name: CI

on:
  push:
    branches:
      - main
  pull_request:
    branches:
      - main

concurrency:
  group: ci-${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

permissions:
  contents: read

jobs:
  linux:
    name: Build & test (Linux)
    runs-on: warp-ubuntu-latest-x64-8x
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          # gpui/ is a git submodule; its crates are workspace members of
          # their own repo and build as path dependencies of oakapp.
          submodules: true

      - name: Install Rust (stable)
        uses: dtolnay/rust-toolchain@stable

      # ------------------------------------------------------------------
      # System dependencies
      # ------------------------------------------------------------------
      - name: Install system dependencies
        run: |
          # Codec/filter libraries for the ffmpeg-sys-next build feature
          # (see tooling/install-deps.sh).
          tooling/install-deps.sh
          # cmake/make for the vendored OpenColorIO build (ocio-sys
          # `bundled`; Ubuntu's libopencolorio-dev is 2.1, older than the
          # bridge's API floor) plus the headless test infra gpui needs:
          # X11, software Mesa Vulkan (lavapipe) and xvfb.
          sudo apt-get install -y \
            cmake \
            libpipewire-0.3-dev libspa-0.2-dev libjack-jackd2-dev \
            libasound2-dev libpulse-dev libsndfile1-dev \
            libgl1-mesa-dev libgl1-mesa-dri mesa-vulkan-drivers \
            libvulkan-dev libxkbcommon-dev libxkbcommon-x11-dev xvfb build-essential clang libclang-dev

      # ------------------------------------------------------------------
      # Build environment
      # ------------------------------------------------------------------
      # ocio-sys builds a stub bridge unless these are set; the oak-core
      # ocioutils tests need the real library.
      # tooling/ocio-env.sh: vendored static OCIO (the [patch.crates-io]
      # ocio-sys tracks shaloong/ocio-rs main, whose vendored sources build
      # on GCC >= 16).
      - name: Configure build environment
        run: |
          bash tooling/ocio-env.sh >> "$GITHUB_ENV"
          echo "FFMPEG_DIR=.cache/ffmpeg" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Caches
      # ------------------------------------------------------------------
      # Covers the whole target/ dir plus ~/.cargo; shared across branches
      # of the same OS.
      - name: Cache cargo artifacts
        uses: Swatinem/rust-cache@v2
        with:
          shared-key: oak-ci-linux
          cache-on-failure: true

      # The project FFmpeg (release/8.0, static, all free codecs + hwaccel)
      # is built by tooling/ffmpeg/build-ffmpeg.sh — 10-20 min on a cold
      # cache. It does not depend on the Rust toolchain, so key it on the
      # script itself and keep it out of rust-cache.
      - name: Cache project FFmpeg
        uses: actions/cache@v4
        with:
          path: .cache/ffmpeg
          key: ffmpeg-${{ runner.os }}-${{ hashFiles('tooling/ffmpeg/build-ffmpeg.sh') }}

      # ------------------------------------------------------------------
      # Project FFmpeg (script + FFMPEG_DIR; see docs/build.md)
      # ------------------------------------------------------------------
      - name: Build project FFmpeg
        run: |
          tooling/ffmpeg/build-ffmpeg.sh
          echo "FFMPEG_DIR=$PWD/.cache/ffmpeg" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Build & test
      # ------------------------------------------------------------------
      # `cargo check` (not build): the Test step links the test binaries
      # anyway, and a full build would codegen every workspace crate twice
      # (once without and once with cfg(test)).
      - name: Build
        run: cargo check --workspace --locked

