);
+                    for (dest, src) in buffer.iter_mut().take(step_size).zip(step_buffer.iter()) {
+                        *dest = (src.max(-1.0).min(1.0) * 32767.0) as i16;
+                    }
+                    buffer = &mut buffer[step_size..];
+                }
+            },
+            move |err| {
+                dbg!(err);
+            },
+            None,
+        )?
+    };
+
+    Ok(Uw8Sound { stream, tx })
+}

src/run_web.rs

@@ -11,6 +11,7 @@ use warp::{http::Response, Filter};
 pub struct RunWebServer {
     cart: Arc<Mutex<Vec<u8>>>,
     tx: broadcast::Sender<()>,
+    socket_addr: SocketAddr,
 }
 
 impl RunWebServer {
@@ -18,8 +19,13 @@ impl RunWebServer {
         let cart = Arc::new(Mutex::new(Vec::new()));
         let (tx, _) = broadcast::channel(1);
 
+        let socket_addr = "127.0.0.1:3030"
+            .parse::<SocketAddr>()
+            .expect("Failed to parse socket address");
+
         let server_cart = cart.clone();
         let server_tx = tx.clone();
+        let server_addr = socket_addr.clone();
         thread::spawn(move || {
             let rt = tokio::runtime::Builder::new_current_thread()
                 .enable_io()
@@ -47,24 +53,26 @@ impl RunWebServer {
                     warp::sse::reply(warp::sse::keep_alive().stream(event_stream(&server_tx)))
                 });
 
-                let socket_addr = "127.0.0.1:3030"
-                    .parse::<SocketAddr>()
-                    .expect("Failed to parse socket address");
-
-                let server_future = warp::serve(html.or(cart).or(events)).bind(socket_addr);
-                println!("Point browser at http://{}", socket_addr);
-                let _ignore_result = webbrowser::open(&format!("http://{}", socket_addr));
+                let server_future = warp::serve(html.or(cart).or(events)).bind(server_addr);
                 server_future.await
             });
         });
 
-        RunWebServer { cart, tx }
+        RunWebServer {
+            cart,
+            tx,
+            socket_addr,
+        }
     }
 }
 
 impl super::Runtime for RunWebServer {
     fn load(&mut self, module_data: &[u8]) -> Result<()> {
         if let Ok(mut lock) = self.cart.lock() {
+            if lock.is_empty() && !module_data.is_empty() {
+                println!("Point browser at http://{}", self.socket_addr);
+                let _ignore_result = webbrowser::open(&format!("http://{}", self.socket_addr));
+            }
             lock.clear();
             lock.extend_from_slice(module_data);
         }

test.cwa

@@ -1,13 +0,0 @@
-import "env.memory" memory(4);
-import "env.printString" fn print(i32);
-
-export fn upd() {
-}
-
-start fn start() {
-    print(0);
-}
-
-data 0 {
-    "Press " i8(0xe0) " and " i8(0xe1) " to adjust, " i8(0xcc) " to commit." i8(0)
-}

test/drawing_test.cwa

@@ -0,0 +1,62 @@
+include "../examples/include/microw8-api.cwa"
+
+global mut counter = 0;
+
+export fn upd() {
+    cls(0);
+    
+    let col: i32 = 1;
+    
+    loop colors {
+        if !testCircle(counter, col) {
+            printInt(counter);
+            return;
+        }
+        counter += 1;
+        branch_if (col +:= 1) < 256: colors;
+    }
+}
+
+fn testCircle(seed: i32, col: i32) -> i32 {
+    randomSeed(seed);
+    let cx = randomf() * 640_f - 160_f;
+    let cy = randomf() * 480_f - 120_f;
+    let radius = randomf() * 4_f;
+    radius *= radius;
+    radius *= radius;
+    
+    circle(cx, cy, radius, col);
+    
+    let min_x = max(0_f, floor(cx - radius - 1_f)) as i32;
+    let min_y = max(0_f, floor(cy - radius - 1_f)) as i32;
+    let max_x = min(320_f, ceil(cx + radius + 1_f)) as i32;
+    let max_y = min(240_f, ceil(cy + radius + 1_f)) as i32;
+    
+    let x = min_x;
+    loop xloop {
+        if x < max_x {
+            let y = min_y;
+            loop yloop {
+                if y < max_y {
+                    let rx = x as f32 + 0.5 - cx;
+                    let ry = y as f32 + 0.5 - cy;
+                    let d = sqrt(rx*rx + ry*ry) - radius;
+                    if abs(d) > 0.001 {
+                        let is_inside = d < 0_f;
+                        let is_plotted = getPixel(x, y) == col;
+                        if is_inside != is_plotted {
+                            return 0;
+                        }
+                    }                    
+                
+                    y += 1;
+                    branch yloop;
+                }
+            }
+            x += 1;
+            branch xloop;
+        }
+    }
+    
+    1
+}

test/frame_time.cwa

@@ -0,0 +1,20 @@
+include "../examples/include/microw8-api.cwa"
+
+global mut pos = 0;
+global mut next = 0;
+
+export fn upd() {
+    let lazy t = 32!32;
+    let lazy tick = t * 6 / 100;
+    let lazy rel = t - tick * 100 / 6;
+
+    setBackgroundColor(select(tick == next, 0, select(tick < next, 0x35, 0x55)));
+    setCursorPosition(pos % 13 * 3, pos / 13 % 30);
+    if rel < 10 {
+        printChar(32);
+    }
+    printInt(rel);
+
+    pos = pos + 1;
+    next = tick + 1;
+}

test/ges_test.cwa

@@ -0,0 +1,31 @@
+import "env.memory" memory(4);
+import "env.pow" fn pow(f32, f32) -> f32;
+import "env.sin" fn sin(f32) -> f32;
+import "env.cls" fn cls(i32);
+import "env.exp" fn exp(f32) -> f32;
+import "env.rectangle" fn rectangle(f32, f32, f32, f32, i32);
+
+include "../platform/src/ges.cwa"
+
+export fn snd(t: i32) -> f32 {
+    sndGes(t)
+}
+
+export fn upd() {
+    80?0 = 32!32 / 200 & 2 | 0x41;
+    80?3 = (32!32 / 400)%8*12/7 + 40;
+    let pulse = (32!32 * 256 / 2000) & 511;
+    if pulse >= 256 {
+        pulse = 511 - pulse;
+    }
+    80?1 = pulse;
+
+    cls(0);
+    rectangle(0.0, 100.0, (pulse * 320 / 256) as f32, 16.0, 15);
+}
+
+data 80 {
+    i8(
+        0x41, 0, 0, 80, 0x70, 0
+    )
+}

test/log.cwa

@@ -0,0 +1,7 @@
+include "../examples/include/microw8-api.cwa"
+
+export fn upd() {
+    printChar('\06f: ');
+    printInt(32!32 * 6 / 100);
+    printChar('\n\4');
+}

test/plot_ges.cwa

@@ -0,0 +1,59 @@
+include "../examples/include/microw8-api.cwa"
+
+export fn upd() {
+    80?0 = (32!32 >> 11 << 6) | 5;
+    80?1 = (sin(time() * 6 as f32) * 95 as f32) as i32 + 128;
+    plotGes();
+}
+
+data 80 { i8 (
+    1, 128, 0, 69, 0, 15,
+    0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0,
+    0, 0, 0, 0, 0, 0,
+    0xff, 0xff,
+    0xc1, 0, 0, 110, 0, 0
+) }
+
+//import "env.gesSnd" fn gesSnd(i32) -> f32;
+
+include "../platform/src/ges.cwa"
+
+export fn snd(t: i32) -> f32 {
+    gesSnd(t)
+}
+
+global mut samplePos: i32 = 0;
+
+const SoundBuffer = 0x30000;
+
+fn plotGes() {
+    rectangle(0 as f32, 10 as f32, 320 as f32, 320 as f32, 0);
+    let count = (time() * 44100 as f32) as i32 * 2 - samplePos;
+    let i: i32;
+    loop genLoop {
+        (i*4)$SoundBuffer = gesSnd(samplePos + i);
+        branch_if (i := i + 1) < count: genLoop;
+    }
+    samplePos = samplePos + count;
+
+    let ch: i32;
+    loop channelLoop {
+        let offset = 159;
+        i = 0;
+
+        loop searchLoop {
+            offset = offset + 1;
+            branch_if (offset * 8 + ch - 8)$SoundBuffer < 0 as f32 | (offset * 8 + ch)$SoundBuffer >= 0 as f32 & offset + 160 < count: searchLoop;
+        }
+
+        offset = ch + (offset - 160) * 8;
+        i = 0;
+        loop plotLoop {
+            setPixel(i, floor((i * 8 + offset)$SoundBuffer * 127 as f32) as i32 + 60 + ch * (120/8), 15);
+            branch_if (i := i + 1) < 320: plotLoop;
+        }
+
+        branch_if (ch := ch + 8) < 16: channelLoop;
+    }
+}

test/start_fn.cwa

@@ -0,0 +1,5 @@
+include "../examples/include/microw8-api.cwa"
+
+export fn start() {
+	printChar('Test');
+}

test/text_modes.cwa

@@ -0,0 +1,13 @@
+include "../examples/include/microw8-api.cwa"
+
+export fn upd() {
+	printString(USER_MEM);
+}
+
+data USER_MEM {
+	i8(12, 31, 5, 6) "Text mode"
+	i8(5, 31, 4, 5) "Graphics mode"
+	i8(6) "Console output\nSecond line\n"
+	i8(4, 31, 4, 12) "Back to text mode"
+	i8(0)
+}

todo.txt

@@ -0,0 +1 @@
+* add support for 16bit sound (not just float)

uw8-tool/Cargo.lock

@@ -4,21 +4,27 @@ version = 3
 
 [[package]]
 name = "anyhow"
-version = "1.0.52"
+version = "1.0.81"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "84450d0b4a8bd1ba4144ce8ce718fbc5d071358b1e5384bace6536b3d1f2d5b3"
+checksum = "0952808a6c2afd1aa8947271f3a60f1a6763c7b912d210184c5149b5cf147247"
 
 [[package]]
 name = "autocfg"
-version = "1.0.1"
+version = "1.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "cdb031dd78e28731d87d56cc8ffef4a8f36ca26c38fe2de700543e627f8a464a"
+checksum = "f1fdabc7756949593fe60f30ec81974b613357de856987752631dea1e3394c80"
+
+[[package]]
+name = "bitflags"
+version = "2.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cf4b9d6a944f767f8e5e0db018570623c85f3d925ac718db4e06d0187adb21c1"
 
 [[package]]
 name = "cc"
-version = "1.0.72"
+version = "1.0.92"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "22a9137b95ea06864e018375b72adfb7db6e6f68cfc8df5a04d00288050485ee"
+checksum = "2678b2e3449475e95b0aa6f9b506a28e61b3dc8996592b983695e8ebb58a8b41"
 
 [[package]]
 name = "cdivsufsort"
@@ -30,32 +36,56 @@ dependencies = [
  "sacabase",
 ]
 
-[[package]]
-name = "cfg-if"
-version = "1.0.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd"
-
 [[package]]
 name = "crossbeam-channel"
-version = "0.5.1"
+version = "0.5.12"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "06ed27e177f16d65f0f0c22a213e17c696ace5dd64b14258b52f9417ccb52db4"
+checksum = "ab3db02a9c5b5121e1e42fbdb1aeb65f5e02624cc58c43f2884c6ccac0b82f95"
 dependencies = [
- "cfg-if",
  "crossbeam-utils",
 ]
 
 [[package]]
 name = "crossbeam-utils"
-version = "0.8.5"
+version = "0.8.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d82cfc11ce7f2c3faef78d8a684447b40d503d9681acebed6cb728d45940c4db"
+checksum = "248e3bacc7dc6baa3b21e405ee045c3047101a49145e7e9eca583ab4c2ca5345"
+
+[[package]]
+name = "equivalent"
+version = "1.0.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5443807d6dff69373d433ab9ef5378ad8df50ca6298caf15de6e52e24aaf54d5"
+
+[[package]]
+name = "fallible-iterator"
+version = "0.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "4443176a9f2c162692bd3d352d745ef9413eec5782a80d8fd6f8a1ac692a07f7"
+
+[[package]]
+name = "gimli"
+version = "0.26.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "22030e2c5a68ec659fde1e949a745124b48e6fa8b045b7ed5bd1fe4ccc5c4e5d"
 dependencies = [
- "cfg-if",
- "lazy_static",
+ "fallible-iterator",
+ "indexmap 1.9.3",
+ "stable_deref_trait",
 ]
 
+[[package]]
+name = "hashbrown"
+version = "0.12.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
+
+[[package]]
+name = "hashbrown"
+version = "0.14.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "290f1a1d9242c78d09ce40a5e87e7554ee637af1351968159f4952f028f75604"
+
 [[package]]
 name = "heck"
 version = "0.3.3"
@@ -72,10 +102,24 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "25a2bc672d1148e28034f176e01fffebb08b35768468cc954630da77a1449005"
 
 [[package]]
-name = "lazy_static"
-version = "1.4.0"
+name = "indexmap"
+version = "1.9.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646"
+checksum = "bd070e393353796e801d209ad339e89596eb4c8d430d18ede6a1cced8fafbd99"
+dependencies = [
+ "autocfg",
+ "hashbrown 0.12.3",
+]
+
+[[package]]
+name = "indexmap"
+version = "2.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "168fb715dda47215e360912c096649d23d58bf392ac62f73919e831745e40f26"
+dependencies = [
+ "equivalent",
+ "hashbrown 0.14.3",
+]
 
 [[package]]
 name = "leb128"
@@ -84,61 +128,63 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
 
 [[package]]
-name = "libc"
-version = "0.2.112"
+name = "lexopt"
+version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1b03d17f364a3a042d5e5d46b053bbbf82c92c9430c592dd4c064dc6ee997125"
+checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
+
+[[package]]
+name = "libc"
+version = "0.2.153"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9c198f91728a82281a64e1f4f9eeb25d82cb32a5de251c6bd1b5154d63a8e7bd"
 
 [[package]]
 name = "log"
-version = "0.4.14"
+version = "0.4.21"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "51b9bbe6c47d51fc3e1a9b945965946b4c44142ab8792c50835a980d362c2710"
-dependencies = [
- "cfg-if",
-]
+checksum = "90ed8c1e510134f979dbc4f070f87d4313098b704861a105fe34231c70a3901c"
 
 [[package]]
 name = "num-traits"
-version = "0.2.14"
+version = "0.2.18"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9a64b1ec5cda2586e284722486d802acf1f7dbdc623e2bfc57e65ca1cd099290"
+checksum = "da0df0e5185db44f69b44f26786fe401b6c293d1907744beaa7fa62b2e5a517a"
 dependencies = [
  "autocfg",
 ]
 
 [[package]]
 name = "pbr"
-version = "1.0.4"
+version = "1.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ff5751d87f7c00ae6403eb1fcbba229b9c76c9a30de8c1cf87182177b168cea2"
+checksum = "ed5827dfa0d69b6c92493d6c38e633bbaa5937c153d0d7c28bf12313f8c6d514"
 dependencies = [
  "crossbeam-channel",
  "libc",
- "time",
  "winapi",
 ]
 
 [[package]]
 name = "pico-args"
-version = "0.4.2"
+version = "0.5.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db8bcd96cb740d03149cbad5518db9fd87126a10ab519c011893b1754134c468"
+checksum = "5be167a7af36ee22fe3115051bc51f6e6c7054c9348e28deb4f49bd6f705a315"
 
 [[package]]
 name = "proc-macro2"
-version = "1.0.36"
+version = "1.0.79"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c7342d5883fbccae1cc37a2353b09c87c9b0f3afd73f5fb9bba687a1f733b029"
+checksum = "e835ff2298f5721608eb1a980ecaee1aef2c132bf95ecc026a11b7bf3c01c02e"
 dependencies = [
- "unicode-xid",
+ "unicode-ident",
 ]
 
 [[package]]
 name = "quote"
-version = "1.0.14"
+version = "1.0.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "47aa80447ce4daf1717500037052af176af5d38cc3e571d9ec1c7353fc10c87d"
+checksum = "291ec9ab5efd934aaf503a6466c5d5251535d108ee747472c3977cc5acc868ef"
 dependencies = [
  "proc-macro2",
 ]
@@ -153,48 +199,80 @@ dependencies = [
 ]
 
