Merge 1d89ef8613 into a651107104

.github/workflows/ci.yml

@@ -30,9 +30,9 @@ jobs:
       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

@@ -1,29 +1,31 @@
 [package]
 name = "uw8"
-version = "0.2.0-rc3"
+version = "0.2.2"
 edition = "2021"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [features]
 default = ["native", "browser"]
-native = ["wasmtime", "minifb", "cpal", "rubato"]
+native = ["wasmtime", "uw8-window", "cpal", "rubato" ]
 browser = ["warp", "tokio", "tokio-stream", "webbrowser"]
 
 [dependencies]
-wasmtime = { version = "0.35.3", optional = true }
+wasmtime = { version = "12.0.0", optional = true }
 anyhow = "1"
-minifb = { version = "0.22", default-features = false, features = ["x11"], optional = true }
+env_logger = "0.10"
-notify = "4"
+log = "0.4"
-pico-args = "0.4"
+uw8-window = { path = "uw8-window", optional = true }
-curlywas = { git = "https://github.com/exoticorn/curlywas.git", rev = "c22297e" }
+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 }
+warp = { version = "0.3.5", optional = true }
-tokio = { version = "1.17.0", features = ["sync", "rt"], optional = true }
+tokio = { version = "1.32.0", features = ["sync", "rt"], optional = true }
-tokio-stream = { version = "0.1.8", features = ["sync"], optional = true }
+tokio-stream = { version = "0.1.14", features = ["sync"], optional = true }
-webbrowser = { version = "0.6.0", optional = true }
+webbrowser = { version = "0.8.11", optional = true }
 ansi_term = "0.12.1"
-cpal = { version = "0.13.5", optional = true }
+cpal = { version = "0.15.2", optional = true }
-rubato = { version = "0.11.0", optional = true }
+rubato = { version = "0.14.0", 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)
+* [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/v0.1.2/microw8-0.1.2-macos.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/v0.1.2/microw8-0.1.2-windows.zip)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.2/microw8-0.2.2-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/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/include/microw8-api.cwa

@@ -34,6 +34,7 @@ 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.exp" fn exp(f32) -> f32;
 import "env.playNote" fn playNote(i32, i32);
+import "env.sndGes" fn sndGes(i32) -> f32;
 
 const TIME_MS = 0x40;
 const GAMEPAD = 0x44;

examples/include/microw8-api.wat

@@ -34,6 +34,7 @@
 (import "env" "circle_outline" (func $circle_outline (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)))
 
 ;; 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

@@ -110,9 +110,9 @@ dependencies = [
 
 [[package]]
 name = "crc32fast"
-version = "1.3.0"
+version = "1.3.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "738c290dfaea84fc1ca15ad9c168d083b05a714e1efddd8edaab678dc28d2836"
+checksum = "b540bd8bc810d3885c6ea91e2018302f68baba2129ab3e88f32389ee9370880d"
 dependencies = [
  "cfg-if",
 ]
@@ -146,37 +146,44 @@ checksum = "7a81dae078cea95a014a339291cec439d2f232ebe854a9d672b796c6afafa9b7"
 [[package]]
 name = "curlywas"
 version = "0.1.0"
-source = "git+https://github.com/exoticorn/curlywas.git?rev=c22297e#c22297ea82977ac06373d629fb273a795322d788"
+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"
+version = "0.4.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "52db5973b6a19247baf19b30f41c23a1bfffc2e9ce0a5db2f60e3cd5dc8895f7"
+checksum = "3f57ccc32870366ae684be48b32a1a2e196f98a42a9b4361fe77e13fd4a34755"
 dependencies = [
  "hashbrown",
 ]
 
 [[package]]
 name = "flate2"
-version = "1.0.22"
+version = "1.0.25"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1e6988e897c1c9c485f43b47a529cef42fde0547f9d8d41a7062518f1d8fc53f"
+checksum = "a8a2db397cb1c8772f31494cb8917e48cd1e64f0fa7efac59fbd741a0a8ce841"
 dependencies = [
- "cfg-if",
  "crc32fast",
- "libc",
  "miniz_oxide",
 ]
 
+[[package]]
+name = "form_urlencoded"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a9c384f161156f5260c24a097c56119f9be8c798586aecc13afbcbe7b7e26bf8"
+dependencies = [
+ "percent-encoding",
+]
+
 [[package]]
 name = "getrandom"
 version = "0.2.3"
@@ -190,9 +197,9 @@ dependencies = [
 
 [[package]]
 name = "hashbrown"
-version = "0.11.2"
+version = "0.12.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ab5ef0d4909ef3724cc8cce6ccc8572c5c817592e9285f5464f8e86f8bd3726e"
+checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
 dependencies = [
  "ahash 0.7.6",
 ]
@@ -212,6 +219,26 @@ version = "2.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "25a2bc672d1148e28034f176e01fffebb08b35768468cc954630da77a1449005"
 
+[[package]]
+name = "idna"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e14ddfc70884202db2244c223200c204c2bda1bc6e0998d11b5e024d657209e6"
+dependencies = [
+ "unicode-bidi",
+ "unicode-normalization",
+]
+
+[[package]]
+name = "indexmap"
+version = "1.9.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1885e79c1fc4b10f0e172c475f458b7f7b93061064d98c3293e98c5ba0c8b399"
+dependencies = [
+ "autocfg",
+ "hashbrown",
+]
+
 [[package]]
 name = "lazy_static"
 version = "1.4.0"
@@ -225,17 +252,24 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
 
 [[package]]
-name = "libc"
+name = "lexopt"
-version = "0.2.112"
+version = "0.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "1b03d17f364a3a042d5e5d46b053bbbf82c92c9430c592dd4c064dc6ee997125"
+checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
+
+[[package]]
+name = "libc"
+version = "0.2.139"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "201de327520df007757c1f0adce6e827fe8562fbc28bfd9c15571c66ca1f5f79"
 
 [[package]]
 name = "lodepng"
-version = "3.4.7"
+version = "3.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "24844d5c0b922ddd52fb5bf0964a4c7f8e799a946ec01bb463771eb04fc1a323"
+checksum = "f0ad39f75bbaa4b10bb6f2316543632a8046a5bcf9c785488d79720b21f044f8"
 dependencies = [
+ "crc32fast",
  "fallible_collections",
  "flate2",
  "libc",
@@ -253,12 +287,11 @@ dependencies = [
 
 [[package]]
 name = "miniz_oxide"
-version = "0.4.4"
+version = "0.6.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "a92518e98c078586bc6c934028adcca4c92a53d6a958196de835170a01d84e4b"
+checksum = "b275950c28b37e794e8c55d88aeb5e139d0ce23fdbbeda68f8d7174abdf9e8fa"
 dependencies = [
  "adler",
- "autocfg",
 ]
 
 [[package]]
@@ -288,12 +321,24 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "percent-encoding"
+version = "2.2.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "478c572c3d73181ff3c2539045f6eb99e5491218eae919370993b890cdbdd98e"
+
 [[package]]
 name = "pico-args"
 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"
@@ -330,9 +375,9 @@ dependencies = [
 