      # xvfb + 24-bit screen: the gpui #[gpui::test] tests open real windows
      # and render through wgpu on Mesa's software Vulkan (lavapipe).
      # The watchdog bounds the step: a deadlocked test produces no output
      # and no failure, so after 1800 s it dumps every hung process's
      # thread stacks and kills the suite.
      - name: Test
        run: |
          sudo apt-get install -y gdb
          run_suite() {
            xvfb-run -a -s "-screen 0 1920x1080x24" cargo test --workspace --locked &
            TEST_PID=$!
            # The watchdog inherits the step's stdout/stderr; detach it so
            # it cannot keep the runner's I/O pipes open after the step
            # ends ("WaitDelay expired before I/O complete" otherwise).
            (
              sleep 1800
              echo "::warning::test suite exceeded 1800s; dumping hung-process stacks"
              for p in $(pgrep -f 'target/debug/deps/|target/debug/oak-worker'); do
                echo "===== thread stacks of pid $p ($(readlink /proc/$p/exe 2>/dev/null)) ====="
                sudo gdb -batch -ex 'thread apply all bt' -p "$p" || true
              done
              pkill -9 -f 'target/debug/deps/' || true
              pkill -9 -f 'target/debug/oak-worker' || true
            ) >/dev/null 2>&1 &
            WATCHDOG_PID=$!
            wait $TEST_PID
            rc=$?
            kill $WATCHDOG_PID 2>/dev/null || true
            # Reap the watchdog so no child holds the step's pipes.
            wait $WATCHDOG_PID 2>/dev/null || true
            return $rc
          }
          # Retry once (same policy as the Windows job): worker-pool
          # startup under full-suite parallelism has flaked once
          # (full_res_worker_outlives_a_dropped_project: the render
          # manager's process dispatcher failed to start while every other
          # test passed). A real regression fails both passes.
          if ! run_suite; then
            echo "first pass failed; retrying once for worker-pool flakes"
            run_suite
          fi

      # A crashing (SIGSEGV) test gives no Rust backtrace; rerun the
      # crashing test binaries under gdb to capture the native stack.
      # `--args` is required — plain `--` makes gdb treat the test args as
      # a core file. The extra probes target loader-stage crashes (the
      # copier_test SIGSEGV happens inside ld.so's dl_main): si_addr/si_code
      # pin down the fault type, the dynsym dump exposes symbols the
      # executable exports for interposition, strace shows the last loader
      # syscalls, and valgrind catches a corrupting static initializer.
      - name: Backtrace on test failure
        if: failure()
        run: |
          sudo apt-get install -y gdb strace valgrind
          for name in node_e2e_test suites_test copier_test; do
            BIN=$(ls -t target/debug/deps/$name-* | grep -v '\.d$' | head -1)
            [ -n "$BIN" ] || continue
            echo "===== $BIN ====="
            file "$BIN" || true
            echo "--- exported defined dynsyms:"
            readelf --dyn-syms -W "$BIN" 2>/dev/null | grep -v ' UND ' | tail -n +4 | head -30 || true
            echo "--- strace tail:"
            strace -f "$BIN" --list 2>&1 | tail -15 || true
            echo "--- valgrind tail:"
            valgrind -q "$BIN" --list 2>&1 | tail -25 || true
            echo "--- gdb:"
            xvfb-run -a gdb -batch \
              -ex run \
              -ex 'bt' \
              -ex 'p $_siginfo.si_code' \
              -ex 'p/x $_siginfo._sifields._sigfault.si_addr' \
              -ex 'x/6i $rip' \
              --args "$BIN" --nocapture || true
          done

      # ------------------------------------------------------------------
      # OFX plugin discovery end-to-end
      # ------------------------------------------------------------------
      # Build a minimal but real OFX plugin into a .ofx.bundle, point
      # OFX_PLUGIN_PATH at it and let the scan_probe example run the full
      # host path (directory scan -> dlopen -> setHost -> load -> describe
      # -> register). The assertion is the plugin's registration line; CI
      # machines have no system-wide OFX plugins, so the fixture is the
      # only discovery.
      - name: Build OFX fixture plugin
        run: crates/oak-plugin/tests/fixtures/build_fixture.sh .cache/ofx-fixture

      - name: Probe OFX plugin discovery
        run: |
          OFX_PLUGIN_PATH="$PWD/.cache/ofx-fixture" \
            cargo run --locked -p oak-plugin --example scan_probe > probe.log 2>&1
          grep -q 'type_id=rs.oak.CiTestPlugin' probe.log
          # A project carrying a plugin node must survive save/load (the
          # serializer resolves plugin types via the dynamic factory).
          OAK_OFX_FIXTURE_DIR="$PWD/.cache/ofx-fixture" \
            cargo test --locked -p oak-plugin --test ofx_roundtrip

  windows:
    name: Build & test (Windows)
    runs-on: warp-windows-latest-x64-16x
    env:
      # The build needs the UCRT64 environment (mingw-w64-ucrt-x86_64-*
      # toolchain), not the base MSYS one.
      MSYSTEM: UCRT64
    defaults:
      run:
        shell: msys2 {0}
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          # gpui/ is a git submodule; its crates are workspace members of
          # their own repo and build as path dependencies of oakapp.
          submodules: true