 [[package]]
-name = "syn"
-version = "1.0.84"
+name = "semver"
+version = "1.0.22"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ecb2e6da8ee5eb9a61068762a32fa9619cc591ceb055b3687f4cd4051ec2e06b"
+checksum = "92d43fe69e652f3df9bdc2b85b2854a0825b86e4fb76bc44d945137d053639ca"
+
+[[package]]
+name = "stable_deref_trait"
+version = "1.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a8f112729512f8e442d81f95a8a7ddf2b7c6b8a1a6f509a95864142b30cab2d3"
+
+[[package]]
+name = "syn"
+version = "1.0.109"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "72b64191b275b66ffe2469e8af2c1cfe3bafa67b529ead792a6d0160888b4237"
 dependencies = [
  "proc-macro2",
  "quote",
- "unicode-xid",
+ "unicode-ident",
 ]
 
 [[package]]
-name = "time"
-version = "0.1.44"
+name = "syn"
+version = "2.0.58"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6db9e6914ab8b1ae1c260a4ae7a49b6c5611b40328a735b21862567685e73255"
+checksum = "44cfb93f38070beee36b3fef7d4f5a16f27751d94b187b666a5cc5e9b0d30687"
 dependencies = [
- "libc",
- "wasi",
- "winapi",
+ "proc-macro2",
+ "quote",
+ "unicode-ident",
 ]
 
+[[package]]
+name = "thiserror"
+version = "1.0.58"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "03468839009160513471e86a034bb2c5c0e4baae3b43f79ffc55c4a5427b3297"
+dependencies = [
+ "thiserror-impl",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.58"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c61f3ba182994efc43764a46c018c347bc492c79f024e705f46567b418f6d4f7"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn 2.0.58",
+]
+
+[[package]]
+name = "unicode-ident"
+version = "1.0.12"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3354b9ac3fae1ff6755cb6db53683adb661634f67557942dea4facebec0fee4b"
+
 [[package]]
 name = "unicode-segmentation"
-version = "1.8.0"
+version = "1.11.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8895849a949e7845e06bd6dc1aa51731a103c42707010a5b591c0038fb73385b"
-
-[[package]]
-name = "unicode-xid"
-version = "0.2.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3"
+checksum = "d4c87d22b6e3f4a18d4d40ef354e97c90fcb14dd91d7dc0aa9d8a1172ebf7202"
 
 [[package]]
 name = "upkr"
-version = "0.1.0"
-source = "git+https://github.com/exoticorn/upkr.git?rev=2e7983fc#2e7983fc650788d98da2eecef2d16f63e849e4a0"
+version = "0.2.1"
+source = "git+https://github.com/exoticorn/upkr.git?rev=080db40d0088bbee2bdf3c5c75288ac7853d6b7a#080db40d0088bbee2bdf3c5c75288ac7853d6b7a"
 dependencies = [
  "anyhow",
  "cdivsufsort",
- "pbr",
- "pico-args",
+ "lexopt",
+ "thiserror",
 ]
 
 [[package]]
@@ -206,22 +284,24 @@ dependencies = [
  "pico-args",
  "upkr",
  "walrus",
- "wasm-encoder",
- "wasmparser 0.81.0",
+ "wasm-encoder 0.201.0",
+ "wasmparser 0.201.0",
 ]
 
 [[package]]
 name = "walrus"
-version = "0.19.0"
+version = "0.20.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4eb08e48cde54c05f363d984bb54ce374f49e242def9468d2e1b6c2372d291f8"
+checksum = "2c03529cd0c4400a2449f640d2f27cd1b48c3065226d15e26d98e4429ab0adb7"
 dependencies = [
  "anyhow",
+ "gimli",
  "id-arena",
  "leb128",
  "log",
  "walrus-macro",
- "wasmparser 0.77.0",
+ "wasm-encoder 0.29.0",
+ "wasmparser 0.80.2",
 ]
 
 [[package]]
@@ -233,35 +313,43 @@ dependencies = [
  "heck",
  "proc-macro2",
  "quote",
- "syn",
+ "syn 1.0.109",
 ]
 
 [[package]]
-name = "wasi"
-version = "0.10.0+wasi-snapshot-preview1"
+name = "wasm-encoder"
+version = "0.29.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1a143597ca7c7793eff794def352d41792a93c481eb1042423ff7ff72ba2c31f"
+checksum = "18c41dbd92eaebf3612a39be316540b8377c871cb9bde6b064af962984912881"
+dependencies = [
+ "leb128",
+]
 
 [[package]]
 name = "wasm-encoder"
-version = "0.8.0"
+version = "0.201.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db0c351632e46cc06a58a696a6c11e4cf90cad4b9f8f07a0b59128d616c29bb0"
+checksum = "b9c7d2731df60006819b013f64ccc2019691deccf6e11a1804bc850cd6748f1a"
 dependencies = [
  "leb128",
 ]
 
 [[package]]
 name = "wasmparser"
-version = "0.77.0"
+version = "0.80.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b35c86d22e720a07d954ebbed772d01180501afe7d03d464f413bb5f8914a8d6"
+checksum = "449167e2832691a1bff24cde28d2804e90e09586a448c8e76984792c44334a6b"
 
 [[package]]
 name = "wasmparser"
-version = "0.81.0"
+version = "0.201.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98930446519f63d00a836efdc22f67766ceae8dbcc1571379f2bcabc6b2b9abc"
+checksum = "84e5df6dba6c0d7fafc63a450f1738451ed7a0b52295d83e868218fa286bf708"
+dependencies = [
+ "bitflags",
+ "indexmap 2.2.6",
+ "semver",
+]
 
 [[package]]
 name = "winapi"

uw8-tool/Cargo.toml

@@ -6,10 +6,10 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-wasmparser = "0.81"
-wasm-encoder = "0.8"
-walrus = "0.19"
+wasmparser = "0.201"
+wasm-encoder = "0.201"
+walrus = { version = "0.20.3", default-features = false }
 anyhow = "1"
-pico-args = "0.4"
-upkr = { git = "https://github.com/exoticorn/upkr.git", rev = "2e7983fc" }
+pico-args = "0.5"
+upkr = { git = "https://github.com/exoticorn/upkr.git", rev = "080db40d0088bbee2bdf3c5c75288ac7853d6b7a" }
 pbr = "1"

uw8-tool/src/base_module.rs

@@ -3,7 +3,7 @@ use std::{collections::HashMap, fs::File, path::Path};
 use anyhow::{bail, Result};
 use std::io::prelude::*;
 use wasm_encoder::{
-    CodeSection, EntityType, Export, ExportSection, Function, FunctionSection, ImportSection,
+    CodeSection, EntityType, ExportKind, ExportSection, Function, FunctionSection, ImportSection,
     Instruction, MemoryType, Module, TypeSection, ValType,
 };
 use ValType::*;
@@ -152,20 +152,38 @@ impl BaseModule {
         add_function(
             &mut functions,
             &type_map,
-            "rectangle_outline",
+            "rectangleOutline",
             &[F32, F32, F32, F32, I32],
             None,
         );
         add_function(
             &mut functions,
             &type_map,
-            "circle_outline",
+            "circleOutline",
             &[F32, F32, F32, I32],
             None,
         );
 
         add_function(&mut functions, &type_map, "exp", &[F32], Some(F32));
 
+        add_function(&mut functions, &type_map, "playNote", &[I32, I32], None);
+        add_function(&mut functions, &type_map, "sndGes", &[I32], Some(F32));
+
+        add_function(
+            &mut functions,
+            &type_map,
+            "blitSprite",
+            &[I32, I32, I32, I32, I32],
+            None,
+        );
+        add_function(
+            &mut functions,
+            &type_map,
+            "grabSprite",
+            &[I32, I32, I32, I32, I32],
+            None,
+        );
+
         for i in functions.len()..64 {
             add_function(
                 &mut functions,
@@ -215,13 +233,13 @@ impl BaseModule {
             let mut imports = ImportSection::new();
 
             for (module, name, type_) in &self.function_imports {
-                imports.import(*module, Some(name.as_str()), EntityType::Function(*type_));
+                imports.import(*module, name.as_str(), EntityType::Function(*type_));
             }
 
             for (module, name, import) in &self.global_imports {
                 imports.import(
                     *module,
-                    Some(name.as_str()),
+                    name.as_str(),
                     EntityType::Global(wasm_encoder::GlobalType {
                         val_type: import.type_,
                         mutable: import.mutable,
@@ -231,11 +249,12 @@ impl BaseModule {
 
             imports.import(
                 "env",
-                Some("memory"),
+                "memory",
                 MemoryType {
                     minimum: self.memory as u64,
                     maximum: None,
                     memory64: false,
+                    shared: false,
                 },
             );
 
@@ -256,7 +275,7 @@ impl BaseModule {
             let mut exports = ExportSection::new();
 
             for (name, fnc) in &self.exports {
-                exports.export(*name, Export::Function(*fnc));
+                exports.export(*name, ExportKind::Func, *fnc);
             }
 
             module.section(&exports);
@@ -284,14 +303,17 @@ impl BaseModule {
 
     pub fn create_binary(path: &Path) -> Result<()> {
         let base1 = BaseModule::for_format_version(1)?.to_wasm();
-        let data = upkr::pack(&base1, 4, None);
+        let data = upkr::pack(&base1, 4, &upkr::Config::default(), None);
         File::create(path)?.write_all(&data)?;
         Ok(())
     }
 
     pub fn write_as_cwa<P: AsRef<Path>>(&self, path: P) -> Result<()> {
         fn inner(mut file: File, base: &BaseModule) -> Result<()> {
-            writeln!(file, "// MicroW8 APIs, to be `include`d in CurlyWas sources")?;
+            writeln!(
+                file,
+                "// MicroW8 APIs, to be `include`d in CurlyWas sources"
+            )?;
             writeln!(file, "import \"env.memory\" memory({});", base.memory)?;
             writeln!(file)?;
             for &(module, ref name, type_id) in &base.function_imports {
@@ -402,5 +424,5 @@ const CONSTANTS: &[(&str, u32)] = &[
     ("BUTTON_A", 4),
     ("BUTTON_B", 5),
     ("BUTTON_X", 6),
-    ("BUTTON_Y", 7)
+    ("BUTTON_Y", 7),
 ];

uw8-tool/src/filter_exports.rs

@@ -1,13 +1,17 @@
-use std::path::Path;
 use anyhow::Result;
+use std::path::Path;
 
 pub fn filter_exports(in_path: &Path, out_path: &Path) -> Result<()> {
     let mut module = walrus::Module::from_file(in_path)?;
 
-    let exports_to_delete: Vec<_> = module.exports.iter().filter_map(|export| match export.name.as_str() {
-        "upd" => None,
-        _ => Some(export.id())
-    }).collect();
+    let exports_to_delete: Vec<_> = module
+        .exports
+        .iter()
+        .filter_map(|export| match export.name.as_str() {
+            "start" | "upd" | "snd" => None,
+            _ => Some(export.id()),
+        })
+        .collect();
 
     for id in exports_to_delete {
         module.exports.delete(id);

uw8-tool/src/lib.rs

@@ -1,7 +1,15 @@
 mod base_module;
-mod pack;
 mod filter_exports;
+mod pack;
 
 pub use base_module::BaseModule;
-pub use pack::{pack, pack_file, unpack, unpack_file, PackConfig};
 pub use filter_exports::filter_exports;
+pub use pack::{pack, pack_file, unpack, unpack_file, PackConfig};
+
+pub fn compressed_size(cart: &[u8]) -> f32 {
+    if cart[0] != 2 {
+        cart.len() as f32
+    } else {
+        upkr::compressed_size(&cart[1..]) + 1.
+    }
+}

uw8-tool/src/pack.rs

@@ -10,7 +10,7 @@ use std::{
 use wasm_encoder as enc;
 use wasmparser::{
     BinaryReader, ExportSectionReader, ExternalKind, FunctionBody, FunctionSectionReader,
-    ImportSectionEntryType, ImportSectionReader, TableSectionReader, TypeSectionReader,
+    ImportSectionReader, TableSectionReader, TypeRef, TypeSectionReader,
 };
 
 pub struct PackConfig {
@@ -63,6 +63,7 @@ pub fn pack(data: &[u8], config: &PackConfig) -> Result<Vec<u8>> {
         uw8.extend_from_slice(&upkr::pack(
             &result[8..],
             level,
+            &upkr::Config::default(),
             Some(&mut |pos| {
                 pb.set(pos as u64);
             }),
@@ -89,7 +90,10 @@ pub fn unpack(data: Vec<u8>) -> Result<Vec<u8>> {
     let (version, data) = match data[0] {
         0 => return Ok(data),
         1 => (1, data[1..].to_vec()),
-        2 => (1, upkr::unpack(&data[1..])),
+        2 => (
+            1,
+            upkr::unpack(&data[1..], &upkr::Config::default(), 4 * 1024 * 1024)?,
+        ),
         other => bail!("Uknown format version {}", other),
     };
 
@@ -132,18 +136,19 @@ pub fn unpack(data: Vec<u8>) -> Result<Vec<u8>> {
     Ok(dest)
 }
 
-fn to_val_type(type_: &wasmparser::Type) -> Result<ValType> {
-    use wasmparser::Type::*;
+fn to_val_type(type_: &wasmparser::ValType) -> Result<ValType> {
+    use wasmparser::ValType::*;
     Ok(match *type_ {
         I32 => ValType::I32,
         I64 => ValType::I64,
         F32 => ValType::F32,
         F64 => ValType::F64,
+        V128 => ValType::V128,
         _ => bail!("Type {:?} isn't a value type", type_),
     })
 }
 
-fn to_val_type_vec(types: &[wasmparser::Type]) -> Result<Vec<ValType>> {
+fn to_val_type_vec(types: &[wasmparser::ValType]) -> Result<Vec<ValType>> {
     types.into_iter().map(to_val_type).collect()
 }
 
@@ -201,7 +206,7 @@ impl<'a> ParsedModule<'a> {
                     import_section = Some(Section::new(range, ImportSection::parse(reader)?));
                 }
                 Payload::GlobalSection(reader) => {
-                    global_section = Some(Section::new(range, reader.get_count()));
+                    global_section = Some(Section::new(range, reader.count()));
                 }
                 Payload::FunctionSection(reader) => {
                     function_section = Some(Section::new(range, read_function_section(reader)?));
@@ -219,17 +224,22 @@ impl<'a> ParsedModule<'a> {
                     validate_table_section(reader)?;
                     table_section = Some(Section::new(range, ()));
                 }
-                Payload::ElementSection(mut reader) => {
-                    let mut elements = Vec::with_capacity(reader.get_count() as usize);
-                    for _ in 0..reader.get_count() {
-                        elements.push(Element::parse(reader.read()?)?);
+                Payload::MemorySection(reader) => {
+                    if reader.count() != 0 {
+                        bail!("Found non-empty MemorySection. Memory has to be imported!");
+                    }
+                }
+                Payload::ElementSection(reader) => {
+                    let mut elements = Vec::with_capacity(reader.count() as usize);
+                    for element in reader {
+                        elements.push(Element::parse(element?)?);
                     }
                     element_section = Some(elements);
                 }
                 Payload::CodeSectionStart { .. } => (),
                 Payload::CodeSectionEntry(body) => function_bodies.push(body),
                 Payload::CustomSection { .. } => (),
-                Payload::End => break,
+                Payload::End(..) => break,
                 other => bail!("Unsupported section: {:?}", other),
             }
 
@@ -457,7 +467,7 @@ impl<'a> ParsedModule<'a> {
             {
                 let mut export_section = enc::ExportSection::new();
                 for (name, fnc) in my_exports {
-                    export_section.export(&name, enc::Export::Function(fnc));
+                    export_section.export(&name, enc::ExportKind::Func, fnc);
                 }
                 module.section(&export_section);
             }
@@ -480,8 +490,7 @@ impl<'a> ParsedModule<'a> {
                 }
                 element_section.active(
                     None,
-                    &wasm_encoder::Instruction::I32Const(element.start_index as i32),
-                    ValType::FuncRef,
+                    &wasm_encoder::ConstExpr::i32_const(element.start_index as i32),
                     wasm_encoder::Elements::Functions(&functions),
                 );
             }
@@ -527,30 +536,32 @@ fn copy_section(module: &mut wasm_encoder::Module, data: &[u8]) -> Result<()> {
 fn read_type_section(reader: TypeSectionReader) -> Result<Vec<base_module::FunctionType>> {
     let mut function_types = vec![];
 