 [[package]]
 name = "rgb"
-version = "0.8.31"
+version = "0.8.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9a374af9a0e5fdcdd98c1c7b64f05004f9ea2555b6c75f211daa81268a3c50f1"
+checksum = "3603b7d71ca82644f79b5a06d1220e9a58ede60bd32255f698cb1af8838b8db3"
 dependencies = [
  "bytemuck",
 ]
@@ -357,6 +402,26 @@ dependencies = [
  "unicode-xid",
 ]
 
+[[package]]
+name = "thiserror"
+version = "1.0.38"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6a9cd18aa97d5c45c6603caea1da6628790b37f7a34b6ca89522331c5180fed0"
+dependencies = [
+ "thiserror-impl",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.38"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fb327af4685e4d03fa8cbcf1716380da910eeb2bb8be417e7f9fd3fb164f36f"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "time"
 version = "0.1.43"
@@ -376,6 +441,36 @@ dependencies = [
  "crunchy",
 ]
 
+[[package]]
+name = "tinyvec"
+version = "1.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "87cc5ceb3875bb20c2890005a4e226a4651264a5c75edb2421b52861a0a0cb50"
+dependencies = [
+ "tinyvec_macros",
+]
+
+[[package]]
+name = "tinyvec_macros"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cda74da7e1a664f795bb1f8a87ec406fb89a02522cf6e50620d016add6dbbf5c"
+
+[[package]]
+name = "unicode-bidi"
+version = "0.3.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d54675592c1dbefd78cbd98db9bacd89886e1ca50692a0692baefffdeb92dd58"
+
+[[package]]
+name = "unicode-normalization"
+version = "0.1.22"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c5713f0fc4b5db668a2ac63cdb7bb4469d8c9fed047b1d0292cc7b0ce2ba921"
+dependencies = [
+ "tinyvec",
+]
+
 [[package]]
 name = "unicode-segmentation"
 version = "1.9.0"
@@ -390,13 +485,24 @@ checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3"
 
 [[package]]
 name = "upkr"
-version = "0.1.0"
+version = "0.2.1"
-source = "git+https://github.com/exoticorn/upkr.git?rev=2e7983fc#2e7983fc650788d98da2eecef2d16f63e849e4a0"
+source = "git+https://github.com/exoticorn/upkr.git?rev=080db40d0088bbee2bdf3c5c75288ac7853d6b7a#080db40d0088bbee2bdf3c5c75288ac7853d6b7a"
 dependencies = [
  "anyhow",
  "cdivsufsort",
- "pbr",
+ "lexopt",
- "pico-args",
+ "thiserror",
+]
+
+[[package]]
+name = "url"
+version = "2.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0d68c799ae75762b8c3fe375feb6600ef5602c883c5d21eb51c09f22b83c4643"
+dependencies = [
+ "form_urlencoded",
+ "idna",
+ "percent-encoding",
 ]
 
 [[package]]
@@ -405,11 +511,11 @@ version = "0.1.0"
 dependencies = [
  "anyhow",
  "pbr",
- "pico-args",
+ "pico-args 0.5.0",
  "upkr",
  "walrus",
- "wasm-encoder 0.8.0",
+ "wasm-encoder 0.22.0",
- "wasmparser 0.81.0",
+ "wasmparser 0.99.0",
 ]
 
 [[package]]
@@ -452,18 +558,18 @@ checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6"
 
 [[package]]
 name = "wasm-encoder"
-version = "0.8.0"
+version = "0.10.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db0c351632e46cc06a58a696a6c11e4cf90cad4b9f8f07a0b59128d616c29bb0"
+checksum = "aa9d9bf45fc46f71c407837c9b30b1e874197f2dc357588430b21e5017d290ab"
 dependencies = [
  "leb128",
 ]
 
 [[package]]
 name = "wasm-encoder"
-version = "0.10.0"
+version = "0.22.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "aa9d9bf45fc46f71c407837c9b30b1e874197f2dc357588430b21e5017d290ab"
+checksum = "ef126be0e14bdf355ac1a8b41afc89195289e5c7179f80118e3abddb472f0810"
 dependencies = [
  "leb128",
 ]
@@ -474,18 +580,22 @@ version = "0.77.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b35c86d22e720a07d954ebbed772d01180501afe7d03d464f413bb5f8914a8d6"
 
-[[package]]
-name = "wasmparser"
-version = "0.81.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98930446519f63d00a836efdc22f67766ceae8dbcc1571379f2bcabc6b2b9abc"
-
 [[package]]
 name = "wasmparser"
 version = "0.83.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "718ed7c55c2add6548cca3ddd6383d738cd73b892df400e96b9aa876f0141d7a"
 
+[[package]]
+name = "wasmparser"
+version = "0.99.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9ef3b717afc67f848f412d4f02c127dd3e35a0eecd58c684580414df4fde01d3"
+dependencies = [
+ "indexmap",
+ "url",
+]
+
 [[package]]
 name = "winapi"
 version = "0.3.9"

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="c22297e" }
+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

@@ -10,7 +10,7 @@ const GesState.Size = GesState.Filter + 8*4;
 const GesStateOffset = 32;
 const GesBufferOffset = 32 + GesState.Size;
 
-export fn gesSnd(t: i32) -> f32 {
+export fn sndGes(t: i32) -> f32 {
     let baseAddr = 0!0x12c78;
     if !(t & 127) {
         let i: i32;
@@ -62,7 +62,6 @@ export fn gesSnd(t: i32) -> f32 {
             let phase = channelState!GesChannelState.Phase;
 
             let inline pulseWidth = channelReg?1;
-            let phaseShift = (pulseWidth - 128) * 255;
             let invPhaseInc = 1 as f32 / phaseInc as f32;
 
             i = 0;
@@ -131,7 +130,7 @@ export fn gesSnd(t: i32) -> f32 {
                 let phaseInc = (freq * (65536.0 / 44100.0)) as i32;
 
                 let phase = channelState!GesChannelState.Phase;
-                if modSrc > ch {
+                if modSrc < ch {
                     phase = phase - (phaseInc << 6);
                 }
 

platform/src/platform.cwa

@@ -62,11 +62,7 @@ export fn cls(col: i32) {
     textCursorX = 0;
     textCursorY = 0;
     outputChannel = 0;
-    col = (col & 255) * 0x1010101;
+    memory.fill(120, col, 320*240);
-    loop pixels {
-        i!120 = col;
-        branch_if (i := i + 4) < 320*240: pixels;
-    }
 }
 
 export fn setPixel(x: i32, y: i32, col: i32) {
@@ -90,12 +86,70 @@ 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;
         }
     }
 }
@@ -318,14 +372,7 @@ export fn printChar(char: i32) {
 global mut controlCodeLength = 0;
 
 fn printSingleChar(char: i32) {
-    if char >= 4 & char <= 6 {
+    if outputChannel >= 2 & (char < 4 | char > 6) {
-        outputChannel = char - 4;
-	    textCursorX = 0;
-        textCursorY = 0;
-        return;
-    }
-
-    if outputChannel >= 2 {
         logChar(char);
         return;
     }
@@ -343,6 +390,15 @@ fn printSingleChar(char: i32) {
         return;
     }
 