      - name: Setup MSYS2
        uses: msys2/setup-msys2@v2
        with:
          msystem: UCRT64
          update: true
          # MSYS2's own Rust targets x86_64-pc-windows-gnu by default —
          # the Windows build is GNU-target (the MSVC linker rejects the
          # Unix-style link args the build scripts emit).
          install: >-
            git
            mingw-w64-ucrt-x86_64-gcc
            mingw-w64-ucrt-x86_64-rust

      # ------------------------------------------------------------------
      # System dependencies
      # ------------------------------------------------------------------
      - name: Install system dependencies
        run: |
          bash ./tooling/install-deps.sh
          pacman -S --needed --noconfirm \
            mingw-w64-ucrt-x86_64-cmake \
            mingw-w64-ucrt-x86_64-opencolorio \
            mingw-w64-ucrt-x86_64-ffnvcodec-headers \
            mingw-w64-ucrt-x86_64-pkgconf \
            mingw-w64-ucrt-x86_64-clang mingw-w64-ucrt-x86_64-clang-libs git

      # ------------------------------------------------------------------
      # Build environment
      # ------------------------------------------------------------------
      # Windows uses the MSYS2 OpenColorIO package (the exact 2.5.2 the
      # bridge targets; the vendored source needs MSVC-only constructs).
      # Dynamic here — the CD packages the DLLs next to the binaries.
      - name: Configure build environment
        run: |
          # System OCIO (MSYS2, the exact 2.5.2 the bridge targets);
          # tooling/ocio-env.sh links it STATICALLY when the package ships
          # libOpenColorIO.a, dynamically otherwise.
          bash tooling/ocio-env.sh >> "$GITHUB_ENV"
          echo "FFMPEG_DIR=.cache/ffmpeg" >> "$GITHUB_ENV"
          # ocio-sys is the GIT fork (Mike-Solar/ocio-rs main, pinned in
          # the root manifest): its build.rs already carries the
          # fork's GNU-toolchain fix, so NO crate-unpack/glob patch step
          # applies here. The old code unpacked `ocio-sys-0.2.1.crate`
          # from the registry cache — impossible for a git dependency
          # (no .crate archive; git sources land under
          # registry/src/git/<hash>), hence the
          # "ls .../ocio-sys-0.2.1/build.rs: No such file" Windows CI
          # failure.
          echo "OCIO_RS_NO_MSVC_INCLUDES=1" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Caches
      # ------------------------------------------------------------------
      # Covers the whole target/ dir plus ~/.cargo; shared across branches
      # of the same OS.
      - name: Cache cargo artifacts
        uses: Swatinem/rust-cache@v2
        with:
          shared-key: oak-ci-windows
          cache-on-failure: true

      # The project FFmpeg (release/8.0, static, all free codecs + hwaccel)
      # is built by tooling/ffmpeg/build-ffmpeg.sh — 10-20 min on a cold
      # cache. It does not depend on the Rust toolchain, so key it on the
      # script itself and keep it out of rust-cache.
      - name: Cache project FFmpeg
        uses: actions/cache@v4
        with:
          path: .cache/ffmpeg
          key: ffmpeg-${{ runner.os }}-${{ hashFiles('tooling/ffmpeg/build-ffmpeg.sh') }}

      # ------------------------------------------------------------------
      # Project FFmpeg (script + FFMPEG_DIR; see docs/build.md)
      # ------------------------------------------------------------------
      - name: Build project FFmpeg
        run: |
          bash tooling/ffmpeg/build-ffmpeg.sh
          echo "FFMPEG_DIR=$(cygpath -m "$PWD/.cache/ffmpeg")" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Build & test
      # ------------------------------------------------------------------
      # `cargo check` (not build): the Test step links the test binaries
      # anyway, and a full build would codegen every workspace crate twice
      # (once without and once with cfg(test)). cargo check performs no
      # final link, so the mingw-w64 link-order workaround is only needed
      # in the Test step.
      - name: Build
        run: |
          cargo check --workspace --locked