-    for type_def in reader {
-        match type_def? {
-            wasmparser::TypeDef::Func(fnc) => {
-                if fnc.returns.len() > 1 {
-                    bail!("Multi-value not supported");
+    for rec_group in reader {
+        for sub_type in rec_group?.into_types() {
+            match sub_type.composite_type {
+                wasmparser::CompositeType::Func(fnc) => {
+                    if fnc.results().len() > 1 {
+                        bail!("Multi-value not supported");
+                    }
+                    let params = to_val_type_vec(fnc.params())?;
+                    let result = to_val_type_vec(fnc.results())?.into_iter().next();
+                    function_types.push(FunctionType { params, result });
                 }
-                let params = to_val_type_vec(&fnc.params)?;
-                let result = to_val_type_vec(&fnc.returns)?.into_iter().next();
-                function_types.push(FunctionType { params, result });
+                _ => bail!("Only function types are supported"),
             }
-            t => bail!("Unsupported type def {:?}", t),
         }
     }
 
     Ok(function_types)
 }
 
-fn validate_table_section(mut reader: TableSectionReader) -> Result<()> {
-    if reader.get_count() != 1 {
+fn validate_table_section(reader: TableSectionReader) -> Result<()> {
+    if reader.count() != 1 {
         bail!("Only up to one table supported");
     }
 
-    let type_ = reader.read()?;
-    if type_.element_type != wasmparser::Type::FuncRef {
+    let table = reader.into_iter().next().unwrap()?;
+    if !table.ty.element_type.is_func_ref() {
         bail!("Only one funcref table is supported");
     }
 
@@ -584,45 +595,38 @@ impl ImportSection {
 
         for import in reader {
             let import = import?;
-            if let Some(field) = import.field {
-                match import.ty {
-                    ImportSectionEntryType::Function(type_) => {
-                        functions.push(FunctionImport {
-                            module: import.module.to_string(),
-                            field: field.to_string(),
-                            type_,
-                        });
-                    }
-                    ImportSectionEntryType::Memory(mem) => {
-                        if import.module != "env" || field != "memory" {
-                            bail!(
-                                "Wrong name of memory import {}.{}, should be env.memory",
-                                import.module,
-                                field
-                            );
-                        }
-                        if mem.memory64 || mem.shared {
-                            bail!("Wrong memory import options: {:?}", import.ty);
-                        }
-                        memory = mem.maximum.unwrap_or(mem.initial) as u32;
-                    }
-                    ImportSectionEntryType::Global(glbl) => {
-                        globals.push(GlobalImport {
-                            module: import.module.to_string(),
-                            field: field.to_string(),
-                            type_: GlobalType {
-                                type_: to_val_type(&glbl.content_type)?,
-                                mutable: glbl.mutable,
-                            },
-                        });
-                    }
-                    _ => bail!("Unsupported import item {:?}", import.ty),
+            match import.ty {
+                TypeRef::Func(type_) => {
+                    functions.push(FunctionImport {
+                        module: import.module.to_string(),
+                        field: import.name.to_string(),
+                        type_,
+                    });
                 }
-            } else {
-                bail!(
-                    "Found import without field, only module '{}'",
-                    import.module
-                );
+                TypeRef::Memory(mem) => {
+                    if import.module != "env" || import.name != "memory" {
+                        bail!(
+                            "Wrong name of memory import {}.{}, should be env.memory",
+                            import.module,
+                            import.name
+                        );
+                    }
+                    if mem.memory64 || mem.shared {
+                        bail!("Wrong memory import options: {:?}", import.ty);
+                    }
+                    memory = mem.maximum.unwrap_or(mem.initial) as u32;
+                }
+                TypeRef::Global(glbl) => {
+                    globals.push(GlobalImport {
+                        module: import.module.to_string(),
+                        field: import.name.to_string(),
+                        type_: GlobalType {
+                            type_: to_val_type(&glbl.content_type)?,
+                            mutable: glbl.mutable,
+                        },
+                    });
+                }
+                _ => bail!("Unsupported import item {:?}", import.ty),
             }
         }
 
@@ -642,40 +646,40 @@ struct Element {
 
 impl Element {
     fn parse(element: wasmparser::Element) -> Result<Element> {
-        if element.ty != wasmparser::Type::FuncRef {
-            bail!("Table element type is not FuncRef");
-        }
+        match element.items {
+            wasmparser::ElementItems::Functions(funcs_reader) => {
+                let start_index = if let wasmparser::ElementKind::Active {
+                    offset_expr,
+                    table_index,
+                } = element.kind
+                {
+                    if table_index.map(|i| i > 0).unwrap_or(false) {
+                        bail!("Only function table with index 0 is supported");
+                    }
+                    let mut init_reader = offset_expr.get_operators_reader();
+                    if let wasmparser::Operator::I32Const { value: start_index } =
+                        init_reader.read()?
+                    {
+                        start_index as u32
+                    } else {
+                        bail!("Table element start index is not a integer constant");
+                    }
+                } else {
+                    bail!("Unsupported table element kind");
+                };
 
-        let start_index = if let wasmparser::ElementKind::Active {
-            init_expr,
-            table_index: 0,
-        } = element.kind
-        {
-            let mut init_reader = init_expr.get_operators_reader();
-            if let wasmparser::Operator::I32Const { value: start_index } = init_reader.read()? {
-                start_index as u32
-            } else {
-                bail!("Table element start index is not a integer constant");
-            }
-        } else {
-            bail!("Unsupported table element kind");
-        };
-
-        let mut items_reader = element.items.get_items_reader()?;
-
-        let mut functions = Vec::with_capacity(items_reader.get_count() as usize);
-        for _ in 0..items_reader.get_count() {
-            if let wasmparser::ElementItem::Func(index) = items_reader.read()? {
-                functions.push(index);
-            } else {
-                bail!("Table element item is not a function");
+                let mut functions = Vec::with_capacity(funcs_reader.count() as usize);
+                for index in funcs_reader {
+                    functions.push(index?);
+                }
+
+                Ok(Element {
+                    start_index,
+                    functions,
+                })
             }
+            _ => bail!("Table element type is not FuncRef"),
         }
-
-        Ok(Element {
-            start_index,
-            functions,
-        })
     }
 }
 
@@ -706,8 +710,8 @@ fn read_export_section(reader: ExportSectionReader) -> Result<Vec<(String, u32)>
     for export in reader {
         let export = export?;
         match export.kind {
-            ExternalKind::Function => {
-                function_exports.push((export.field.to_string(), export.index));
+            ExternalKind::Func => {
+                function_exports.push((export.name.to_string(), export.index));
             }
             _ => (), // just ignore all other kinds since MicroW8 doesn't expect any exports other than functions
         }
@@ -728,13 +732,11 @@ fn remap_function(
     }
     let mut function = enc::Function::new(locals);
 
-    let block_type = |ty: wasmparser::TypeOrFuncType| -> Result<enc::BlockType> {
+    let block_type = |ty: wasmparser::BlockType| -> Result<enc::BlockType> {
         Ok(match ty {
-            wasmparser::TypeOrFuncType::Type(wasmparser::Type::EmptyBlockType) => {
-                enc::BlockType::Empty
-            }
-            wasmparser::TypeOrFuncType::Type(ty) => enc::BlockType::Result(to_val_type(&ty)?),
-            wasmparser::TypeOrFuncType::FuncType(ty) => enc::BlockType::FunctionType(
+            wasmparser::BlockType::Empty => enc::BlockType::Empty,
+            wasmparser::BlockType::Type(ty) => enc::BlockType::Result(to_val_type(&ty)?),
+            wasmparser::BlockType::FuncType(ty) => enc::BlockType::FunctionType(
                 *type_map
                     .get(&ty)
                     .ok_or_else(|| anyhow!("Function type index out of range: {}", ty))?,
@@ -748,7 +750,7 @@ fn remap_function(
             .ok_or_else(|| anyhow!("Global index out of range: {}", idx))?)
     };
 