+    if char >= 4 & char <= 6 {
+        outputChannel = char - 4;
+        if !outputChannel {
+            textCursorX = 0;
+            textCursorY = 0;
+        }
+        return;
+    }
+
     if char == 7 {
         80?0 = 80?0 ^ 2;
         return;
@@ -546,13 +602,6 @@ export fn endFrame() {
     68!4 = 68!0;
 }
 
-fn memclr(base: i32, size: i32) {
-    loop bytes {
-        (base + (size := size - 1))?0 = 0;
-        branch_if size: bytes;
-    }
-}
-
 start fn setup() {
     let i: i32 = 12*16*3-1;
     let avg: f32;
@@ -585,9 +634,9 @@ start fn setup() {
         branch_if (i := i - 1) >= 0: expand_sweetie;
     }
 
-    memclr(0, 64);
+    memory.fill(0, 0, 64);
-    memclr(112, 8);
+    memory.fill(112, 0, 8);
-    memclr(0x14000, 0x2c000);
+    memory.fill(0x14000, 0, 0x2c000);
 
 
     cls(0);

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](v0.2.0#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
-* [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
+* [Fireworks](v0.2.0#AgwvgP+M59snqjl4CMKw5sqm1Zw9yJCbSviMjeLUdHus2a3yl/a99+uiBeqZgP/2jqSjrLjRk73COMM6OSLpsxK8ugT1kuk/q4hQUqqPpGozHoa0laulzGGcahzdfdJsYaK1sIdeIYS9M5PnJx/Wk9H+PvWEPy2Zvv7I6IW7Fg==) (127 bytes): Some fireworks to welcome 2022.
-* [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
+* [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](v0.1.2#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
+* [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](v0.1.2#At/p39+IBnj6ry1TRe7jzVy2A4tXgBvmoW2itzoyF2aM28pGy5QDiKxqrk8l9sbWZLtnAb+jgOfU+9QhpuyCAkhN6gPOU481IUL/df96vNe3h288Dqwhd3sfFpothIVFsMwRK72kW2hiR7zWsaXyy5pNmjR6BJk4piWx9ApT1ZwoUajhk6/zij6itq/FD1U3jj/J3MOwqZ2ef8Bv6ZPQlJIYVf62icGa69wS6SI1qBpIFiF14F8PcztRVbKIxLpT4ArCS6nz6FPnyUkqATGSBNPJ): Just a simple viewer for the default font and palette.
+* [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:
 
@@ -29,64 +31,50 @@ Examplers for older versions:
 
 ## Versions
 
-### v0.2.0-rc3
+### v0.2.2
 
-* [Web runtime](v0.2.0-rc3)
+* [Web runtime](v0.2.2)
-* [Linux](https://github.com/exoticorn/microw8/releases/download/v0.2.0-rc3/microw8-0.2.0-rc3-linux.tgz)
+* [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/v0.2.0-rc3/microw8-0.2.0-rc3-macos.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/v0.2.0-rc3/microw8-0.2.0-rc3-windows.zip)
+* [Windows](https://github.com/exoticorn/microw8/releases/download/v0.2.2/microw8-0.2.2-windows.zip)
 
 Changes:
 
-* improve timing stability some more. essentially now guaranteeing that "frame = time_ms * 6 / 100" returns
+* call `start` function after loading cart if the cart exports one
-  consecutive frame numbers, provided the module can be run at 60 fps
+* 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 `else if`
-* * add support for escape sequences in strings
+  * add support for escape sequences in strings
-* * add support for char literals
+  * add support for char literals
-* * add support for binop-assignment, eg. `+=`, `^=`, `<<=` etc. (also support for the tee operator: `+:=`)
+  * 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-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
 