      - name: Test
        run: |
          # The GitHub image injects the MSVC INCLUDE/LIB into the
          # environment; clear them in-step (they poison the MinGW
          # compiles with MSVC SDK headers).
          unset INCLUDE LIB
          # mingw-w64 >= Nov 2025 forwards _assert to __msvcrt_assert inside
          # libmingwex.a; rustc's link order puts -lmingwex last, so any
          # binary that pulls _assert.o leaves _fileno/_setmode/
          # __imp___msvcrt_assert unresolved. A trailing -lmsvcrt re-scans
          # the CRT import lib after libmingwex.
          export RUSTFLAGS="-C link-args=-lmsvcrt"
          if ! cargo test --workspace --locked; then
            # Retry once: a few gpui keystroke tests flake on Windows CI —
            # a synthetic keystroke is occasionally never delivered (the
            # undo/redo pair and a plain 's' toggle both failed once,
            # each identically to its pass state). A real regression
            # fails both passes.
            echo "first pass failed; retrying once for gpui keystroke flakes"
            cargo test --workspace --locked
          fi

  macos:
    name: Build & test (macOS)
    runs-on: warp-macos-26-arm64-6x
    steps:
      - name: Checkout
        uses: actions/checkout@v4
        with:
          # gpui/ is a git submodule; its crates are workspace members of
          # their own repo and build as path dependencies of oakapp.
          submodules: true

      - name: Install Rust (stable)
        uses: dtolnay/rust-toolchain@stable

      # ------------------------------------------------------------------
      # System dependencies
      # ------------------------------------------------------------------
      - name: Install system dependencies
        run: |
          tooling/install-deps.sh
          brew install cmake

      # ------------------------------------------------------------------
      # Build environment
      # ------------------------------------------------------------------
      - name: Configure build environment
        run: |
          # Vendored static OCIO (same as every non-Windows platform via
          # tooling/ocio-env.sh); no OCIO_INSTALL_DIR override.
          bash tooling/ocio-env.sh >> "$GITHUB_ENV"
          # Homebrew quirks: lame.pc / snappy / libopenjp2.pc live off the
          # default pkg-config search paths (see docs/build.md).
          {
            echo "CFLAGS=-I/opt/homebrew/include"
            echo "LDFLAGS=-L/opt/homebrew/lib"
            echo "PKG_CONFIG_PATH=/opt/homebrew/lib/pkgconfig/openjpeg"
          } >> "$GITHUB_ENV"
          echo "FFMPEG_DIR=.cache/ffmpeg" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Caches
      # ------------------------------------------------------------------
      # Covers the whole target/ dir plus ~/.cargo; shared across branches
      # of the same OS.
      - name: Cache cargo artifacts
        uses: Swatinem/rust-cache@v2
        with:
          shared-key: oak-ci-macos
          cache-on-failure: true

      # The project FFmpeg (release/8.0, static, all free codecs + hwaccel)
      # is built by tooling/ffmpeg/build-ffmpeg.sh — 10-20 min on a cold
      # cache. It does not depend on the Rust toolchain, so key it on the
      # script itself and keep it out of rust-cache.
      - name: Cache project FFmpeg
        uses: actions/cache@v4
        with:
          path: .cache/ffmpeg
          key: ffmpeg-${{ runner.os }}-${{ hashFiles('tooling/ffmpeg/build-ffmpeg.sh') }}

      # ------------------------------------------------------------------
      # Project FFmpeg (script + FFMPEG_DIR; see docs/build.md)
      # ------------------------------------------------------------------
      - name: Build project FFmpeg
        run: |
          tooling/ffmpeg/build-ffmpeg.sh
          echo "FFMPEG_DIR=$PWD/.cache/ffmpeg" >> "$GITHUB_ENV"

      # ------------------------------------------------------------------
      # Build & test
      # ------------------------------------------------------------------
      # `cargo check` (not build): the Test step links the test binaries
      # anyway, and a full build would codegen every workspace crate twice
      # (once without and once with cfg(test)).
      - name: Build
        run: cargo check --workspace --locked

      # The runner's GUI session doubles as the display for the gpui
      # #[gpui::test] windows; wgpu renders through Metal.
      - name: Test
        run: |
          # Retry once (same policy as the other platforms): worker-pool
          # startup under full-suite parallelism has flaked on Linux. A
          # real regression fails both passes.
          if ! cargo test --workspace --locked; then
            echo "first pass failed; retrying once for worker-pool flakes"
            cargo test --workspace --locked
          fi