-    fn mem(m: wasmparser::MemoryImmediate) -> enc::MemArg {
+    fn mem(m: wasmparser::MemArg) -> enc::MemArg {
         enc::MemArg {
             offset: m.offset,
             align: m.align as u32,
@@ -763,24 +765,31 @@ fn remap_function(
         function.instruction(&match op? {
             De::Unreachable => En::Unreachable,
             De::Nop => En::Nop,
-            De::Block { ty } => En::Block(block_type(ty)?),
-            De::Loop { ty } => En::Loop(block_type(ty)?),
-            De::If { ty } => En::If(block_type(ty)?),
+            De::Block { blockty } => En::Block(block_type(blockty)?),
+            De::Loop { blockty } => En::Loop(block_type(blockty)?),
+            De::If { blockty } => En::If(block_type(blockty)?),
             De::Else => En::Else,
             De::Try { .. } | De::Catch { .. } | De::Throw { .. } | De::Rethrow { .. } => todo!(),
             De::End => En::End,
             De::Br { relative_depth } => En::Br(relative_depth),
             De::BrIf { relative_depth } => En::BrIf(relative_depth),
-            De::BrTable { .. } => todo!(),
+            De::BrTable { targets } => En::BrTable(
+                targets.targets().collect::<Result<Vec<u32>, _>>()?.into(),
+                targets.default(),
+            ),
             De::Return => En::Return,
             De::Call { function_index } => En::Call(
                 *function_map
                     .get(&function_index)
                     .ok_or_else(|| anyhow!("Function index out of range: {}", function_index))?,
             ),
-            De::CallIndirect { index, table_index } => En::CallIndirect {
+            De::CallIndirect {
+                type_index,
+                table_index,
+                table_byte: _, // what is this supposed to be?
+            } => En::CallIndirect {
                 ty: *type_map
-                    .get(&index)
+                    .get(&type_index)
                     .ok_or_else(|| anyhow!("Unknown function type in call indirect"))?,
                 table: table_index,
             },
@@ -800,16 +809,16 @@ fn remap_function(
             De::I64Load { memarg } => En::I64Load(mem(memarg)),
             De::F32Load { memarg } => En::F32Load(mem(memarg)),
             De::F64Load { memarg } => En::F64Load(mem(memarg)),
-            De::I32Load8S { memarg } => En::I32Load8_S(mem(memarg)),
-            De::I32Load8U { memarg } => En::I32Load8_U(mem(memarg)),
-            De::I32Load16S { memarg } => En::I32Load16_S(mem(memarg)),
-            De::I32Load16U { memarg } => En::I32Load16_U(mem(memarg)),
-            De::I64Load8S { memarg } => En::I64Load8_S(mem(memarg)),
-            De::I64Load8U { memarg } => En::I64Load8_U(mem(memarg)),
-            De::I64Load16S { memarg } => En::I64Load16_S(mem(memarg)),
-            De::I64Load16U { memarg } => En::I64Load16_U(mem(memarg)),
-            De::I64Load32S { memarg } => En::I64Load32_S(mem(memarg)),
-            De::I64Load32U { memarg } => En::I64Load32_U(mem(memarg)),
+            De::I32Load8S { memarg } => En::I32Load8S(mem(memarg)),
+            De::I32Load8U { memarg } => En::I32Load8U(mem(memarg)),
+            De::I32Load16S { memarg } => En::I32Load16S(mem(memarg)),
+            De::I32Load16U { memarg } => En::I32Load16U(mem(memarg)),
+            De::I64Load8S { memarg } => En::I64Load8S(mem(memarg)),
+            De::I64Load8U { memarg } => En::I64Load8U(mem(memarg)),
+            De::I64Load16S { memarg } => En::I64Load16S(mem(memarg)),
+            De::I64Load16U { memarg } => En::I64Load16U(mem(memarg)),
+            De::I64Load32S { memarg } => En::I64Load32S(mem(memarg)),
+            De::I64Load32U { memarg } => En::I64Load32U(mem(memarg)),
             De::I32Store { memarg } => En::I32Store(mem(memarg)),
             De::I64Store { memarg } => En::I64Store(mem(memarg)),
             De::F32Store { memarg } => En::F32Store(mem(memarg)),
@@ -828,7 +837,7 @@ fn remap_function(
             De::RefNull { .. } | De::RefIsNull { .. } | De::RefFunc { .. } => todo!(),
             De::I32Eqz => En::I32Eqz,
             De::I32Eq => En::I32Eq,
-            De::I32Ne => En::I32Neq,
+            De::I32Ne => En::I32Ne,
             De::I32LtS => En::I32LtS,
             De::I32LtU => En::I32LtU,
             De::I32GtS => En::I32GtS,
@@ -839,7 +848,7 @@ fn remap_function(
             De::I32GeU => En::I32GeU,
             De::I64Eqz => En::I64Eqz,
             De::I64Eq => En::I64Eq,
-            De::I64Ne => En::I64Neq,
+            De::I64Ne => En::I64Ne,
             De::I64LtS => En::I64LtS,
             De::I64LtU => En::I64LtU,
             De::I64GtS => En::I64GtS,
@@ -849,13 +858,13 @@ fn remap_function(
             De::I64GeS => En::I64GeS,
             De::I64GeU => En::I64GeU,
             De::F32Eq => En::F32Eq,
-            De::F32Ne => En::F32Neq,
+            De::F32Ne => En::F32Ne,
             De::F32Lt => En::F32Lt,
             De::F32Gt => En::F32Gt,
             De::F32Le => En::F32Le,
             De::F32Ge => En::F32Ge,
             De::F64Eq => En::F64Eq,
-            De::F64Ne => En::F64Neq,
+            De::F64Ne => En::F64Ne,
             De::F64Lt => En::F64Lt,
             De::F64Gt => En::F64Gt,
             De::F64Le => En::F64Le,
@@ -962,6 +971,246 @@ fn remap_function(
             De::I64TruncSatF32U => En::I64TruncSatF32U,
             De::I64TruncSatF64S => En::I64TruncSatF64S,
             De::I64TruncSatF64U => En::I64TruncSatF64U,
+            De::MemoryCopy { src_mem, dst_mem } => En::MemoryCopy { src_mem, dst_mem },
+            De::MemoryFill { mem } => En::MemoryFill(mem),
+
+            De::V128Const { value } => En::V128Const(value.i128()),
+            De::V128Load { memarg } => En::V128Load(mem(memarg)),
+            De::V128Store { memarg } => En::V128Store(mem(memarg)),
+            De::V128Load8x8S { memarg } => En::V128Load8x8S(mem(memarg)),
+            De::V128Load8x8U { memarg } => En::V128Load8x8U(mem(memarg)),
+            De::V128Load16x4S { memarg } => En::V128Load16x4S(mem(memarg)),
+            De::V128Load16x4U { memarg } => En::V128Load16x4U(mem(memarg)),
+            De::V128Load32x2S { memarg } => En::V128Load32x2S(mem(memarg)),
+            De::V128Load32x2U { memarg } => En::V128Load32x2U(mem(memarg)),
+            De::V128Load8Splat { memarg } => En::V128Load8Splat(mem(memarg)),
+            De::V128Load16Splat { memarg } => En::V128Load16Splat(mem(memarg)),
+            De::V128Load32Splat { memarg } => En::V128Load32Splat(mem(memarg)),
+            De::V128Load64Splat { memarg } => En::V128Load64Splat(mem(memarg)),
+            De::V128Load32Zero { memarg } => En::V128Load32Zero(mem(memarg)),
+            De::V128Load64Zero { memarg } => En::V128Load64Zero(mem(memarg)),
+
+            De::V128Load8Lane { memarg, lane } => En::V128Load8Lane { memarg: mem(memarg), lane },
+            De::V128Load16Lane { memarg, lane } => En::V128Load16Lane { memarg: mem(memarg), lane },
+            De::V128Load32Lane { memarg, lane } => En::V128Load32Lane { memarg: mem(memarg), lane },
+            De::V128Load64Lane { memarg, lane } => En::V128Load64Lane { memarg: mem(memarg), lane },
+            De::V128Store8Lane { memarg, lane } => En::V128Store8Lane { memarg: mem(memarg), lane },
+            De::V128Store16Lane { memarg, lane } => En::V128Store16Lane { memarg: mem(memarg), lane },
+            De::V128Store32Lane { memarg, lane } => En::V128Store32Lane { memarg: mem(memarg), lane },
+            De::V128Store64Lane { memarg, lane } => En::V128Store64Lane { memarg: mem(memarg), lane },
+
+            De::I8x16ExtractLaneS { lane } => En::I8x16ExtractLaneS(lane),
+            De::I8x16ExtractLaneU { lane } => En::I8x16ExtractLaneU(lane),
+            De::I8x16ReplaceLane { lane } => En::I8x16ReplaceLane(lane),
+            De::I16x8ExtractLaneS { lane } => En::I16x8ExtractLaneS(lane),
+            De::I16x8ExtractLaneU { lane } => En::I16x8ExtractLaneU(lane),
+            De::I16x8ReplaceLane { lane } => En::I16x8ReplaceLane(lane),
+            De::I32x4ExtractLane { lane } => En::I32x4ExtractLane(lane),
+            De::I32x4ReplaceLane { lane } => En::I32x4ReplaceLane(lane),
+            De::I64x2ExtractLane { lane } => En::I64x2ExtractLane(lane),
+            De::I64x2ReplaceLane { lane } => En::I64x2ReplaceLane(lane),
+            De::F32x4ExtractLane { lane } => En::F32x4ExtractLane(lane),
+            De::F32x4ReplaceLane { lane } => En::F32x4ReplaceLane(lane),
+            De::F64x2ExtractLane { lane } => En::F64x2ExtractLane(lane),
+            De::F64x2ReplaceLane { lane } => En::F64x2ReplaceLane(lane),
+
+            De::I8x16Splat => En::I8x16Splat,
+            De::I16x8Splat => En::I16x8Splat,
+            De::I32x4Splat => En::I32x4Splat,
+            De::I64x2Splat => En::I64x2Splat,
+            De::F32x4Splat => En::F32x4Splat,
+            De::F64x2Splat => En::F64x2Splat,
+            De::I8x16Swizzle => En::I8x16Swizzle,
+            De::I8x16Add => En::I8x16Add,
+            De::I16x8Add => En::I16x8Add,
+            De::I32x4Add => En::I32x4Add,
+            De::I64x2Add => En::I64x2Add,
+            De::F32x4Add => En::F32x4Add,
+            De::F64x2Add => En::F64x2Add,
+            De::I8x16Sub => En::I8x16Sub,
+            De::I16x8Sub => En::I16x8Sub,
+            De::I32x4Sub => En::I32x4Sub,
+            De::I64x2Sub => En::I64x2Sub,
+            De::F32x4Sub => En::F32x4Sub,
+            De::F64x2Sub => En::F64x2Sub,
+            De::I16x8Mul => En::I16x8Mul,
+            De::I32x4Mul => En::I32x4Mul,
+            De::I64x2Mul => En::I64x2Mul,
+            De::F32x4Mul => En::F32x4Mul,
+            De::F64x2Mul => En::F64x2Mul,
+            De::I32x4DotI16x8S => En::I32x4DotI16x8S,
+            De::I8x16Neg => En::I8x16Neg,
+            De::I16x8Neg => En::I16x8Neg,
+            De::I32x4Neg => En::I32x4Neg,
+            De::I64x2Neg => En::I64x2Neg,
+            De::F32x4Neg => En::F32x4Neg,
+            De::F64x2Neg => En::F64x2Neg,
+            De::I16x8ExtMulLowI8x16S => En::I16x8ExtMulLowI8x16S,
+            De::I16x8ExtMulHighI8x16S => En::I16x8ExtMulHighI8x16S,
+            De::I16x8ExtMulLowI8x16U => En::I16x8ExtMulLowI8x16U,
+            De::I16x8ExtMulHighI8x16U => En::I16x8ExtMulHighI8x16U,
+            De::I32x4ExtMulLowI16x8S => En::I32x4ExtMulLowI16x8S,
+            De::I32x4ExtMulHighI16x8S => En::I32x4ExtMulHighI16x8S,
+            De::I32x4ExtMulLowI16x8U => En::I32x4ExtMulLowI16x8U,
+            De::I32x4ExtMulHighI16x8U => En::I32x4ExtMulHighI16x8U,
+            De::I64x2ExtMulLowI32x4S => En::I64x2ExtMulLowI32x4S,
+            De::I64x2ExtMulHighI32x4S => En::I64x2ExtMulHighI32x4S,
+            De::I64x2ExtMulLowI32x4U => En::I64x2ExtMulLowI32x4U,
+            De::I64x2ExtMulHighI32x4U => En::I64x2ExtMulHighI32x4U,
+            De::I16x8ExtAddPairwiseI8x16S => En::I16x8ExtAddPairwiseI8x16S,
+            De::I16x8ExtAddPairwiseI8x16U => En::I16x8ExtAddPairwiseI8x16U,
+            De::I32x4ExtAddPairwiseI16x8S => En::I32x4ExtAddPairwiseI16x8S,
+            De::I32x4ExtAddPairwiseI16x8U => En::I32x4ExtAddPairwiseI16x8U,
+            De::I8x16AddSatS => En::I8x16AddSatS,
+            De::I8x16AddSatU => En::I8x16AddSatU,
+            De::I16x8AddSatS => En::I16x8AddSatS,
+            De::I16x8AddSatU => En::I16x8AddSatU,
+            De::I8x16SubSatS => En::I8x16SubSatS,
+            De::I8x16SubSatU => En::I8x16SubSatU,
+            De::I16x8SubSatS => En::I16x8SubSatS,
+            De::I16x8SubSatU => En::I16x8SubSatU,
+            De::I16x8Q15MulrSatS => En::I16x8Q15MulrSatS,
+            De::I8x16MinS => En::I8x16MinS,
+            De::I8x16MinU => En::I8x16MinU,
+            De::I16x8MinS => En::I16x8MinS,
+            De::I16x8MinU => En::I16x8MinU,
+            De::I32x4MinS => En::I32x4MinS,
+            De::I32x4MinU => En::I32x4MinU,
+            De::F32x4Min => En::F32x4Min,
+            De::F64x2Min => En::F64x2Min,
+            De::F32x4PMin => En::F32x4PMin,
+            De::F64x2PMin => En::F64x2PMin,
+            De::I8x16MaxS => En::I8x16MaxS,
+            De::I8x16MaxU => En::I8x16MaxU,
+            De::I16x8MaxS => En::I16x8MaxS,
+            De::I16x8MaxU => En::I16x8MaxU,
+            De::I32x4MaxS => En::I32x4MaxS,
+            De::I32x4MaxU => En::I32x4MaxU,
+            De::F32x4Max => En::F32x4Max,
+            De::F64x2Max => En::F64x2Max,
+            De::F32x4PMax => En::F32x4PMax,
+            De::F64x2PMax => En::F64x2PMax,
+            De::I8x16AvgrU => En::I8x16AvgrU,
+            De::I16x8AvgrU => En::I16x8AvgrU,
+            De::I8x16Abs => En::I8x16Abs,
+            De::I16x8Abs => En::I16x8Abs,
+            De::I32x4Abs => En::I32x4Abs,
+            De::I64x2Abs => En::I64x2Abs,
+            De::F32x4Abs => En::F32x4Abs,
+            De::F64x2Abs => En::F64x2Abs,
+            De::I8x16Shl => En::I8x16Shl,
+            De::I16x8Shl => En::I16x8Shl,
+            De::I32x4Shl => En::I32x4Shl,
+            De::I64x2Shl => En::I64x2Shl,
+            De::I8x16ShrS => En::I8x16ShrS,
+            De::I8x16ShrU => En::I8x16ShrU,
+            De::I16x8ShrS => En::I16x8ShrS,
+            De::I16x8ShrU => En::I16x8ShrU,
+            De::I32x4ShrS => En::I32x4ShrS,
+            De::I32x4ShrU => En::I32x4ShrU,
+            De::I64x2ShrS => En::I64x2ShrS,
+            De::I64x2ShrU => En::I64x2ShrU,
+            De::V128And => En::V128And,
+            De::V128Or => En::V128Or,
+            De::V128Xor => En::V128Xor,
+            De::V128Not => En::V128Not,
+            De::V128AndNot => En::V128AndNot,
+            De::V128Bitselect => En::V128Bitselect,
+            De::I8x16Popcnt => En::I8x16Popcnt,
+            De::V128AnyTrue => En::V128AnyTrue,
+            De::I8x16AllTrue => En::I8x16AllTrue,
+            De::I16x8AllTrue => En::I16x8AllTrue,
+            De::I32x4AllTrue => En::I32x4AllTrue,
+            De::I64x2AllTrue => En::I64x2AllTrue,
+            De::I8x16Bitmask => En::I8x16Bitmask,
+            De::I16x8Bitmask => En::I16x8Bitmask,
+            De::I32x4Bitmask => En::I32x4Bitmask,
+            De::I64x2Bitmask => En::I64x2Bitmask,
+            De::I8x16Eq => En::I8x16Eq,
+            De::I16x8Eq => En::I16x8Eq,
+            De::I32x4Eq => En::I32x4Eq,
+            De::I64x2Eq => En::I64x2Eq,
+            De::F32x4Eq => En::F32x4Eq,
+            De::F64x2Eq => En::F64x2Eq,
+            De::I8x16Ne => En::I8x16Ne,
+            De::I16x8Ne => En::I16x8Ne,
+            De::I32x4Ne => En::I32x4Ne,
+            De::I64x2Ne => En::I64x2Ne,
+            De::F32x4Ne => En::F32x4Ne,
+            De::F64x2Ne => En::F64x2Ne,
+            De::I8x16LtS => En::I8x16LtS,
+            De::I8x16LtU => En::I8x16LtU,
+            De::I16x8LtS => En::I16x8LtS,
+            De::I16x8LtU => En::I16x8LtU,
+            De::I32x4LtS => En::I32x4LtS,
+            De::I32x4LtU => En::I32x4LtU,
+            De::F32x4Lt => En::F32x4Lt,
+            De::F64x2Lt => En::F64x2Lt,
+            De::I8x16LeS => En::I8x16LeS,
+            De::I8x16LeU => En::I8x16LeU,
+            De::I16x8LeS => En::I16x8LeS,
+            De::I16x8LeU => En::I16x8LeU,
+            De::I32x4LeS => En::I32x4LeS,
+            De::I32x4LeU => En::I32x4LeU,
+            De::F32x4Le => En::F32x4Le,
+            De::F64x2Le => En::F64x2Le,
+            De::I8x16GtS => En::I8x16GtS,
+            De::I8x16GtU => En::I8x16GtU,
+            De::I16x8GtS => En::I16x8GtS,
+            De::I16x8GtU => En::I16x8GtU,
+            De::I32x4GtS => En::I32x4GtS,
+            De::I32x4GtU => En::I32x4GtU,
+            De::F32x4Gt => En::F32x4Gt,
+            De::F64x2Gt => En::F64x2Gt,
+            De::I8x16GeS => En::I8x16GeS,
+            De::I8x16GeU => En::I8x16GeU,
+            De::I16x8GeS => En::I16x8GeS,
+            De::I16x8GeU => En::I16x8GeU,
+            De::I32x4GeS => En::I32x4GeS,
+            De::I32x4GeU => En::I32x4GeU,
+            De::F32x4Ge => En::F32x4Ge,
+            De::F64x2Ge => En::F64x2Ge,
+            De::F32x4Div => En::F32x4Div,
+            De::F64x2Div => En::F64x2Div,
+            De::F32x4Sqrt => En::F32x4Sqrt,
+            De::F64x2Sqrt => En::F64x2Sqrt,
+            De::F32x4Ceil => En::F32x4Ceil,
+            De::F64x2Ceil => En::F64x2Ceil,
+            De::F32x4Floor => En::F32x4Floor,
+            De::F64x2Floor => En::F64x2Floor,
+            De::F32x4Trunc => En::F32x4Trunc,
+            De::F64x2Trunc => En::F64x2Trunc,
+            De::F32x4Nearest => En::F32x4Nearest,
+            De::F64x2Nearest => En::F64x2Nearest,
+            De::F32x4ConvertI32x4S => En::F32x4ConvertI32x4S,
+            De::F32x4ConvertI32x4U => En::F32x4ConvertI32x4U,
+            De::F64x2ConvertLowI32x4S => En::F64x2ConvertLowI32x4S,
+            De::F64x2ConvertLowI32x4U => En::F64x2ConvertLowI32x4U,
+            De::I32x4TruncSatF32x4S => En::I32x4TruncSatF32x4S,
+            De::I32x4TruncSatF32x4U => En::I32x4TruncSatF32x4U,
+            De::I32x4TruncSatF64x2SZero => En::I32x4TruncSatF64x2SZero,
+            De::I32x4TruncSatF64x2UZero => En::I32x4TruncSatF64x2UZero,
+            De::F32x4DemoteF64x2Zero => En::F32x4DemoteF64x2Zero,
+            De::F64x2PromoteLowF32x4 => En::F64x2PromoteLowF32x4,
+            De::I8x16NarrowI16x8S => En::I8x16NarrowI16x8S,
+            De::I8x16NarrowI16x8U => En::I8x16NarrowI16x8U,
+            De::I16x8NarrowI32x4S => En::I16x8NarrowI32x4S,
+            De::I16x8NarrowI32x4U => En::I16x8NarrowI32x4U,
+            De::I16x8ExtendLowI8x16S => En::I16x8ExtendLowI8x16S,
+            De::I16x8ExtendHighI8x16S => En::I16x8ExtendHighI8x16S,
+            De::I16x8ExtendLowI8x16U => En::I16x8ExtendLowI8x16U,
+            De::I16x8ExtendHighI8x16U => En::I16x8ExtendHighI8x16U,
+            De::I32x4ExtendLowI16x8S => En::I32x4ExtendLowI16x8S,
+            De::I32x4ExtendHighI16x8S => En::I32x4ExtendHighI16x8S,
+            De::I32x4ExtendLowI16x8U => En::I32x4ExtendLowI16x8U,
+            De::I32x4ExtendHighI16x8U => En::I32x4ExtendHighI16x8U,
+            De::I64x2ExtendLowI32x4S => En::I64x2ExtendLowI32x4S,
+            De::I64x2ExtendHighI32x4S => En::I64x2ExtendHighI32x4S,
+            De::I64x2ExtendLowI32x4U => En::I64x2ExtendLowI32x4U,
+            De::I64x2ExtendHighI32x4U => En::I64x2ExtendHighI32x4U,
+
+            De::I8x16Shuffle { lanes } => En::I8x16Shuffle(lanes),
+
             other => bail!("Unsupported instruction {:?}", other),
         });
     }

uw8-window/.gitignore

@@ -0,0 +1 @@
+/target

uw8-window/Cargo.toml

@@ -0,0 +1,18 @@
+[package]
+name = "uw8-window"
+version = "0.1.0"
+edition = "2021"
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+winit = "0.29.15"
+env_logger = "0.11.3"
+log = "0.4"
+pico-args = "0.5"
+wgpu = "0.19.3"
+pollster = "0.3.0"
+bytemuck = { version = "1.15", features = [ "derive" ] }
+anyhow = "1"
+minifb = { version = "0.25.0", default-features = false, features = ["x11"] }
+winapi = { version = "0.3.9", features = [ "timeapi" ] }

uw8-window/src/cpu.rs

@@ -0,0 +1,83 @@
+use std::time::Instant;
+
+use crate::{Input, WindowImpl};
+use anyhow::Result;
+use minifb::{Key, WindowOptions};
+
+static GAMEPAD_KEYS: &[Key] = &[
+    Key::Up,
+    Key::Down,
+    Key::Left,
+    Key::Right,
+    Key::Z,
+    Key::X,
+    Key::A,
+    Key::S,
+];
+
+pub struct Window {
+    window: minifb::Window,
+    buffer: Vec<u32>,
+}
+
+impl Window {
+    pub fn new() -> Result<Window> {
+        #[cfg(target_os = "windows")]
+        unsafe {
+            winapi::um::timeapi::timeBeginPeriod(1);
+        }
+
+        let buffer: Vec<u32> = vec![0; 320 * 240];
+
+        let options = WindowOptions {
+            scale: minifb::Scale::X2,
+            scale_mode: minifb::ScaleMode::AspectRatioStretch,
+            resize: true,
+            ..Default::default()
+        };
+        let window = minifb::Window::new("MicroW8", 320, 240, options).unwrap();
+
+        Ok(Window { window, buffer })
+    }
+}
+
+impl WindowImpl for Window {
+    fn begin_frame(&mut self) -> Input {
+        let mut gamepads = [0u8; 4];
+        for key in self.window.get_keys() {
+            if let Some(index) = GAMEPAD_KEYS
+                .iter()
+                .enumerate()
+                .find(|(_, &k)| k == key)
+                .map(|(i, _)| i)
+            {
+                gamepads[0] |= 1 << index;
+            }
+        }
+
+        Input {
+            gamepads,
+            reset: self.window.is_key_pressed(Key::R, minifb::KeyRepeat::No),
+        }
+    }
+
+    fn end_frame(&mut self, framebuffer: &[u8], palette: &[u8], next_frame: Instant) {
+        for (i, &color_index) in framebuffer.iter().enumerate() {
+            let offset = color_index as usize * 4;
+            self.buffer[i] = 0xff000000
+                | ((palette[offset] as u32) << 16)
+                | ((palette[offset + 1] as u32) << 8)
+                | palette[offset + 2] as u32;
+        }
+        self.window
+            .update_with_buffer(&self.buffer, 320, 240)
+            .unwrap();
+        if let Some(sleep) = next_frame.checked_duration_since(Instant::now()) {
+            std::thread::sleep(sleep);
+        }
+    }
+
+    fn is_open(&self) -> bool {
+        self.window.is_open() && !self.window.is_key_down(Key::Escape)
+    }
+}