 ### Older versions
 

site/content/docs.md

@@ -12,6 +12,9 @@ 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
 
 ```
@@ -225,30 +228,45 @@ Moving/printing past any border does not cause any special operation, the cursor
 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                            |
+| Code  | Parameters | Operation                                  |
-| ----- | ---------- | ------------------------------------ |
+| ----- | ---------- | ------------------------------------------ |
-| 0     | -          | Nop                                  |
+| 0     | -          | Nop                                        |
-| 1     | char       | Print char (including control chars) |
+| 1     | char       | Print char (including control chars)       |
-| 2-3   | -          | Reserved                             |
+| 2-3   | -          | Reserved                                   |
-| 4     | -          | Switch to normal mode                |
+| 4     | -          | Switch to normal mode, reset cursor to 0,0 |
-| 5     | -          | Switch to graphics mode              |
+| 5     | -          | Switch to graphics mode                    |
-| 6     | -          | Reserved                             |
+| 6     | -          | Switch output to (debug) console           |
-| 7     | -          | Bell / trigger sound channel 0       |
+| 7     | -          | Bell / trigger sound channel 0             |
-| 8     | -          | Move cursor left                     |
+| 8     | -          | Move cursor left                           |
-| 9     | -          | Move cursor right                    |
+| 9     | -          | Move cursor right                          |
-| 10    | -          | Move cursor down                     |
+| 10    | -          | Move cursor down                           |
-| 11    | -          | Move cursor up                       |
+| 11    | -          | Move cursor up                             |
-| 12    | -          | do `cls(background_color)`           |
+| 12    | -          | do `cls(background_color)`                 |
-| 13    | -          | Move cursor to the left border       |
+| 13    | -          | Move cursor to the left border             |
-| 14    | color      | Set the background color             |
+| 14    | color      | Set the background color                   |
-| 15    | color      | Set the text color                   |
+| 15    | color      | Set the text color                         |
-| 16-23 | -          | Reserved                             |
+| 16-23 | -          | Reserved                                   |
-| 24    | -          | Swap text/background colors          |
+| 24    | -          | Swap text/background colors                |
-| 25-30 | -          | Reserved                             |
+| 25-30 | -          | Reserved                                   |
-| 31    | x, y       | Set cursor position (*)              |
+| 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 implementation 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.
@@ -421,6 +439,30 @@ 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

@@ -2,6 +2,69 @@
 description = "Versions"
 +++
 
+### 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)

site/templates/index.html

@@ -4,7 +4,7 @@
         <section>
             <h1 class="text-center heading-text">A WebAssembly based fantasy console</h1>
         </section>
-        <a href="v0.1.2">
+        <a href="v0.2.2">
             <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 self.rx.try_recv() {
-        match event {
+            Ok(event) => match event.kind {
-            Ok(DebouncedEvent::Create(path) | DebouncedEvent::Write(path)) => {
+                DebouncedEventKind::Any => {
-                let handle = same_file::Handle::from_path(&path)?;
+                    let handle = same_file::Handle::from_path(&event.path)?;
-                for file in &self.watched_files {
+                    for file in &self.watched_files {
-                    if handle == same_file::Handle::from_path(file)? {
+                        if handle == same_file::Handle::from_path(file)? {
-                        return Ok(Some(path));
+                            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

@@ -13,6 +13,7 @@ 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
@@ -52,6 +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();
@@ -79,7 +81,17 @@ fn run(mut args: Arguments) -> Result<()> {
     #[cfg(not(feature = "native"))]
     let run_browser = args.contains(["-b", "--browser"]) || true;
 
-    let disable_audio = args.contains(["-m", "--disable-audio"]);
+    #[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()))?;
 
@@ -92,7 +104,7 @@ fn run(mut args: Arguments) -> Result<()> {
         unimplemented!();
         #[cfg(feature = "native")]
         {
-            let mut microw8 = MicroW8::new()?;
+            let mut microw8 = MicroW8::new(timeout, window_config)?;
             if disable_audio {
                 microw8.disable_audio();
             }
@@ -105,10 +117,6 @@ 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() {
@@ -167,7 +175,11 @@ fn load_cart(filename: &Path, config: &Config) -> (Result<Vec<u8>>, Vec<PathBuf>
 
         if let Some(ref pack_config) = config.pack {
             cart = uw8_tool::pack(&cart, pack_config)?;
-            println!("packed size: {} bytes", cart.len());
+            println!(
+                "\npacked size: {} bytes ({:.2})",
+                cart.len(),
+                uw8_tool::compressed_size(&cart)
+            );
         }
 
         if let Some(ref path) = config.output_path {

src/run_native.rs

@@ -4,31 +4,21 @@ use std::{thread, time::Instant};
 
 use anyhow::{anyhow, Result};
 use cpal::traits::*;
-use minifb::{Key, Window, WindowOptions};
 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>,
-    timeout: u32,
     disable_audio: bool,
+    module_data: Option<Vec<u8>>,
+    timeout: u32,
+    instance: Option<UW8Instance>,
 }
 
 struct UW8Instance {
@@ -37,10 +27,8 @@ struct UW8Instance {
     end_frame: TypedFunc<(), ()>,
     update: Option<TypedFunc<(), ()>>,
     start_time: Instant,
-    next_frame: Instant,
-    module: Vec<u8>,
     watchdog: Arc<Mutex<UW8WatchDog>>,
-    sound: Option<Uw8Sound>,
+    sound_tx: Option<mpsc::SyncSender<RegisterUpdate>>,
 }
 
 impl Drop for UW8Instance {
@@ -52,44 +40,36 @@ impl Drop for UW8Instance {
 }
 
 struct UW8WatchDog {
-    interupt: wasmtime::InterruptHandle,
+    engine: Engine,
-    timeout: u32,
     stop: bool,
 }
 
 impl MicroW8 {
-    pub fn new() -> Result<MicroW8> {
+    pub fn new(timeout: Option<u32>, window_config: WindowConfig) -> Result<MicroW8> {
-        let engine = wasmtime::Engine::new(wasmtime::Config::new().interruptable(true))?;
+        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 {
+        let window = Window::new(window_config)?;
-            scale: minifb::Scale::X2,
-            scale_mode: minifb::ScaleMode::AspectRatioStretch,
-            resize: true,
-            ..Default::default()
-        };
-        let window = Window::new("MicroW8", 320, 240, options)?;
 
         Ok(MicroW8 {
+            window,
+            stream: None,
             engine,
             loader_module,
-            window,
-            window_buffer: vec![0u32; 320 * 240],
-            instance: None,
-            timeout: 30,
             disable_audio: false,
+            module_data: None,
+            timeout: timeout.unwrap_or(0),
+            instance: None,
         })
     }
 
-    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;
     }
@@ -97,25 +77,23 @@ impl MicroW8 {
 
 impl super::Runtime for MicroW8 {
     fn is_open(&self) -> bool {
-        self.window.is_open() && !self.window.is_key_down(Key::Escape)
+        self.window.is_open()
-    }
-
-    fn set_timeout(&mut self, timeout: u32) {
-        self.timeout = timeout;
     }
 
     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);
@@ -133,8 +111,7 @@ impl super::Runtime for MicroW8 {
         let platform_instance = instantiate_platform(&mut linker, &mut store, &platform_module)?;
 
         let watchdog = Arc::new(Mutex::new(UW8WatchDog {
-            interupt: store.interrupt_handle()?,
+            engine: self.engine.clone(),
-            timeout: self.timeout,
             stop: false,
         }));
 
@@ -142,16 +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.engine.increment_epoch();
-                        watchdog.timeout -= 1;
-                        if watchdog.timeout == 0 {
-                            watchdog.interupt.interrupt();
-                        }
-                    }
                 } else {
                     break;
                 }
@@ -159,23 +131,24 @@ impl super::Runtime for MicroW8 {
         }
 
         let instance = linker.instantiate(&mut store, &module)?;
-        if let Ok(mut watchdog) = watchdog.lock() {
+        let end_frame = platform_instance.get_typed_func::<(), ()>(&mut store, "endFrame")?;
-            watchdog.timeout = 0;
+        let update = instance.get_typed_func::<(), ()>(&mut store, "upd").ok();
-        }
-        let end_frame = platform_instance.get_typed_func::<(), (), _>(&mut store, "endFrame")?;
-        let update = instance.get_typed_func::<(), (), _>(&mut store, "upd").ok();
 
-        let sound = if self.disable_audio {
+        if let Some(start) = instance.get_typed_func::<(), ()>(&mut store, "start").ok() {
-            None
+            start.call(&mut store, ())?;
+        }
+
+        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)
+                    (Some(sound.tx), Some(sound.stream))
                 }
                 Err(err) => {
                     eprintln!("Failed to init sound: {}", err);
-                    None
+                    (None, None)
                 }
             }
         };
@@ -186,57 +159,45 @@ impl super::Runtime for MicroW8 {
             end_frame,
             update,
             start_time: Instant::now(),
-            next_frame: Instant::now(),
-            module: module_data.into(),
             watchdog,
-            sound,
+            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() {
-            {
-                if let Some(sleep) = instance.next_frame.checked_duration_since(Instant::now()) {
-                    std::thread::sleep(sleep);
-                }
-            }
-
-            let now = Instant::now();
             let time = (now - instance.start_time).as_millis() as i32;
-            {
+            let next_frame = {
                 let offset = ((time as u32 as i64 * 6) % 100 - 50) / 6;
-                instance.next_frame = now + Duration::from_millis((16 - offset) as u64);
+                let max = now + Duration::from_millis(17);
-            }
+                let next_center = now + Duration::from_millis((16 - offset) as u64);
+                next_center.min(max)
+            };
 
             {
-                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 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() {
+            instance.store.set_epoch_deadline(self.timeout as u64);
-                watchdog.timeout = self.timeout;
-            }
             if let Some(ref update) = instance.update {
-                result = update.call(&mut instance.store, ());
+                if let Err(err) = update.call(&mut instance.store, ()) {
-            }
+                    result = Err(err);
-            if let Ok(mut watchdog) = instance.watchdog.lock() {
+                }
-                watchdog.timeout = 0;
             }
             instance.end_frame.call(&mut instance.store, ())?;
 
@@ -244,33 +205,23 @@ impl super::Runtime for MicroW8 {
 
             let mut sound_regs = [0u8; 32];
             sound_regs.copy_from_slice(&memory[80..112]);
-            if let Some(ref sound) = instance.sound {
+            if let Some(ref sound_tx) = instance.sound_tx {
-                sound.tx.send(RegisterUpdate {
+                let _ = sound_tx.send(RegisterUpdate {
                     time,
                     data: sound_regs,
-                })?;
+                });
             }
 
-            let framebuffer = &memory[120..(120 + 320 * 240)];
+            let framebuffer_mem = &memory[120..(120 + 320 * 240)];
-            let palette = &memory[0x13000..];
+            let palette_mem = &memory[0x13000..];
-            for (i, &color_index) in framebuffer.iter().enumerate() {
+            self.window
-                let offset = color_index as usize * 4;
+                .end_frame(framebuffer_mem, palette_mem, next_frame);
-                self.window_buffer[i] = 0xff000000
-                    | ((palette[offset] as u32) << 16)
-                    | ((palette[offset + 1] as u32) << 8)
-                    | palette[offset + 2] as u32;
-            }
 
-            if self.window.is_key_pressed(Key::R, minifb::KeyRepeat::No) {
+            if result.is_ok() {
-                self.load(&instance.module)?;
-            } else if result.is_ok() {
                 self.instance = Some(instance);
             }
         }
 
-        self.window
-            .update_with_buffer(&self.window_buffer, 320, 240)?;
-
         result?;
         Ok(())
     }
@@ -304,15 +255,12 @@ fn add_native_functions(
         }
     })?;
     for i in 0..16 {
-        linker.define(
+        let global = wasmtime::Global::new(
-            "env",
+            &mut *store,
-            &format!("g_reserved{}", i),
+            GlobalType::new(ValType::I32, Mutability::Const),
-            wasmtime::Global::new(
+            0.into(),
-                &mut *store,
-                GlobalType::new(ValType::I32, Mutability::Const),
-                0.into(),
-            )?,
         )?;
+        linker.define(&store, "env", &format!("g_reserved{}", i), global)?;
     }
 
     Ok(())
@@ -325,14 +273,18 @@ fn instantiate_platform(
 ) -> Result<wasmtime::Instance> {
     let platform_instance = linker.instantiate(&mut *store, &platform_module)?;
 
-    for export in platform_instance.exports(&mut *store) {
+    let exports: Vec<(String, Func)> = platform_instance
-        linker.define(
+        .exports(&mut *store)
-            "env",
+        .map(|e| {
-            export.name(),
+            (
-            export
+                e.name().to_owned(),
-                .into_func()
+                e.into_func()
-                .expect("platform surely only exports functions"),
+                    .expect("platform surely only exports functions"),
-        )?;
+            )
+        })
+        .collect();
+    for (name, func) in exports {
+        linker.define(&store, "env", &name, func)?;
     }
 
     Ok(platform_instance)
@@ -354,19 +306,20 @@ fn init_sound(
     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("env", "memory", memory)?;
+    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")
+        .get_typed_func::<(i32,), f32>(&mut store, "snd")
-        .or_else(|_| platform_instance.get_typed_func::<(i32,), f32, _>(&mut store, "gesSnd"))?;
+        .or_else(|_| platform_instance.get_typed_func::<(i32,), f32>(&mut store, "sndGes"))?;
 
     let host = cpal::default_host();
     let device = host
@@ -375,33 +328,37 @@ fn init_sound(
     let mut configs: Vec<_> = device
         .supported_output_configs()?
         .filter(|config| {
-            config.channels() == 2 && config.sample_format() == cpal::SampleFormat::F32
+            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);
-        if rate >= 44100 {
+        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 output config"))?;
+        .ok_or_else(|| anyhow!("Could not find float or 16bit signed output config"))?;
     let sample_rate = cpal::SampleRate(44100)
         .max(config.min_sample_rate())
-        .max(config.max_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(65536.max(min).min(max))
+            cpal::BufferSize::Fixed(256.max(min).min(max))
         }
     };
+    let sample_format = config.sample_format();
     let config = cpal::StreamConfig {
         buffer_size,
         ..config.config()
@@ -421,8 +378,8 @@ fn init_sound(
         None
     } else {
         let rs = rubato::FftFixedIn::new(44100, sample_rate, 128, 1, 2)?;
-        let input_buffers = rs.input_buffer_allocate();
+        let input_buffers = rs.input_buffer_allocate(true);
-        let output_buffers = rs.output_buffer_allocate();
+        let output_buffers = rs.output_buffer_allocate(true);
         Some(Resampler {
             resampler: rs,
             input_buffers,
@@ -434,95 +391,130 @@ fn init_sound(
     let mut sample_index = 0;
     let mut pending_updates: Vec<RegisterUpdate> = Vec::with_capacity(30);
     let mut current_time = 0;
-    let stream = device.build_output_stream(
+
-        &config,
+    let mut callback = move |mut outer_buffer: &mut [f32], _: &_| {
-        move |mut outer_buffer: &mut [f32], _| {
+        let mut first_update = true;
-            let mut first_update = true;
+        while let Ok(update) = rx.try_recv() {
-            while let Ok(update) = rx.try_recv() {
+            if first_update {
-                if first_update {
+                current_time += update.time.wrapping_sub(current_time) / 8;
-                    current_time += update.time.wrapping_sub(current_time) / 8;