uw8-window/src/gpu/crt.rs

@@ -0,0 +1,143 @@
+use wgpu::util::DeviceExt;
+use winit::dpi::PhysicalSize;
+
+use super::Filter;
+
+pub struct CrtFilter {
+    uniform_buffer: wgpu::Buffer,
+    bind_group: wgpu::BindGroup,
+    pipeline: wgpu::RenderPipeline,
+}
+
+impl CrtFilter {
+    pub fn new(
+        device: &wgpu::Device,
+        screen: &wgpu::TextureView,
+        resolution: PhysicalSize<u32>,
+        surface_format: wgpu::TextureFormat,
+    ) -> CrtFilter {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(resolution),
+        };
+
+        let uniform_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: None,
+            contents: bytemuck::cast_slice(&[uniforms]),
+            usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+        });
+
+        let crt_bind_group_layout =
+            device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+                entries: &[
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 0,
+                        visibility: wgpu::ShaderStages::FRAGMENT,
+                        ty: wgpu::BindingType::Texture {
+                            multisampled: false,
+                            view_dimension: wgpu::TextureViewDimension::D2,
+                            sample_type: wgpu::TextureSampleType::Float { filterable: false },
+                        },
+                        count: None,
+                    },
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 1,
+                        visibility: wgpu::ShaderStages::VERTEX_FRAGMENT,
+                        ty: wgpu::BindingType::Buffer {
+                            ty: wgpu::BufferBindingType::Uniform,
+                            has_dynamic_offset: false,
+                            min_binding_size: None,
+                        },
+                        count: None,
+                    },
+                ],
+                label: None,
+            });
+
+        let crt_bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &crt_bind_group_layout,
+            entries: &[
+                wgpu::BindGroupEntry {
+                    binding: 0,
+                    resource: wgpu::BindingResource::TextureView(&screen),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 1,
+                    resource: uniform_buffer.as_entire_binding(),
+                },
+            ],
+            label: None,
+        });
+
+        let crt_shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: None,
+            source: wgpu::ShaderSource::Wgsl(include_str!("crt.wgsl").into()),
+        });
+
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: None,
+                bind_group_layouts: &[&crt_bind_group_layout],
+                push_constant_ranges: &[],
+            });
+
+        let render_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: None,
+            layout: Some(&render_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &crt_shader,
+                entry_point: "vs_main",
+                buffers: &[],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &crt_shader,
+                entry_point: "fs_main",
+                targets: &[Some(wgpu::ColorTargetState {
+                    format: surface_format,
+                    blend: None,
+                    write_mask: wgpu::ColorWrites::ALL,
+                })],
+            }),
+            primitive: Default::default(),
+            depth_stencil: None,
+            multisample: Default::default(),
+            multiview: None,
+        });
+
+        CrtFilter {
+            uniform_buffer,
+            bind_group: crt_bind_group,
+            pipeline: render_pipeline,
+        }
+    }
+}
+
+impl Filter for CrtFilter {
+    fn resize(&mut self, queue: &wgpu::Queue, new_size: PhysicalSize<u32>) {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(new_size),
+        };
+        queue.write_buffer(&self.uniform_buffer, 0, bytemuck::cast_slice(&[uniforms]));
+    }
+
+    fn render<'a>(&'a self, render_pass: &mut wgpu::RenderPass<'a>) {
+        render_pass.set_pipeline(&self.pipeline);
+        render_pass.set_bind_group(0, &self.bind_group, &[]);
+        render_pass.draw(0..6, 0..1);
+    }
+}
+
+fn texture_scale_from_resolution(res: PhysicalSize<u32>) -> [f32; 4] {
+    let scale = ((res.width as f32) / 160.0).min((res.height as f32) / 120.0);
+    [
+        res.width as f32 / scale,
+        res.height as f32 / scale,
+        2.0 / scale,
+        0.0,
+    ]
+}
+
+#[repr(C)]
+#[derive(Debug, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
+struct Uniforms {
+    texture_scale: [f32; 4],
+}

uw8-window/src/gpu/crt.wgsl

@@ -0,0 +1,75 @@
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) tex_coords: vec2<f32>,
+}
+
+struct Uniforms {
+    texture_scale: vec4<f32>,
+}
+
+@group(0) @binding(1) var<uniform> uniforms: Uniforms;
+
+@vertex
+fn vs_main(
+    @builtin(vertex_index) in_vertex_index: u32,
+) -> VertexOutput {
+    var out: VertexOutput;
+    let i = in_vertex_index / 3u + in_vertex_index % 3u;
+    let x = -1.0 + f32(i % 2u) * 322.0;
+    let y = -1.0 + f32(i / 2u) * 242.0;
+    out.clip_position = vec4<f32>((vec2<f32>(x, y) - vec2<f32>(160.0, 120.0)) / uniforms.texture_scale.xy, 0.0, 1.0);
+    out.tex_coords = vec2<f32>(x, y);
+    return out;
+}
+
+@group(0) @binding(0) var screen_texture: texture_2d<f32>;
+
+fn sample_pixel(coords: vec2<i32>, offset: vec4<f32>) -> vec3<f32> {
+    let is_outside = any(vec2<u32>(coords) >= vec2<u32>(320u, 240u));
+    if(is_outside) {
+        return vec3<f32>(0.0);
+    } else {
+        let f = max(vec4<f32>(0.008) / offset - vec4<f32>(0.0024), vec4<f32>(0.0));
+        return textureLoad(screen_texture, coords, 0).rgb * (f.x + f.y + f.z + f.w);
+    }
+}
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    let pixelf = floor(in.tex_coords);
+    let o = vec2<f32>(0.5) - (in.tex_coords - pixelf);
+    let pixel = vec2<i32>(pixelf);
+    
+    let offset_x = o.xxxx + vec4<f32>(-0.125, 0.375, 0.125, -0.375) * uniforms.texture_scale.z;
+    let offset_y = o.yyyy + vec4<f32>(-0.375, -0.125, 0.375, 0.125) * uniforms.texture_scale.z;
+    
+    var offset_x0 = max(abs(offset_x + vec4<f32>(-1.0)) - vec4<f32>(0.5), vec4<f32>(0.0));
+    var offset_x1 = max(abs(offset_x) - vec4<f32>(0.5), vec4<f32>(0.0));
+    var offset_x2 = max(abs(offset_x + vec4<f32>(1.0)) - vec4<f32>(0.5), vec4<f32>(0.0));
+    
+    offset_x0 = offset_x0 * offset_x0;
+    offset_x1 = offset_x1 * offset_x1;
+    offset_x2 = offset_x2 * offset_x2;
+    
+    var offset_yr = offset_y + vec4<f32>(-1.0);
+    offset_yr = vec4<f32>(0.02) + offset_yr * offset_yr;
+    
+    var acc = sample_pixel(pixel + vec2<i32>(-1, -1), offset_x0 + offset_yr);
+    acc = acc + sample_pixel(pixel + vec2<i32>(0, -1), offset_x1 + offset_yr);
+    acc = acc + sample_pixel(pixel + vec2<i32>(1, -1), offset_x2 + offset_yr);
+
+    offset_yr = vec4<f32>(0.02) + offset_y * offset_y;
+    
+    acc = acc + sample_pixel(pixel + vec2<i32>(-1, 0), offset_x0 + offset_yr);
+    acc = acc + sample_pixel(pixel, offset_x1 + offset_yr);
+    acc = acc + sample_pixel(pixel + vec2<i32>(1, 0), offset_x2 + offset_yr);
+
+    offset_yr = offset_y + vec4<f32>(1.0);
+    offset_yr = vec4<f32>(0.02) + offset_yr * offset_yr;
+    
+    acc = acc + sample_pixel(pixel + vec2<i32>(-1, 1), offset_x0 + offset_yr);
+    acc = acc + sample_pixel(pixel + vec2<i32>(0, 1), offset_x1 + offset_yr);
+    acc = acc + sample_pixel(pixel + vec2<i32>(1, 1), offset_x2 + offset_yr);
+
+    return vec4<f32>(acc, 1.0);
+}

uw8-window/src/gpu/fast_crt.rs

@@ -0,0 +1,162 @@
+use wgpu::util::DeviceExt;
+use winit::dpi::PhysicalSize;
+
+use super::Filter;
+
+pub struct FastCrtFilter {
+    uniform_buffer: wgpu::Buffer,
+    bind_group: wgpu::BindGroup,
+    pipeline: wgpu::RenderPipeline,
+}
+
+impl FastCrtFilter {
+    pub fn new(
+        device: &wgpu::Device,
+        screen: &wgpu::TextureView,
+        resolution: PhysicalSize<u32>,
+        surface_format: wgpu::TextureFormat,
+        chromatic: bool,
+    ) -> FastCrtFilter {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(resolution),
+        };
+
+        let uniform_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: None,
+            contents: bytemuck::cast_slice(&[uniforms]),
+            usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+        });
+
+        let bind_group_layout = device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+            entries: &[
+                wgpu::BindGroupLayoutEntry {
+                    binding: 0,
+                    visibility: wgpu::ShaderStages::FRAGMENT,
+                    ty: wgpu::BindingType::Texture {
+                        multisampled: false,
+                        view_dimension: wgpu::TextureViewDimension::D2,
+                        sample_type: wgpu::TextureSampleType::Float { filterable: true },
+                    },
+                    count: None,
+                },
+                wgpu::BindGroupLayoutEntry {
+                    binding: 1,
+                    visibility: wgpu::ShaderStages::FRAGMENT,
+                    ty: wgpu::BindingType::Sampler(wgpu::SamplerBindingType::Filtering),
+                    count: None,
+                },
+                wgpu::BindGroupLayoutEntry {
+                    binding: 2,
+                    visibility: wgpu::ShaderStages::VERTEX_FRAGMENT,
+                    ty: wgpu::BindingType::Buffer {
+                        ty: wgpu::BufferBindingType::Uniform,
+                        has_dynamic_offset: false,
+                        min_binding_size: None,
+                    },
+                    count: None,
+                },
+            ],
+            label: None,
+        });
+
+        let sampler = device.create_sampler(&wgpu::SamplerDescriptor {
+            mag_filter: wgpu::FilterMode::Linear,
+            ..Default::default()
+        });
+
+        let bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &bind_group_layout,
+            entries: &[
+                wgpu::BindGroupEntry {
+                    binding: 0,
+                    resource: wgpu::BindingResource::TextureView(&screen),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 1,
+                    resource: wgpu::BindingResource::Sampler(&sampler),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 2,
+                    resource: uniform_buffer.as_entire_binding(),
+                },
+            ],
+            label: None,
+        });
+
+        let shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: None,
+            source: wgpu::ShaderSource::Wgsl(include_str!("fast_crt.wgsl").into()),
+        });
+
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: None,
+                bind_group_layouts: &[&bind_group_layout],
+                push_constant_ranges: &[],
+            });
+
+        let render_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: None,
+            layout: Some(&render_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &shader,
+                entry_point: "vs_main",
+                buffers: &[],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &shader,
+                entry_point: if chromatic {
+                    "fs_main_chromatic"
+                } else {
+                    "fs_main"
+                },
+                targets: &[Some(wgpu::ColorTargetState {
+                    format: surface_format,
+                    blend: None,
+                    write_mask: wgpu::ColorWrites::ALL,
+                })],
+            }),
+            primitive: Default::default(),
+            depth_stencil: None,
+            multisample: Default::default(),
+            multiview: None,
+        });
+
+        FastCrtFilter {
+            uniform_buffer,
+            bind_group,
+            pipeline: render_pipeline,
+        }
+    }
+}
+
+impl Filter for FastCrtFilter {
+    fn resize(&mut self, queue: &wgpu::Queue, new_size: PhysicalSize<u32>) {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(new_size),
+        };
+        queue.write_buffer(&self.uniform_buffer, 0, bytemuck::cast_slice(&[uniforms]));
+    }
+
+    fn render<'a>(&'a self, render_pass: &mut wgpu::RenderPass<'a>) {
+        render_pass.set_pipeline(&self.pipeline);
+        render_pass.set_bind_group(0, &self.bind_group, &[]);
+        render_pass.draw(0..6, 0..1);
+    }
+}
+
+fn texture_scale_from_resolution(res: PhysicalSize<u32>) -> [f32; 4] {
+    let scale = ((res.width as f32) / 160.0).min((res.height as f32) / 120.0);
+    [
+        scale / res.width as f32,
+        scale / res.height as f32,
+        2.0 / scale,
+        0.0,
+    ]
+}
+
+#[repr(C)]
+#[derive(Debug, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
+struct Uniforms {
+    texture_scale: [f32; 4],
+}

uw8-window/src/gpu/fast_crt.wgsl

@@ -0,0 +1,66 @@
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) tex_coords: vec2<f32>,
+}
+
+struct Uniforms {
+    texture_scale: vec4<f32>,
+}
+
+@group(0) @binding(2) var<uniform> uniforms: Uniforms;
+
+@vertex
+fn vs_main(
+    @builtin(vertex_index) in_vertex_index: u32,
+) -> VertexOutput {
+    var out: VertexOutput;
+    let i = in_vertex_index / 3u + in_vertex_index % 3u;
+    let x = 0.0 + f32(i % 2u) * 320.0;
+    let y = 0.0 + f32(i / 2u) * 240.0;
+    out.clip_position = vec4<f32>((vec2<f32>(x, y) - vec2<f32>(160.0, 120.0)) * uniforms.texture_scale.xy, 0.0, 1.0);
+    out.tex_coords = vec2<f32>(x, y);
+    return out;
+}
+
+@group(0) @binding(0) var screen_texture: texture_2d<f32>;
+@group(0) @binding(1) var linear_sampler: sampler;
+
+fn row_factor(offset: f32) -> f32 {
+    return 1.0 / (1.0 + offset * offset * 16.0);
+}
+
+fn col_factor(offset: f32) -> f32 {
+    let o = max(0.0, abs(offset) - 0.4);
+    return 1.0 / (1.0 + o * o * 16.0);
+}
+
+fn sample_screen(tex_coords: vec2<f32>) -> vec4<f32> {
+    let base = round(tex_coords) - vec2<f32>(0.5);
+    let frac = tex_coords - base;
+    
+    let top_factor = row_factor(frac.y);
+    let bottom_factor = row_factor(frac.y - 1.0);
+    
+    let v = base.y + bottom_factor / (bottom_factor + top_factor);
+    
+    let left_factor = col_factor(frac.x);
+    let right_factor = col_factor(frac.x - 1.0);
+    
+    let u = base.x + right_factor / (right_factor + left_factor);
+    
+    return textureSample(screen_texture, linear_sampler, vec2<f32>(u, v) / vec2<f32>(320.0, 240.0)) * (top_factor + bottom_factor) * (left_factor + right_factor) * 1.1;
+}
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    return sample_screen(in.tex_coords);
+}
+
+@fragment
+fn fs_main_chromatic(in: VertexOutput) -> @location(0) vec4<f32> {
+    let r = sample_screen(in.tex_coords + vec2<f32>(0.2, 0.2)).r;
+    let g = sample_screen(in.tex_coords + vec2<f32>(0.07, -0.27)).g;
+    let b = sample_screen(in.tex_coords + vec2<f32>(-0.27, 0.07)).b;
+    return vec4<f32>(r, g, b, 1.0);
+}
+