+                first_update = false;
-                    first_update = false;
+            }
-                }
+            pending_updates.push(update);
-                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();
             }
 
-            while !outer_buffer.is_empty() {
+            let duration = if let Some(update) = pending_updates.first() {
-                while pending_updates
+                ((update.time.wrapping_sub(current_time) as usize) * sample_rate + 999) / 1000
-                    .first()
+            } else {
-                    .into_iter()
+                outer_buffer.len()
-                    .any(|u| u.time.wrapping_sub(current_time) <= 0)
+            };
-                {
+            let step_size = (duration.max(64) * 2).min(outer_buffer.len());
-                    let update = pending_updates.remove(0);
-                    memory.write(&mut store, 80, &update.data).unwrap();
-                }
 
-                let duration = if let Some(update) = pending_updates.first() {
+            let mut buffer = &mut outer_buffer[..step_size];
-                    ((update.time.wrapping_sub(current_time) as usize) * sample_rate + 999) / 1000
+
+            {
+                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 {
-                    outer_buffer.len()
+                    s.max(-1.0).min(1.0)
-                };
-                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());
                 }
+            }
 
-                if let Some(ref mut resampler) = resampler {
+            if let Some(ref mut resampler) = resampler {
-                    while !buffer.is_empty() {
+                while !buffer.is_empty() {
-                        let copy_size = resampler.output_buffers[0]
+                    let copy_size = resampler.output_buffers[0]
-                            .len()
+                        .len()
-                            .saturating_sub(resampler.output_index)
+                        .saturating_sub(resampler.output_index)
-                            .min(buffer.len() / 2);
+                        .min(buffer.len() / 2);
-                        if copy_size == 0 {
+                    if copy_size == 0 {
-                            resampler.input_buffers[0].clear();
+                        resampler.input_buffers[0].clear();
-                            resampler.input_buffers[1].clear();
+                        resampler.input_buffers[1].clear();
-                            for _ in 0..resampler.resampler.input_frames_next() {
+                        for _ in 0..resampler.resampler.input_frames_next() {
-                                resampler.input_buffers[0]
+                            resampler.input_buffers[0].push(clamp_sample(
-                                    .push(snd.call(&mut store, (sample_index,)).unwrap_or(0.0));
+                                snd.call(&mut store, (sample_index,)).unwrap_or(0.0),
-                                resampler.input_buffers[1]
+                            ));
-                                    .push(snd.call(&mut store, (sample_index + 1,)).unwrap_or(0.0));
+                            resampler.input_buffers[1].push(clamp_sample(
-                                sample_index = sample_index.wrapping_add(2);
+                                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 {
+                        resampler
-                    for v in buffer {
+                            .resampler
-                        *v = snd.call(&mut store, (sample_index,)).unwrap_or(0.0);
+                            .process_into_buffer(
-                        sample_index = sample_index.wrapping_add(1);
+                                &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 {
-                outer_buffer = &mut outer_buffer[step_size..];
+                for v in buffer {
-                current_time =
+                    *v = clamp_sample(snd.call(&mut store, (sample_index,)).unwrap_or(0.0));
-                    current_time.wrapping_add((step_size * 500 / sample_rate).max(1) as i32);
+                    sample_index = sample_index.wrapping_add(1);
+                }
             }
-        },
+
-        move |err| {
+            outer_buffer = &mut outer_buffer[step_size..];
-            dbg!(err);
+            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"
+                let server_future = warp::serve(html.or(cart).or(events)).bind(server_addr);
-                    .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));
                 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);
         }
@@ -86,4 +94,4 @@ impl Default for RunWebServer {
     fn default() -> RunWebServer {
         RunWebServer::new()
     }
-}
+}

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/ges_test.cwa

@@ -2,17 +2,18 @@ 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 {
-    gesSnd(t)
+    sndGes(t)
 }
 
 export fn upd() {
     80?0 = 32!32 / 200 & 2 | 0x41;
-    80?3 = (32!32 / 400)%7*12/7 + 40;
+    80?3 = (32!32 / 400)%8*12/7 + 40;
     let pulse = (32!32 * 256 / 2000) & 511;
     if pulse >= 256 {
         pulse = 511 - pulse;

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

@@ -56,6 +56,21 @@ dependencies = [
  "lazy_static",
 ]
 
+[[package]]
+name = "form_urlencoded"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a9c384f161156f5260c24a097c56119f9be8c798586aecc13afbcbe7b7e26bf8"
+dependencies = [
+ "percent-encoding",
+]
+
+[[package]]
+name = "hashbrown"
+version = "0.12.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "8a9ee70c43aaf417c914396645a0fa852624801b24ebb7ae78fe8272889ac888"
+
 [[package]]
 name = "heck"
 version = "0.3.3"
@@ -71,6 +86,26 @@ version = "2.2.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "25a2bc672d1148e28034f176e01fffebb08b35768468cc954630da77a1449005"
 
+[[package]]
+name = "idna"
+version = "0.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e14ddfc70884202db2244c223200c204c2bda1bc6e0998d11b5e024d657209e6"
+dependencies = [
+ "unicode-bidi",
+ "unicode-normalization",
+]
+
+[[package]]
+name = "indexmap"
+version = "1.9.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1885e79c1fc4b10f0e172c475f458b7f7b93061064d98c3293e98c5ba0c8b399"
+dependencies = [
+ "autocfg",
+ "hashbrown",
+]
+
 [[package]]
 name = "lazy_static"
 version = "1.4.0"
@@ -83,6 +118,12 @@ version = "0.2.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "884e2677b40cc8c339eaefcb701c32ef1fd2493d71118dc0ca4b6a736c93bd67"
 
+[[package]]
+name = "lexopt"
+version = "0.2.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "478ee9e62aaeaf5b140bd4138753d1f109765488581444218d3ddda43234f3e8"
+
 [[package]]
 name = "libc"
 version = "0.2.112"
@@ -120,10 +161,16 @@ dependencies = [
 ]
 
 [[package]]
-name = "pico-args"
+name = "percent-encoding"
-version = "0.4.2"
+version = "2.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db8bcd96cb740d03149cbad5518db9fd87126a10ab519c011893b1754134c468"
+checksum = "478c572c3d73181ff3c2539045f6eb99e5491218eae919370993b890cdbdd98e"
+
+[[package]]
+name = "pico-args"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5be167a7af36ee22fe3115051bc51f6e6c7054c9348e28deb4f49bd6f705a315"
 
 [[package]]
 name = "proc-macro2"
@@ -163,6 +210,26 @@ dependencies = [
  "unicode-xid",
 ]
 
+[[package]]
+name = "thiserror"
+version = "1.0.38"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "6a9cd18aa97d5c45c6603caea1da6628790b37f7a34b6ca89522331c5180fed0"
+dependencies = [
+ "thiserror-impl",
+]
+
+[[package]]
+name = "thiserror-impl"
+version = "1.0.38"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1fb327af4685e4d03fa8cbcf1716380da910eeb2bb8be417e7f9fd3fb164f36f"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "time"
 version = "0.1.44"
@@ -174,6 +241,36 @@ dependencies = [
  "winapi",
 ]
 
+[[package]]
+name = "tinyvec"
+version = "1.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "87cc5ceb3875bb20c2890005a4e226a4651264a5c75edb2421b52861a0a0cb50"
+dependencies = [
+ "tinyvec_macros",
+]
+
+[[package]]
+name = "tinyvec_macros"
+version = "0.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "cda74da7e1a664f795bb1f8a87ec406fb89a02522cf6e50620d016add6dbbf5c"
+
+[[package]]
+name = "unicode-bidi"
+version = "0.3.10"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d54675592c1dbefd78cbd98db9bacd89886e1ca50692a0692baefffdeb92dd58"
+
+[[package]]
+name = "unicode-normalization"
+version = "0.1.22"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "5c5713f0fc4b5db668a2ac63cdb7bb4469d8c9fed047b1d0292cc7b0ce2ba921"
+dependencies = [
+ "tinyvec",
+]
+
 [[package]]
 name = "unicode-segmentation"
 version = "1.8.0"
@@ -188,13 +285,24 @@ checksum = "8ccb82d61f80a663efe1f787a51b16b5a51e3314d6ac365b08639f52387b33f3"
 
 [[package]]
 name = "upkr"
-version = "0.1.0"
+version = "0.2.1"
-source = "git+https://github.com/exoticorn/upkr.git?rev=2e7983fc#2e7983fc650788d98da2eecef2d16f63e849e4a0"
+source = "git+https://github.com/exoticorn/upkr.git?rev=080db40d0088bbee2bdf3c5c75288ac7853d6b7a#080db40d0088bbee2bdf3c5c75288ac7853d6b7a"
 dependencies = [
  "anyhow",
  "cdivsufsort",
- "pbr",
+ "lexopt",
- "pico-args",
+ "thiserror",
+]
+
+[[package]]
+name = "url"
+version = "2.3.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0d68c799ae75762b8c3fe375feb6600ef5602c883c5d21eb51c09f22b83c4643"
+dependencies = [
+ "form_urlencoded",
+ "idna",
+ "percent-encoding",
 ]
 
 [[package]]
@@ -207,7 +315,7 @@ dependencies = [
  "upkr",
  "walrus",
  "wasm-encoder",
- "wasmparser 0.81.0",
+ "wasmparser 0.99.0",
 ]
 
 [[package]]
@@ -244,9 +352,9 @@ checksum = "1a143597ca7c7793eff794def352d41792a93c481eb1042423ff7ff72ba2c31f"
 
 [[package]]
 name = "wasm-encoder"
-version = "0.8.0"
+version = "0.22.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "db0c351632e46cc06a58a696a6c11e4cf90cad4b9f8f07a0b59128d616c29bb0"
+checksum = "ef126be0e14bdf355ac1a8b41afc89195289e5c7179f80118e3abddb472f0810"
 dependencies = [
  "leb128",
 ]
@@ -259,9 +367,13 @@ checksum = "b35c86d22e720a07d954ebbed772d01180501afe7d03d464f413bb5f8914a8d6"
 
 [[package]]
 name = "wasmparser"
-version = "0.81.0"
+version = "0.99.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "98930446519f63d00a836efdc22f67766ceae8dbcc1571379f2bcabc6b2b9abc"
+checksum = "9ef3b717afc67f848f412d4f02c127dd3e35a0eecd58c684580414df4fde01d3"
+dependencies = [
+ "indexmap",
+ "url",
+]
 
 [[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"
+wasmparser = "0.99"
-wasm-encoder = "0.8"
+wasm-encoder = "0.22"
 walrus = "0.19"
 anyhow = "1"
-pico-args = "0.4"
+pico-args = "0.5"
-upkr = { git = "https://github.com/exoticorn/upkr.git", rev = "2e7983fc" }
+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::*;
@@ -167,6 +167,7 @@ impl BaseModule {
         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));
 
         for i in functions.len()..64 {
             add_function(
@@ -217,13 +218,13 @@ 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,
@@ -233,11 +234,12 @@ impl BaseModule {
 
             imports.import(
                 "env",
-                Some("memory"),
+                "memory",
                 MemoryType {
                     minimum: self.memory as u64,
                     maximum: None,
                     memory64: false,
+                    shared: false,
                 },
             );
 
@@ -258,7 +260,7 @@ 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);
@@ -286,7 +288,7 @@ 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(())
     }

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() {
+    let exports_to_delete: Vec<_> = module
-        "upd" => None,
+        .exports
-        _ => Some(export.id())
+        .iter()
-    }).collect();
+        .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);
@@ -18,4 +22,4 @@ pub fn filter_exports(in_path: &Path, out_path: &Path) -> Result<()> {
     module.emit_wasm_file(out_path)?;
 
     Ok(())
-}
+}

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> {
+fn to_val_type(type_: &wasmparser::ValType) -> Result<ValType> {
-    use wasmparser::Type::*;
+    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) => {
+                Payload::MemorySection(reader) => {
-                    let mut elements = Vec::with_capacity(reader.get_count() as usize);
+                    if reader.count() != 0 {
-                    for _ in 0..reader.get_count() {
+                        bail!("Found non-empty MemorySection. Memory has to be imported!");
-                        elements.push(Element::parse(reader.read()?)?);
+                    }
+                }
+                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);
             }
@@ -480,7 +490,7 @@ impl<'a> ParsedModule<'a> {
                 }
                 element_section.active(
                     None,
-                    &wasm_encoder::Instruction::I32Const(element.start_index as i32),
+                    &wasm_encoder::ConstExpr::i32_const(element.start_index as i32),
                     ValType::FuncRef,
                     wasm_encoder::Elements::Functions(&functions),
                 );
@@ -529,28 +539,27 @@ fn read_type_section(reader: TypeSectionReader) -> Result<Vec<base_module::Funct
 
     for type_def in reader {
         match type_def? {
-            wasmparser::TypeDef::Func(fnc) => {
+            wasmparser::Type::Func(fnc) => {
-                if fnc.returns.len() > 1 {
+                if fnc.results().len() > 1 {
                     bail!("Multi-value not supported");
                 }
-                let params = to_val_type_vec(&fnc.params)?;
+                let params = to_val_type_vec(fnc.params())?;
-                let result = to_val_type_vec(&fnc.returns)?.into_iter().next();
+                let result = to_val_type_vec(fnc.results())?.into_iter().next();
                 function_types.push(FunctionType { params, result });
             }
-            t => bail!("Unsupported type def {:?}", t),
         }
     }
 
     Ok(function_types)
 }
 
-fn validate_table_section(mut reader: TableSectionReader) -> Result<()> {
+fn validate_table_section(reader: TableSectionReader) -> Result<()> {
-    if reader.get_count() != 1 {
+    if reader.count() != 1 {
         bail!("Only up to one table supported");
     }
 
-    let type_ = reader.read()?;
+    let type_ = reader.into_iter().next().unwrap()?;
-    if type_.element_type != wasmparser::Type::FuncRef {
+    if type_.element_type != wasmparser::ValType::FuncRef {
         bail!("Only one funcref table is supported");
     }
 
@@ -584,45 +593,38 @@ impl ImportSection {
 
         for import in reader {
             let import = import?;
-            if let Some(field) = import.field {
+            match import.ty {
-                match import.ty {
+                TypeRef::Func(type_) => {
-                    ImportSectionEntryType::Function(type_) => {
+                    functions.push(FunctionImport {
-                        functions.push(FunctionImport {
+                        module: import.module.to_string(),
-                            module: import.module.to_string(),
+                        field: import.name.to_string(),
-                            field: field.to_string(),
+                        type_,
-                            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),
                 }
-            } else {
+                TypeRef::Memory(mem) => {
-                bail!(
+                    if import.module != "env" || import.name != "memory" {
-                    "Found import without field, only module '{}'",
+                        bail!(
-                    import.module
+                            "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 +644,37 @@ struct Element {
 
 impl Element {
     fn parse(element: wasmparser::Element) -> Result<Element> {
-        if element.ty != wasmparser::Type::FuncRef {
+        match element.items {
-            bail!("Table element type is not FuncRef");
+            wasmparser::ElementItems::Functions(funcs_reader) => {
-        }
+                let start_index = if let wasmparser::ElementKind::Active {
+                    offset_expr,
+                    table_index: 0,
+                } = element.kind
+                {
+                    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 {
+                let mut functions = Vec::with_capacity(funcs_reader.count() as usize);
-            init_expr,
+                for index in funcs_reader {
-            table_index: 0,
+                    functions.push(index?);
-        } = element.kind
+                }
-        {
+
-            let mut init_reader = init_expr.get_operators_reader();
+                Ok(Element {
-            if let wasmparser::Operator::I32Const { value: start_index } = init_reader.read()? {
+                    start_index,
-                start_index as u32
+                    functions,
-            } 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");
             }
+            _ => bail!("Table element type is not FuncRef"),
         }
-
-        Ok(Element {
-            start_index,
-            functions,
-        })
     }
 }
 
@@ -706,8 +705,8 @@ fn read_export_section(reader: ExportSectionReader) -> Result<Vec<(String, u32)>
     for export in reader {