uw8-window/src/gpu/mod.rs

@@ -0,0 +1,565 @@
+use crate::{Input, WindowConfig, WindowImpl};
+use anyhow::{anyhow, Result};
+use std::{sync::Arc, time::Instant};
+
+use winit::{
+    dpi::PhysicalSize,
+    event::{Event, WindowEvent},
+    event_loop::{ControlFlow, EventLoop},
+    keyboard::{Key, KeyCode, NamedKey, PhysicalKey},
+    platform::pump_events::{EventLoopExtPumpEvents, PumpStatus},
+    window::{Fullscreen, WindowBuilder},
+};
+
+mod crt;
+mod fast_crt;
+mod square;
+
+use crt::CrtFilter;
+use fast_crt::FastCrtFilter;
+use square::SquareFilter;
+
+pub struct Window {
+    _instance: wgpu::Instance,
+    surface: wgpu::Surface<'static>,
+    _adapter: wgpu::Adapter,
+    device: wgpu::Device,
+    queue: wgpu::Queue,
+    palette_screen_mode: PaletteScreenMode,
+    surface_config: wgpu::SurfaceConfiguration,
+    filter: Box<dyn Filter>,
+    event_loop: EventLoop<()>,
+    window: Arc<winit::window::Window>,
+    gamepads: [u8; 4],
+    next_frame: Instant,
+    is_fullscreen: bool,
+    is_open: bool,
+}
+
+impl Window {
+    pub fn new(window_config: WindowConfig) -> Result<Window> {
+        async fn create(window_config: WindowConfig) -> Result<Window> {
+            let event_loop = EventLoop::new()?;
+            let window = WindowBuilder::new()
+                .with_inner_size(PhysicalSize::new(
+                    (320. * window_config.scale).round() as u32,
+                    (240. * window_config.scale).round() as u32,
+                ))
+                .with_min_inner_size(PhysicalSize::new(320u32, 240))
+                .with_title("MicroW8")
+                .with_fullscreen(if window_config.fullscreen {
+                    Some(Fullscreen::Borderless(None))
+                } else {
+                    None
+                })
+                .build(&event_loop)?;
+
+            window.set_cursor_visible(false);
+
+            let window = Arc::new(window);
+
+            let instance = wgpu::Instance::new(Default::default());
+            let surface = instance.create_surface(window.clone())?;
+            let adapter = instance
+                .request_adapter(&wgpu::RequestAdapterOptions {
+                    power_preference: wgpu::PowerPreference::LowPower,
+                    compatible_surface: Some(&surface),
+                    force_fallback_adapter: false,
+                })
+                .await
+                .ok_or_else(|| anyhow!("Request adapter failed"))?;
+
+            let (device, queue) = adapter
+                .request_device(&wgpu::DeviceDescriptor::default(), None)
+                .await?;
+
+            let palette_screen_mode = PaletteScreenMode::new(&device);
+
+            let surface_config = wgpu::SurfaceConfiguration {
+                present_mode: wgpu::PresentMode::AutoNoVsync,
+                ..surface
+                    .get_default_config(
+                        &adapter,
+                        window.inner_size().width,
+                        window.inner_size().height,
+                    )
+                    .expect("Surface incompatible with adapter")
+            };
+
+            let filter: Box<dyn Filter> = create_filter(
+                &device,
+                &palette_screen_mode.screen_view,
+                window.inner_size(),
+                surface_config.format,
+                window_config.filter,
+            );
+
+            surface.configure(&device, &surface_config);
+
+            Ok(Window {
+                event_loop,
+                window,
+                _instance: instance,
+                surface,
+                _adapter: adapter,
+                device,
+                queue,
+                palette_screen_mode,
+                surface_config,
+                filter,
+                gamepads: [0; 4],
+                next_frame: Instant::now(),
+                is_fullscreen: window_config.fullscreen,
+                is_open: true,
+            })
+        }
+
+        pollster::block_on(create(window_config))
+    }
+}
+
+impl WindowImpl for Window {
+    fn begin_frame(&mut self) -> Input {
+        let mut reset = false;
+        self.event_loop
+            .set_control_flow(ControlFlow::WaitUntil(self.next_frame));
+        while self.is_open {
+            let timeout = self.next_frame.saturating_duration_since(Instant::now());
+            let status = self.event_loop.pump_events(Some(timeout), |event, elwt| {
+                let mut new_filter = None;
+                match event {
+                    Event::WindowEvent { event, .. } => match event {
+                        WindowEvent::Resized(new_size) => {
+                            self.surface_config.width = new_size.width;
+                            self.surface_config.height = new_size.height;
+                            self.surface.configure(&self.device, &self.surface_config);
+                            self.filter.resize(&self.queue, new_size);
+                        }
+                        WindowEvent::CloseRequested => {
+                            elwt.exit();
+                        }
+                        WindowEvent::KeyboardInput { event, .. } => {
+                            fn gamepad_button(input: &winit::event::KeyEvent) -> u8 {
+                                match input.physical_key {
+                                    PhysicalKey::Code(KeyCode::KeyZ) => 16,
+                                    PhysicalKey::Code(KeyCode::KeyX) => 32,
+                                    PhysicalKey::Code(KeyCode::KeyA) => 64,
+                                    PhysicalKey::Code(KeyCode::KeyS) => 128,
+                                    _ => match input.logical_key {
+                                        Key::Named(NamedKey::ArrowUp) => 1,
+                                        Key::Named(NamedKey::ArrowDown) => 2,
+                                        Key::Named(NamedKey::ArrowLeft) => 4,
+                                        Key::Named(NamedKey::ArrowRight) => 8,
+                                        _ => 0,
+                                    },
+                                }
+                            }
+                            if event.state == winit::event::ElementState::Pressed {
+                                match event.logical_key {
+                                    Key::Named(NamedKey::Escape) => {
+                                        elwt.exit();
+                                    }
+                                    Key::Character(ref c) => match c.as_str() {
+                                        "f" => {
+                                            let fullscreen = if self.window.fullscreen().is_some() {
+                                                None
+                                            } else {
+                                                Some(Fullscreen::Borderless(None))
+                                            };
+                                            self.is_fullscreen = fullscreen.is_some();
+                                            self.window.set_fullscreen(fullscreen);
+                                        }
+                                        "r" => reset = true,
+                                        "1" => new_filter = Some(1),
+                                        "2" => new_filter = Some(2),
+                                        "3" => new_filter = Some(3),
+                                        "4" => new_filter = Some(4),
+                                        "5" => new_filter = Some(5),
+                                        _ => (),
+                                    },
+                                    _ => (),
+                                }
+
+                                self.gamepads[0] |= gamepad_button(&event);
+                            } else {
+                                self.gamepads[0] &= !gamepad_button(&event);
+                            }
+                        }
+                        _ => (),
+                    },
+                    _ => (),
+                }
+                if let Some(new_filter) = new_filter {
+                    self.filter = create_filter(
+                        &self.device,
+                        &self.palette_screen_mode.screen_view,
+                        self.window.inner_size(),
+                        self.surface_config.format,
+                        new_filter,
+                    );
+                }
+            });
+            match status {
+                PumpStatus::Exit(_) => self.is_open = false,
+                _ => (),
+            }
+
+            if Instant::now() >= self.next_frame {
+                break;
+            }
+        }
+        Input {
+            gamepads: self.gamepads,
+            reset,
+        }
+    }
+
+    fn end_frame(&mut self, framebuffer: &[u8], palette: &[u8], next_frame: Instant) {
+        self.next_frame = next_frame;
+        self.palette_screen_mode
+            .write_framebuffer(&self.queue, framebuffer);
+        self.palette_screen_mode.write_palette(&self.queue, palette);
+
+        let output = self.surface.get_current_texture().unwrap();
+        let view = output
+            .texture
+            .create_view(&wgpu::TextureViewDescriptor::default());
+        let mut encoder = self
+            .device
+            .create_command_encoder(&wgpu::CommandEncoderDescriptor { label: None });
+
+        self.palette_screen_mode.resolve_screen(&mut encoder);
+
+        {
+            let mut render_pass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
+                label: None,
+                color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                    view: &view,
+                    resolve_target: None,
+                    ops: wgpu::Operations {
+                        load: wgpu::LoadOp::Clear(wgpu::Color {
+                            r: 0.0,
+                            g: 0.0,
+                            b: 0.0,
+                            a: 1.0,
+                        }),
+                        store: wgpu::StoreOp::Store,
+                    },
+                })],
+                depth_stencil_attachment: None,
+                timestamp_writes: None,
+                occlusion_query_set: None,
+            });
+
+            self.filter.render(&mut render_pass);
+        }
+
+        self.queue.submit(std::iter::once(encoder.finish()));
+        output.present();
+    }
+
+    fn is_open(&self) -> bool {
+        self.is_open
+    }
+}
+
+fn create_filter(
+    device: &wgpu::Device,
+    screen_texture: &wgpu::TextureView,
+    window_size: PhysicalSize<u32>,
+    surface_format: wgpu::TextureFormat,
+    filter: u32,
+) -> Box<dyn Filter> {
+    match filter {
+        1 => Box::new(SquareFilter::new(
+            device,
+            screen_texture,
+            window_size,
+            surface_format,
+        )),
+        2 => Box::new(FastCrtFilter::new(
+            device,
+            screen_texture,
+            window_size,
+            surface_format,
+            false,
+        )),
+        3 => Box::new(CrtFilter::new(
+            device,
+            screen_texture,
+            window_size,
+            surface_format,
+        )),
+        4 => Box::new(FastCrtFilter::new(
+            device,
+            screen_texture,
+            window_size,
+            surface_format,
+            true,
+        )),
+        _ => Box::new(AutoCrtFilter::new(
+            device,
+            screen_texture,
+            window_size,
+            surface_format,
+        )),
+    }
+}
+
+trait Filter {
+    fn resize(&mut self, queue: &wgpu::Queue, new_size: PhysicalSize<u32>);
+    fn render<'a>(&'a self, render_pass: &mut wgpu::RenderPass<'a>);
+}
+
+struct AutoCrtFilter {
+    small: CrtFilter,
+    large: FastCrtFilter,
+    resolution: PhysicalSize<u32>,
+}
+
+impl AutoCrtFilter {
+    fn new(
+        device: &wgpu::Device,
+        screen: &wgpu::TextureView,
+        resolution: PhysicalSize<u32>,
+        surface_format: wgpu::TextureFormat,
+    ) -> AutoCrtFilter {
+        let small = CrtFilter::new(device, screen, resolution, surface_format);
+        let large = FastCrtFilter::new(device, screen, resolution, surface_format, true);
+        AutoCrtFilter {
+            small,
+            large,
+            resolution,
+        }
+    }
+}
+
+impl Filter for AutoCrtFilter {
+    fn resize(&mut self, queue: &wgpu::Queue, new_size: PhysicalSize<u32>) {
+        self.small.resize(queue, new_size);
+        self.large.resize(queue, new_size);
+        self.resolution = new_size;
+    }
+
+    fn render<'a>(&'a self, render_pass: &mut wgpu::RenderPass<'a>) {
+        if self.resolution.width < 960 || self.resolution.height < 720 {
+            self.small.render(render_pass);
+        } else {
+            self.large.render(render_pass);
+        }
+    }
+}
+
+struct PaletteScreenMode {
+    framebuffer: wgpu::Texture,
+    palette: wgpu::Texture,
+    screen_view: wgpu::TextureView,
+    bind_group: wgpu::BindGroup,
+    pipeline: wgpu::RenderPipeline,
+}
+
+impl PaletteScreenMode {
+    fn new(device: &wgpu::Device) -> PaletteScreenMode {
+        let framebuffer_texture = device.create_texture(&wgpu::TextureDescriptor {
+            size: wgpu::Extent3d {
+                width: 320,
+                height: 240,
+                depth_or_array_layers: 1,
+            },
+            mip_level_count: 1,
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D2,
+            format: wgpu::TextureFormat::R8Uint,
+            usage: wgpu::TextureUsages::TEXTURE_BINDING | wgpu::TextureUsages::COPY_DST,
+            label: None,
+            view_formats: &[],
+        });
+
+        let palette_texture = device.create_texture(&wgpu::TextureDescriptor {
+            size: wgpu::Extent3d {
+                width: 256,
+                height: 1,
+                depth_or_array_layers: 1,
+            },
+            mip_level_count: 1,
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D1,
+            format: wgpu::TextureFormat::Rgba8UnormSrgb,
+            usage: wgpu::TextureUsages::TEXTURE_BINDING | wgpu::TextureUsages::COPY_DST,
+            label: None,
+            view_formats: &[],
+        });
+
+        let screen_texture = device.create_texture(&wgpu::TextureDescriptor {
+            size: wgpu::Extent3d {
+                width: 320,
+                height: 240,
+                depth_or_array_layers: 1,
+            },
+            mip_level_count: 1,
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D2,
+            format: wgpu::TextureFormat::Rgba8UnormSrgb,
+            usage: wgpu::TextureUsages::TEXTURE_BINDING | wgpu::TextureUsages::RENDER_ATTACHMENT,
+            label: None,
+            view_formats: &[],
+        });
+
+        let framebuffer_texture_view =
+            framebuffer_texture.create_view(&wgpu::TextureViewDescriptor::default());
+        let palette_texture_view =
+            palette_texture.create_view(&wgpu::TextureViewDescriptor::default());
+        let screen_texture_view =
+            screen_texture.create_view(&wgpu::TextureViewDescriptor::default());
+
+        let palette_bind_group_layout =
+            device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+                entries: &[
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 0,
+                        visibility: wgpu::ShaderStages::FRAGMENT,
+                        ty: wgpu::BindingType::Texture {
+                            multisampled: false,
+                            view_dimension: wgpu::TextureViewDimension::D2,
+                            sample_type: wgpu::TextureSampleType::Uint,
+                        },
+                        count: None,
+                    },
+                    wgpu::BindGroupLayoutEntry {
+                        binding: 1,
+                        visibility: wgpu::ShaderStages::FRAGMENT,
+                        ty: wgpu::BindingType::Texture {
+                            multisampled: false,
+                            view_dimension: wgpu::TextureViewDimension::D1,
+                            sample_type: wgpu::TextureSampleType::Float { filterable: false },
+                        },
+                        count: None,
+                    },
+                ],
+                label: None,
+            });
+
+        let palette_bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &palette_bind_group_layout,
+            entries: &[
+                wgpu::BindGroupEntry {
+                    binding: 0,
+                    resource: wgpu::BindingResource::TextureView(&framebuffer_texture_view),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 1,
+                    resource: wgpu::BindingResource::TextureView(&palette_texture_view),
+                },
+            ],
+            label: None,
+        });
+
+        let palette_shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: None,
+            source: wgpu::ShaderSource::Wgsl(include_str!("palette.wgsl").into()),
+        });
+
+        let palette_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: None,
+                bind_group_layouts: &[&palette_bind_group_layout],
+                push_constant_ranges: &[],
+            });
+
+        let palette_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: None,
+            layout: Some(&palette_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &palette_shader,
+                entry_point: "vs_main",
+                buffers: &[],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &palette_shader,
+                entry_point: "fs_main",
+                targets: &[Some(wgpu::ColorTargetState {
+                    format: wgpu::TextureFormat::Rgba8UnormSrgb,
+                    blend: None,
+                    write_mask: wgpu::ColorWrites::ALL,
+                })],
+            }),
+            primitive: Default::default(),
+            depth_stencil: None,
+            multisample: Default::default(),
+            multiview: None,
+        });
+
+        PaletteScreenMode {
+            framebuffer: framebuffer_texture,
+            palette: palette_texture,
+            screen_view: screen_texture_view,
+            bind_group: palette_bind_group,
+            pipeline: palette_pipeline,
+        }
+    }
+
+    fn write_framebuffer(&self, queue: &wgpu::Queue, pixels: &[u8]) {
+        queue.write_texture(
+            wgpu::ImageCopyTexture {
+                texture: &self.framebuffer,
+                mip_level: 0,
+                origin: wgpu::Origin3d::ZERO,
+                aspect: wgpu::TextureAspect::All,
+            },
+            &bytemuck::cast_slice(pixels),
+            wgpu::ImageDataLayout {
+                offset: 0,
+                bytes_per_row: Some(320),
+                rows_per_image: None,
+            },
+            wgpu::Extent3d {
+                width: 320,
+                height: 240,
+                depth_or_array_layers: 1,
+            },
+        );
+    }
+
+    fn write_palette(&self, queue: &wgpu::Queue, palette: &[u8]) {
+        queue.write_texture(
+            wgpu::ImageCopyTexture {
+                texture: &self.palette,
+                mip_level: 0,
+                origin: wgpu::Origin3d::ZERO,
+                aspect: wgpu::TextureAspect::All,
+            },
+            &bytemuck::cast_slice(palette),
+            wgpu::ImageDataLayout {
+                offset: 0,
+                bytes_per_row: Some(256 * 4),
+                rows_per_image: None,
+            },
+            wgpu::Extent3d {
+                width: 256,
+                height: 1,
+                depth_or_array_layers: 1,
+            },
+        );
+    }
+
+    fn resolve_screen(&self, encoder: &mut wgpu::CommandEncoder) {
+        let mut render_pass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
+            label: None,
+            color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                view: &self.screen_view,
+                resolve_target: None,
+                ops: wgpu::Operations {
+                    load: wgpu::LoadOp::Load,
+                    store: wgpu::StoreOp::Store,
+                },
+            })],
+            depth_stencil_attachment: None,
+            timestamp_writes: None,
+            occlusion_query_set: None,
+        });
+
+        render_pass.set_pipeline(&self.pipeline);
+        render_pass.set_bind_group(0, &self.bind_group, &[]);
+        render_pass.draw(0..3, 0..1);
+    }
+}