         let export = export?;
         match export.kind {
-            ExternalKind::Function => {
+            ExternalKind::Func => {
-                function_exports.push((export.field.to_string(), export.index));
+                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 +727,11 @@ 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) => {
+            wasmparser::BlockType::Empty => enc::BlockType::Empty,
-                enc::BlockType::Empty
+            wasmparser::BlockType::Type(ty) => enc::BlockType::Result(to_val_type(&ty)?),
-            }
+            wasmparser::BlockType::FuncType(ty) => enc::BlockType::FunctionType(
-            wasmparser::TypeOrFuncType::Type(ty) => enc::BlockType::Result(to_val_type(&ty)?),
-            wasmparser::TypeOrFuncType::FuncType(ty) => enc::BlockType::FunctionType(
                 *type_map
                     .get(&ty)
                     .ok_or_else(|| anyhow!("Function type index out of range: {}", ty))?,
@@ -748,7 +745,7 @@ 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 +760,31 @@ fn remap_function(
         function.instruction(&match op? {
             De::Unreachable => En::Unreachable,
             De::Nop => En::Nop,
-            De::Block { ty } => En::Block(block_type(ty)?),
+            De::Block { blockty } => En::Block(block_type(blockty)?),
-            De::Loop { ty } => En::Loop(block_type(ty)?),
+            De::Loop { blockty } => En::Loop(block_type(blockty)?),
-            De::If { ty } => En::If(block_type(ty)?),
+            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 +804,16 @@ 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::I32Load8S { memarg } => En::I32Load8S(mem(memarg)),
-            De::I32Load8U { memarg } => En::I32Load8_U(mem(memarg)),
+            De::I32Load8U { memarg } => En::I32Load8U(mem(memarg)),
-            De::I32Load16S { memarg } => En::I32Load16_S(mem(memarg)),
+            De::I32Load16S { memarg } => En::I32Load16S(mem(memarg)),
-            De::I32Load16U { memarg } => En::I32Load16_U(mem(memarg)),
+            De::I32Load16U { memarg } => En::I32Load16U(mem(memarg)),
-            De::I64Load8S { memarg } => En::I64Load8_S(mem(memarg)),
+            De::I64Load8S { memarg } => En::I64Load8S(mem(memarg)),
-            De::I64Load8U { memarg } => En::I64Load8_U(mem(memarg)),
+            De::I64Load8U { memarg } => En::I64Load8U(mem(memarg)),
-            De::I64Load16S { memarg } => En::I64Load16_S(mem(memarg)),
+            De::I64Load16S { memarg } => En::I64Load16S(mem(memarg)),
-            De::I64Load16U { memarg } => En::I64Load16_U(mem(memarg)),
+            De::I64Load16U { memarg } => En::I64Load16U(mem(memarg)),
-            De::I64Load32S { memarg } => En::I64Load32_S(mem(memarg)),
+            De::I64Load32S { memarg } => En::I64Load32S(mem(memarg)),
-            De::I64Load32U { memarg } => En::I64Load32_U(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 +832,7 @@ 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 +843,7 @@ 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 +853,13 @@ 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 +966,246 @@ 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 = "0.28.6"
+env_logger = "0.10"
+log = "0.4"
+pico-args = "0.5"
+wgpu = "0.17"
+pollster = "0.3.0"
+bytemuck = { version = "1.13", 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 +1,554 @@
+use crate::{Input, WindowConfig, WindowImpl};
+use anyhow::{anyhow, Result};
+use std::time::Instant;
+
+use winit::{
+    dpi::PhysicalSize,
+    event::{Event, VirtualKeyCode, WindowEvent},
+    event_loop::{ControlFlow, EventLoop},
+    window::{Fullscreen, WindowBuilder},
+};
+
+use winit::platform::run_return::EventLoopExtRunReturn;
+
+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,
+    _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: 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 instance = wgpu::Instance::new(Default::default());
+            let surface = unsafe { instance.create_surface(&window) }?;
+            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.run_return(|event, _, control_flow| {
+            *control_flow = ControlFlow::WaitUntil(self.next_frame);
+            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 => {
+                        self.is_open = false;
+                        *control_flow = ControlFlow::Exit;
+                    }
+                    WindowEvent::KeyboardInput { input, .. } => {
+                        fn gamepad_button(input: &winit::event::KeyboardInput) -> u8 {
+                            match input.scancode {
+                                44 => 16,
+                                45 => 32,
+                                30 => 64,
+                                31 => 128,
+                                _ => match input.virtual_keycode {
+                                    Some(VirtualKeyCode::Up) => 1,
+                                    Some(VirtualKeyCode::Down) => 2,
+                                    Some(VirtualKeyCode::Left) => 4,
+                                    Some(VirtualKeyCode::Right) => 8,
+                                    _ => 0,
+                                },
+                            }
+                        }
+                        if input.state == winit::event::ElementState::Pressed {
+                            match input.virtual_keycode {
+                                Some(VirtualKeyCode::Escape) => {
+                                    self.is_open = false;
+                                    *control_flow = ControlFlow::Exit;
+                                }
+                                Some(VirtualKeyCode::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);
+                                }
+                                Some(VirtualKeyCode::R) => reset = true,
+                                Some(VirtualKeyCode::Key1) => new_filter = Some(1),
+                                Some(VirtualKeyCode::Key2) => new_filter = Some(2),
+                                Some(VirtualKeyCode::Key3) => new_filter = Some(3),
+                                Some(VirtualKeyCode::Key4) => new_filter = Some(4),
+                                Some(VirtualKeyCode::Key5) => new_filter = Some(5),
+                                _ => (),
+                            }
+
+                            self.gamepads[0] |= gamepad_button(&input);
+                        } else {
+                            self.gamepads[0] &= !gamepad_button(&input);
+                        }
+                    }
+                    _ => (),
+                },
+                Event::RedrawEventsCleared => {
+                    if Instant::now() >= self.next_frame
+                        // workaround needed on Wayland until the next winit release
+                        && self.window.fullscreen().is_some() == self.is_fullscreen
+                    {
+                        *control_flow = ControlFlow::Exit
+                    }
+                }
+                _ => (),
+            }
+            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,
+                );
+            }
+        });
+        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: true,
+                    },
+                })],
+                depth_stencil_attachment: 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: true,
+                },
+            })],
+            depth_stencil_attachment: 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/src/audiolet.js

@@ -63,7 +63,7 @@ class APU extends AudioWorkletProcessor {
 
         this.memory = memory;
 
-        this.snd = instance.exports.snd || platform_instance.exports.gesSnd;
+        this.snd = instance.exports.snd || platform_instance.exports.sndGes;
 
         this.port.postMessage(2);
     }

web/src/index.html

@@ -10,7 +10,7 @@
     </head>
     <body>
         <div id="uw8">
-            <a href="https://exoticorn.github.io/microw8">MicroW8</a> 0.2.0-rc3
+            <a href="https://exoticorn.github.io/microw8">MicroW8</a> 0.2.2
         </div>
         <div id="centered">
             <canvas class="screen" id="screen" width="320" height="240">

web/src/microw8.js

@@ -263,6 +263,10 @@ export default function MicroW8(screen, config = {}) {
             window.addEventListener('blur', () => updateVisibility(false), { signal: abortController.signal });
             updateVisibility(document.hasFocus());
 
+            if (instance.exports.start) {
+                instance.exports.start();
+            }
+
             function mainloop() {
                 if (!keepRunning) {
                     return;