uw8-window/src/gpu/palette.wgsl

@@ -0,0 +1,24 @@
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) tex_coords: vec2<f32>,
+}
+
+@vertex
+fn vs_main(@builtin(vertex_index) vertex_index: u32) -> VertexOutput {
+    var out: VertexOutput;
+    let x = (1.0 - f32(vertex_index)) * 3.0;
+    let y = f32(vertex_index & 1u) * 3.0 - 1.0;
+    out.clip_position = vec4<f32>(x, y, 0.0, 1.0);
+    out.tex_coords = vec2<f32>((x + 1.0) * 160.0, (y + 1.0) * 120.0);
+    return out;
+}
+
+@group(0) @binding(0) var framebuffer_texture: texture_2d<u32>;
+@group(0) @binding(1) var palette_texture: texture_1d<f32>;
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    let texel = vec2<i32>(floor(in.tex_coords));
+    let index = textureLoad(framebuffer_texture, texel, 0).r;
+    return textureLoad(palette_texture, i32(index), 0);
+}

uw8-window/src/gpu/square.rs

@@ -0,0 +1,157 @@
+use wgpu::util::DeviceExt;
+use winit::dpi::PhysicalSize;
+
+use super::Filter;
+
+pub struct SquareFilter {
+    uniform_buffer: wgpu::Buffer,
+    bind_group: wgpu::BindGroup,
+    pipeline: wgpu::RenderPipeline,
+}
+
+impl SquareFilter {
+    pub fn new(
+        device: &wgpu::Device,
+        screen: &wgpu::TextureView,
+        resolution: PhysicalSize<u32>,
+        surface_format: wgpu::TextureFormat,
+    ) -> SquareFilter {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(resolution),
+        };
+
+        let uniform_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: None,
+            contents: bytemuck::cast_slice(&[uniforms]),
+            usage: wgpu::BufferUsages::UNIFORM | wgpu::BufferUsages::COPY_DST,
+        });
+
+        let bind_group_layout = device.create_bind_group_layout(&wgpu::BindGroupLayoutDescriptor {
+            entries: &[
+                wgpu::BindGroupLayoutEntry {
+                    binding: 0,
+                    visibility: wgpu::ShaderStages::FRAGMENT,
+                    ty: wgpu::BindingType::Texture {
+                        multisampled: false,
+                        view_dimension: wgpu::TextureViewDimension::D2,
+                        sample_type: wgpu::TextureSampleType::Float { filterable: true },
+                    },
+                    count: None,
+                },
+                wgpu::BindGroupLayoutEntry {
+                    binding: 1,
+                    visibility: wgpu::ShaderStages::FRAGMENT,
+                    ty: wgpu::BindingType::Sampler(wgpu::SamplerBindingType::Filtering),
+                    count: None,
+                },
+                wgpu::BindGroupLayoutEntry {
+                    binding: 2,
+                    visibility: wgpu::ShaderStages::VERTEX_FRAGMENT,
+                    ty: wgpu::BindingType::Buffer {
+                        ty: wgpu::BufferBindingType::Uniform,
+                        has_dynamic_offset: false,
+                        min_binding_size: None,
+                    },
+                    count: None,
+                },
+            ],
+            label: None,
+        });
+
+        let sampler = device.create_sampler(&wgpu::SamplerDescriptor {
+            mag_filter: wgpu::FilterMode::Linear,
+            ..Default::default()
+        });
+
+        let bind_group = device.create_bind_group(&wgpu::BindGroupDescriptor {
+            layout: &bind_group_layout,
+            entries: &[
+                wgpu::BindGroupEntry {
+                    binding: 0,
+                    resource: wgpu::BindingResource::TextureView(&screen),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 1,
+                    resource: wgpu::BindingResource::Sampler(&sampler),
+                },
+                wgpu::BindGroupEntry {
+                    binding: 2,
+                    resource: uniform_buffer.as_entire_binding(),
+                },
+            ],
+            label: None,
+        });
+
+        let shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: None,
+            source: wgpu::ShaderSource::Wgsl(include_str!("square.wgsl").into()),
+        });
+
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: None,
+                bind_group_layouts: &[&bind_group_layout],
+                push_constant_ranges: &[],
+            });
+
+        let render_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: None,
+            layout: Some(&render_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &shader,
+                entry_point: "vs_main",
+                buffers: &[],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &shader,
+                entry_point: "fs_main",
+                targets: &[Some(wgpu::ColorTargetState {
+                    format: surface_format,
+                    blend: None,
+                    write_mask: wgpu::ColorWrites::ALL,
+                })],
+            }),
+            primitive: Default::default(),
+            depth_stencil: None,
+            multisample: Default::default(),
+            multiview: None,
+        });
+
+        SquareFilter {
+            uniform_buffer,
+            bind_group,
+            pipeline: render_pipeline,
+        }
+    }
+}
+
+impl Filter for SquareFilter {
+    fn resize(&mut self, queue: &wgpu::Queue, new_size: PhysicalSize<u32>) {
+        let uniforms = Uniforms {
+            texture_scale: texture_scale_from_resolution(new_size),
+        };
+        queue.write_buffer(&self.uniform_buffer, 0, bytemuck::cast_slice(&[uniforms]));
+    }
+
+    fn render<'a>(&'a self, render_pass: &mut wgpu::RenderPass<'a>) {
+        render_pass.set_pipeline(&self.pipeline);
+        render_pass.set_bind_group(0, &self.bind_group, &[]);
+        render_pass.draw(0..6, 0..1);
+    }
+}
+
+fn texture_scale_from_resolution(res: PhysicalSize<u32>) -> [f32; 4] {
+    let scale = ((res.width as f32) / 160.0).min((res.height as f32) / 120.0);
+    [
+        scale / res.width as f32,
+        scale / res.height as f32,
+        2.0 / scale,
+        0.0,
+    ]
+}
+
+#[repr(C)]
+#[derive(Debug, Copy, Clone, bytemuck::Pod, bytemuck::Zeroable)]
+struct Uniforms {
+    texture_scale: [f32; 4],
+}

uw8-window/src/gpu/square.wgsl

@@ -0,0 +1,44 @@
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) tex_coords: vec2<f32>,
+}
+
+struct Uniforms {
+    texture_scale: vec4<f32>,
+}
+
+@group(0) @binding(2) var<uniform> uniforms: Uniforms;
+
+@vertex
+fn vs_main(
+    @builtin(vertex_index) in_vertex_index: u32,
+) -> VertexOutput {
+    var out: VertexOutput;
+    let i = in_vertex_index / 3u + in_vertex_index % 3u;
+    let x = 0.0 + f32(i % 2u) * 320.0;
+    let y = 0.0 + f32(i / 2u) * 240.0;
+    out.clip_position = vec4<f32>((vec2<f32>(x, y) - vec2<f32>(160.0, 120.0)) * uniforms.texture_scale.xy, 0.0, 1.0);
+    out.tex_coords = vec2<f32>(x, y);
+    return out;
+}
+
+@group(0) @binding(0) var screen_texture: texture_2d<f32>;
+@group(0) @binding(1) var linear_sampler: sampler;
+
+fn aa_tex_coord(c: f32) -> f32 {
+    let low = c - uniforms.texture_scale.z * 0.5;
+    let high = c + uniforms.texture_scale.z * 0.5;
+    let base = floor(low);
+    let center = base + 0.5;
+    let next = base + 1.0;
+    if high > next {
+        return center + (high - next) / (high - low);
+    } else {
+        return center;
+    }
+}
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    return textureSample(screen_texture, linear_sampler, vec2<f32>(aa_tex_coord(in.tex_coords.x), aa_tex_coord(in.tex_coords.y)) / vec2<f32>(320.0, 240.0));
+}

uw8-window/src/lib.rs

@@ -0,0 +1,124 @@
+use anyhow::Result;
+use std::time::Instant;
+
+mod cpu;
+mod gpu;
+
+pub struct Window {
+    inner: Box<dyn WindowImpl>,
+    fps_counter: Option<FpsCounter>,
+}
+
+struct FpsCounter {
+    start: Instant,
+    num_frames: u32,
+}
+
+impl Window {
+    pub fn new(mut config: WindowConfig) -> Result<Window> {
+        let fps_counter = if config.fps_counter {
+            Some(FpsCounter {
+                start: Instant::now(),
+                num_frames: 0,
+            })
+        } else {
+            None
+        };
+        config.scale = config.scale.max(1.).min(20.);
+        if config.enable_gpu {
+            match gpu::Window::new(config) {
+                Ok(window) => {
+                    return Ok(Window {
+                        inner: Box::new(window),
+                        fps_counter,
+                    })
+                }
+                Err(err) => eprintln!(
+                    "Failed to create gpu window: {}\nFalling back tp cpu window",
+                    err
+                ),
+            }
+        }
+        cpu::Window::new().map(|window| Window {
+            inner: Box::new(window),
+            fps_counter,
+        })
+    }
+
+    pub fn begin_frame(&mut self) -> Input {
+        self.inner.begin_frame()
+    }
+    pub fn end_frame(&mut self, framebuffer: &[u8], palette: &[u8], next_frame: Instant) {
+        self.inner.end_frame(framebuffer, palette, next_frame);
+        if let Some(ref mut fps_counter) = self.fps_counter {
+            fps_counter.num_frames += 1;
+            let elapsed = fps_counter.start.elapsed().as_secs_f32();
+            if elapsed >= 1.0 {
+                println!("fps: {:.1}", fps_counter.num_frames as f32 / elapsed);
+                fps_counter.num_frames = 0;
+                fps_counter.start = Instant::now();
+            }
+        }
+    }
+
+    pub fn is_open(&self) -> bool {
+        self.inner.is_open()
+    }
+}
+
+#[derive(Debug)]
+pub struct WindowConfig {
+    enable_gpu: bool,
+    filter: u32,
+    fullscreen: bool,
+    fps_counter: bool,
+    scale: f32,
+}
+
+impl Default for WindowConfig {
+    fn default() -> WindowConfig {
+        WindowConfig {
+            enable_gpu: true,
+            filter: 5,
+            fullscreen: false,
+            fps_counter: false,
+            scale: 2.,
+        }
+    }
+}
+
+impl WindowConfig {
+    pub fn parse_arguments(&mut self, args: &mut pico_args::Arguments) {
+        self.enable_gpu = !args.contains("--no-gpu");
+        if let Some(filter) = args.opt_value_from_str::<_, String>("--filter").unwrap() {
+            self.filter = match filter.as_str() {
+                "1" | "nearest" => 1,
+                "2" | "fast_crt" => 2,
+                "3" | "ss_crt" => 3,
+                "4" | "chromatic" => 4,
+                "5" | "auto_crt" => 5,
+                o => {
+                    println!("Unknown --filter '{}'", o);
+                    std::process::exit(1);
+                }
+            }
+        }
+        self.fullscreen = args.contains("--fullscreen");
+        self.fps_counter = args.contains("--fps");
+        self.scale = args
+            .opt_value_from_str("--scale")
+            .unwrap()
+            .unwrap_or(self.scale);
+    }
+}
+
+pub struct Input {
+    pub gamepads: [u8; 4],
+    pub reset: bool,
+}
+
+trait WindowImpl {
+    fn begin_frame(&mut self) -> Input;
+    fn end_frame(&mut self, framebuffer: &[u8], palette: &[u8], next_frame: Instant);
+    fn is_open(&self) -> bool;
+}

uw8-window/src/main.rs

@@ -0,0 +1,59 @@
+use std::time::Instant;
+use uw8_window::WindowConfig;
+
+fn main() {
+    env_logger::Builder::from_env(env_logger::Env::default().default_filter_or("info")).init();
+
+    let mut args = pico_args::Arguments::from_env();
+
+    let mut framebuffer = vec![0u8; 320 * 240];
+    let mut start_time = Instant::now();
+
+    let mut palette = vec![0u32; 256];
+    for i in 0..256 {
+        let v = i & 15;
+        let r = ((i >> 2) & 12) * v;
+        let g = ((i >> 3) & 12) * v;
+        let b = ((i >> 4) & 12) * v;
+        palette[i as usize] = r + (g << 8) + (b << 16);
+    }
+
+    let mut fps_start = Instant::now();
+    let mut fps_counter = 0;
+
+    let mut window_config = WindowConfig::default();
+    window_config.parse_arguments(&mut args);
+
+    let mut window = uw8_window::Window::new(window_config).unwrap();
+
+    while window.is_open() {
+        let input = window.begin_frame();
+        if input.reset {
+            start_time = Instant::now();
+        }
+        draw_frame(&mut framebuffer, start_time.elapsed().as_secs_f32());
+        window.end_frame(&framebuffer, bytemuck::cast_slice(&palette), Instant::now());
+
+        fps_counter += 1;
+        let elapsed = fps_start.elapsed().as_secs_f32();
+        if elapsed >= 1.0 {
+            println!("{:.1} fps", fps_counter as f32 / elapsed);
+            fps_start = Instant::now();
+            fps_counter = 0;
+        }
+    }
+}
+
+fn draw_frame(framebuffer: &mut [u8], time: f32) {
+    for x in 0..320 {
+        let xr = x as f32 - 160.0;
+        for y in 0..240 {
+            let yr = y as f32 - 120.0;
+            let f = 8192.0 / (xr * xr + yr * yr);
+            let u = xr * f + 512.0 + time * 32.0;
+            let v = yr * f + time * 29.0;
+            let c = (u.floor() as i32 ^ v.floor() as i32) as u32;
+            framebuffer[x + y * 320] = c as u8;
+        }
+    }
+}

web/opus-repro.html

@@ -0,0 +1,57 @@
+<html>
+    <button onclick="go()">Go!</button>
+    <canvas id="screen" width="320" height="240"></canvas>
+    <video id="video"></video>
+    <script>
+        function go() {
+            let audioContext = new AudioContext({sampleRate: 44100});
+
+            let oscillator = new OscillatorNode(audioContext);
+            let gain = new GainNode(audioContext, {gain: 1});
+            oscillator.connect(gain);
+            gain.connect(audioContext.destination);
+            for(let i = 0; i < 8; ++i ) {
+                gain.gain.setValueAtTime(1, i / 2);
+                gain.gain.setValueAtTime(0, i / 2 + 0.3);
+            }
+            oscillator.start();
+            oscillator.stop(4);
+
+            let screen = document.getElementById('screen');
+            let context = screen.getContext('2d');
+            let startTime = Date.now();
+            let drawFrame = () => {
+                let time = Date.now() - startTime;
+                context.fillStyle = 'white';
+                context.fillRect(0, 0, 320, 240);
+                if(time < 4000) {
+                    if(time % 500 < 300) {
+                        context.fillStyle = 'black';
+                        context.fillRect(time / 15, 50, 50, 50);
+                    }
+                    window.requestAnimationFrame(drawFrame);
+                }
+            };
+            drawFrame();
+
+            let stream = screen.captureStream();
+            let audioStreamNode = audioContext.createMediaStreamDestination();
+            gain.connect(audioStreamNode);
+            stream.addTrack(audioStreamNode.stream.getAudioTracks()[0]);
+            let recorder = new MediaRecorder(stream, { mimeType: 'video/webm' });
+
+            let chunks = [];
+            recorder.ondataavailable = e => chunks.push(e.data);
+            recorder.onstop = () => {
+                let blob = new Blob(chunks, {type: 'video/webm'});
+                let url = URL.createObjectURL(blob);
+                let video = document.getElementById('video');
+                video.src = url;
+                video.play();
+            };
+
+            recorder.start();
+            setTimeout(() => recorder.stop(), 4000);
+        }
+    </script>
+</html>

web/package.json

@@ -4,5 +4,6 @@
     "@parcel/optimizer-data-url": "^2.0.0",
     "@parcel/transformer-inline-string": "^2.0.0",
     "parcel": "^2.0.0"
-  }
+  },
+  "dependencies": {}
 }

web/run

@@ -0,0 +1,2 @@
+#!/bin/bash
+rm -rf .parcel-cache && yarn parcel src/index.html

web/src/audiolet.js

@@ -0,0 +1,94 @@
+let U8 = (...a) => new Uint8Array(...a);
+class APU extends AudioWorkletProcessor {
+    constructor() {
+        super();
+        this.sampleIndex = 0;
+        this.currentTime = 0;
+        this.isFirstMessage = true;
+        this.pendingUpdates = [];
+        this.port.onmessage = (ev) => {
+            if(this.memory) {
+                if(this.isFirstMessage)
+                {
+                    this.currentTime += (ev.data.t - this.currentTime) / 8;
+                    this.isFirstMessage = false;
+                }
+                this.pendingUpdates.push(ev.data);
+            } else {
+                this.load(ev.data[0], ev.data[1]);
+            }
+        };
+    }
+
+    async load(platform_data, data) {
+        let memory = new WebAssembly.Memory({ initial: 4, maximum: 4 });
+
+        let importObject = {
+            env: {
+                memory
+            },
+        };
+
+        for (let n of ['acos', 'asin', 'atan', 'atan2', 'cos', 'exp', 'log', 'sin', 'tan', 'pow']) {
+            importObject.env[n] = Math[n];
+        }
+
+        for (let i = 9; i < 64; ++i) {
+            importObject.env['reserved' + i] = () => { };
+        }
+
+        let logLine = '';
+        importObject.env['logChar'] = (c) => {
+            if(c == 10) {
+                console.log(logLine);
+                logLine = '';
+            } else {
+                logLine += String.fromCharCode(c);
+            }
+        };
+
+        for (let i = 0; i < 16; ++i) {
+            importObject.env['g_reserved' + i] = 0;
+        }
+
+        let instantiate = async (data) => (await WebAssembly.instantiate(data, importObject)).instance;
+
+        let platform_instance = await instantiate(platform_data);
+
+        for (let name in platform_instance.exports) {
+            importObject.env[name] = platform_instance.exports[name]
+        }
+
+        let instance = await instantiate(data);
+
+        this.memory = memory;
+
+        this.snd = instance.exports.snd || platform_instance.exports.sndGes;
+
+        this.port.postMessage(2);
+    }
+
+    process(inputs, outputs, parameters) {
+        this.isFirstMessage = true;
+        if(this.snd) {
+            while(this.pendingUpdates.length > 0 && this.pendingUpdates[0].t <= this.currentTime) {
+                U8(this.memory.buffer, 80, 32).set(U8(this.pendingUpdates.shift().r));
+            }
+            let u32Mem = new Uint32Array(this.memory.buffer);
+            u32Mem[16] = this.currentTime;
+            let channels = outputs[0];
+            let index = this.sampleIndex;
+            let numSamples = channels[0].length;
+            for(let i = 0; i < numSamples; ++i) {
+                channels[0][i] = this.snd(index++);
+                channels[1][i] = this.snd(index++);
+            }
+            this.sampleIndex = index & 0xffffffff;
+            this.currentTime += numSamples / 44.1;
+        }
+
+        return true;
+    }
+}
+
+registerProcessor('apu', APU);

web/src/index.html

@@ -10,13 +10,14 @@
     </head>
     <body>
         <div id="uw8">
-            <a href="https://exoticorn.github.io/microw8">MicroW8</a> 0.1.2
+            <a href="https://exoticorn.github.io/microw8">MicroW8</a> 0.3.0
         </div>
         <div id="centered">
-            <canvas id="screen" width="320" height="240">
+            <canvas class="screen" id="screen" width="320" height="240">
             </canvas>
-        <div id="timer" hidden="true"></div>        
-        <div id="message"></div>
+            <button class="screen" id="start" style="display:none">Click to start</button>
+            <div id="timer" hidden="true"></div>        
+            <div id="message"></div>
             <button id="cartButton" style="visibility:hidden">Load cart...</button>
         </div>
         <div id="footer">

web/src/main.js

@@ -12,6 +12,7 @@ let uw8 = MicroW8(document.getElementById('screen'), {
     setMessage,
     keyboardElement: window,
     timerElement: document.getElementById("timer"),
+    startButton: document.getElementById("start")
 });
 
 function runModuleFromHash() {
@@ -79,7 +80,9 @@ if(location.hash.length != 0) {
             url += 'cart.uw8';
         }
         try {
-            await uw8.runModuleFromURL(url, true);
+            if(!await uw8.runModuleFromURL(url, true)) {
+                setupLoad();
+            }
         } catch(e) {
             setupLoad();
         }

web/src/microw8.js

@@ -1,5 +1,15 @@
 import loaderUrl from "data-url:../../platform/bin/loader.wasm";
 import platformUrl from "data-url:../../platform/bin/platform.uw8";
+import audioWorkletUrl from "data-url:./audiolet.js";
+
+class AudioNode extends AudioWorkletNode {
+    constructor(context) {
+        super(context, 'apu', {outputChannelCount: [2]});
+    }
+}
+
+let U8 = (...a) => new Uint8Array(...a);
+let U32 = (...a) => new Uint32Array(...a);
 
 export default function MicroW8(screen, config = {}) {
     if(!config.setMessage) {
@@ -18,9 +28,6 @@ export default function MicroW8(screen, config = {}) {
     
     let currentData;
     
-    let U8 = (d) => new Uint8Array(d);
-    let U32 = (d) => new Uint32Array(d);
-    
     let pad = 0;
     let keyboardElement = config.keyboardElement == undefined ? screen : config.keyboardElement;
     if(keyboardElement) {
@@ -84,12 +91,24 @@ export default function MicroW8(screen, config = {}) {
         keyboardElement.onkeyup = keyHandler;
     }
 
+    let audioContext;
+    let audioNode;
+    
     async function runModule(data, keepUrl) {
         if (cancelFunction) {
             cancelFunction();
             cancelFunction = null;
         }
     
+        audioContext = new AudioContext({sampleRate: 44100});
+        let keepRunning = true;
+        let abortController = new AbortController();
+        cancelFunction = () => {
+            audioContext.close();
+            keepRunning = false;
+            abortController.abort();
+        };
+
         let cartridgeSize = data.byteLength;
     
         config.setMessage(cartridgeSize);
@@ -97,6 +116,39 @@ export default function MicroW8(screen, config = {}) {
             return;
         }
     
+        await audioContext.audioWorklet.addModule(audioWorkletUrl);
+        audioNode = new AudioNode(audioContext);
+
+        let audioReadyFlags = 0;
+        let audioReadyResolve;
+        let audioReadyPromise = new Promise(resolve => audioReadyResolve = resolve);
+        let updateAudioReady = (f) => {
+            audioReadyFlags |= f;
+            if(audioReadyFlags == 3 && audioReadyResolve) {
+                audioReadyResolve(true);
+                audioReadyResolve = null;
+            }
+        };
+        let audioStateChange = () => {
+            if(audioContext.state == 'suspended') {
+                if(config.startButton) {
+                    config.startButton.style = '';
+                    screen.style = 'display:none';
+                }
+                (config.startButton || screen).onclick = () => {
+                    audioContext.resume();
+                };
+            } else {
+                if(config.startButton) {
+                    config.startButton.style = 'display:none';
+                    screen.style = '';
+                }
+                updateAudioReady(1);
+            }
+        };
+        audioContext.onstatechange = audioStateChange;
+        audioStateChange();
+
         currentData = data;
     
         let newURL = window.location.pathname;
@@ -119,7 +171,7 @@ export default function MicroW8(screen, config = {}) {
             if(!devkitMode) {
                 memSize.maximum = 4;
             }
-            let memory = new WebAssembly.Memory({ initial: 4, maximum: devkitMode ? 16 : 4 });
+            let memory = new WebAssembly.Memory(memSize);
             let memU8 = U8(memory.buffer);
     
             let importObject = {
@@ -142,9 +194,9 @@ export default function MicroW8(screen, config = {}) {
     
             let instantiate = async (data) => (await WebAssembly.instantiate(data, importObject)).instance;
     
-            let loadModuleURL = async (url) => instantiate(loadModuleData(await (await fetch(url)).arrayBuffer()));
+            let loadModuleURL = async (url) => loadModuleData(await (await fetch(url)).arrayBuffer());
     
-            loader = await loadModuleURL(loaderUrl);
+            loader = await instantiate(await loadModuleURL(loaderUrl));
     
             for (let n of ['acos', 'asin', 'atan', 'atan2', 'cos', 'exp', 'log', 'sin', 'tan', 'pow']) {
                 importObject.env[n] = Math[n];
@@ -154,13 +206,28 @@ export default function MicroW8(screen, config = {}) {
                 importObject.env['reserved' + i] = () => { };
             }
     
+            let logLine = '';
+            importObject.env['logChar'] = (c) => {
+                if(c == 10) {
+                    console.log(logLine);
+                    logLine = '';
+                } else {
+                    logLine += String.fromCharCode(c);
+                }
+            };
+    
             for (let i = 0; i < 16; ++i) {
                 importObject.env['g_reserved' + i] = 0;
             }
     
             data = loadModuleData(data);
     
-            let platform_instance = await loadModuleURL(platformUrl);
+            let platform_data = await loadModuleURL(platformUrl);
+
+            audioNode.port.onmessage = (e) => updateAudioReady(e.data);
+            audioNode.port.postMessage([platform_data, data]);
+
+            let platform_instance = await instantiate(platform_data);
     
             for (let name in platform_instance.exports) {
                 importObject.env[name] = platform_instance.exports[name]
@@ -170,13 +237,35 @@ export default function MicroW8(screen, config = {}) {
     
             let buffer = U32(imageData.data.buffer);
 
+            await audioReadyPromise;
+    
             let startTime = Date.now();
     
-            let keepRunning = true;
-            cancelFunction = () => keepRunning = false;
-    
             const timePerFrame = 1000 / 60;
-            let nextFrame = startTime;
+
+            audioNode.connect(audioContext.destination);
+
+            let isPaused = false;
+            let pauseTime = startTime;
+            let updateVisibility = isVisible => {
+                let now = Date.now();
+                if(isVisible) {
+                    isPaused = false;
+                    audioContext.resume();
+                    startTime += now - pauseTime;
+                } else {
+                    isPaused = true;
+                    audioContext.suspend();
+                    pauseTime = now;
+                }
+            };
+            window.addEventListener('focus', () => updateVisibility(true), { signal: abortController.signal });
+            window.addEventListener('blur', () => updateVisibility(false), { signal: abortController.signal });
+            updateVisibility(document.hasFocus());
+
+            if (instance.exports.start) {
+                instance.exports.start();
+            }
 
             function mainloop() {
                 if (!keepRunning) {
@@ -184,9 +273,9 @@ export default function MicroW8(screen, config = {}) {
                 }
     
                 try {
-                    let now = Date.now();
                     let restart = false;
-                    if (now >= nextFrame) {
+                    let thisFrame;
+                    if (!isPaused) {
                         let gamepads = navigator.getGamepads();
                         let gamepad = 0;
                         for (let i = 0; i < 4; ++i) {
@@ -214,23 +303,36 @@ export default function MicroW8(screen, config = {}) {
                         }
     
                         let u32Mem = U32(memory.buffer);
-                        u32Mem[16] = now - startTime;
+                        let time = Date.now() - startTime;
+                        u32Mem[16] = time;
                         u32Mem[17] = pad | gamepad;
-                        instance.exports.upd();
+                        if(instance.exports.upd) {
+                            instance.exports.upd();
+                        }
                         platform_instance.exports.endFrame();
 
-                        let palette = U32(memory.buffer.slice(0x13000, 0x13000 + 1024));
+                        let soundRegisters = new ArrayBuffer(32);
+                        U8(soundRegisters).set(U8(memory.buffer, 80, 32));
+                        audioNode.port.postMessage({t: time, r: soundRegisters}, [soundRegisters]);
+    
+                        let palette = U32(memory.buffer, 0x13000, 1024);
                         for (let i = 0; i < 320 * 240; ++i) {
                             buffer[i] = palette[memU8[i + 120]] | 0xff000000;
                         }
                         canvasCtx.putImageData(imageData, 0, 0);
-                        nextFrame = Math.max(nextFrame + timePerFrame, now);
+
+                        let timeOffset = ((time * 6) % 100 - 50) / 6;
+                        thisFrame = startTime + time - timeOffset / 8;
+                    } else {
+                        thisFrame = Date.now();
                     }
+                    let now = Date.now();
+                    let nextFrame = Math.max(thisFrame + timePerFrame, now);
     
                     if (restart) {
                         runModule(currentData);
                     } else {
-                        window.requestAnimationFrame(mainloop);
+                        window.setTimeout(mainloop, nextFrame - now)
                     }
                 } catch (err) {
                     config.setMessage(cartridgeSize, err.toString());
@@ -253,14 +355,25 @@ export default function MicroW8(screen, config = {}) {
     
     let videoRecorder;
     let videoStartTime;
+    let videoAudioSourceNode;
+    let videoAudioStreamNode;
     function recordVideo() {
         if(videoRecorder) {
             videoRecorder.stop();
             videoRecorder = null;
+            videoAudioSourceNode.disconnect(videoAudioStreamNode);
+            videoAudioSourceNode = null;
+            videoAudioStreamNode = null;
             return;
         }
 
-        videoRecorder = new MediaRecorder(screen.captureStream(), {
+        let stream = screen.captureStream();
+        videoAudioStreamNode = audioContext.createMediaStreamDestination();
+        videoAudioSourceNode = audioNode;
+        audioNode.connect(videoAudioStreamNode);
+        stream.addTrack(videoAudioStreamNode.stream.getAudioTracks()[0]);
+    
+        videoRecorder = new MediaRecorder(stream, {
             mimeType: 'video/webm',
             videoBitsPerSecond: 25000000
         });
@@ -305,10 +418,11 @@ export default function MicroW8(screen, config = {}) {
     async function runModuleFromURL(url, keepUrl) {
         let response = await fetch(url);
         let type = response.headers.get('Content-Type');
-        if(type && type.includes('html')) {
-            throw false;
+        if((type && type.includes('html')) || response.status != 200) {
+            return false;
         }
         runModule(await response.arrayBuffer(), keepUrl || devkitMode);
+        return true;
     }
 
     return {

web/src/style.css

@@ -37,7 +37,7 @@ a:hover {
     color: #405040;
 }
 
-#screen {
+.screen {
     width: 320px;
     height: 240px;
     image-rendering: pixelated;
@@ -45,9 +45,16 @@ a:hover {
     margin-bottom: 8px;
     border: 4px solid #303040;
     box-shadow: 5px 5px 20px black;
+}
+
+#screen {
     cursor: none;
 }
 
+#start {
+    font-size: 150%;
+}
+
 #timer::before {
     content: '';
     display: inline-block;
@@ -84,21 +91,21 @@ button:active {
 }
 
 @media (min-width: 680px) and (min-height: 620px) {
-    #screen {
+    .screen {
         width: 640px;
         height: 480px;
     }
 }
 
 @media (min-width: 1000px) and (min-height: 800px) {
-    #screen {
+    .screen {
         width: 960px;
         height: 720px;
     }
 }
 
 @media (width:640px) and (height:480px) {
-    #screen {
+    .screen {
         width: 640px;
         height: 480px;
         border: 0;