ci.yml

name: CI

on:
  push:
    # `feat/expr-compiler` is the long-lived integration branch for the
    # Nx.Defn.Expr -> MLX compiler effort; milestone branches merge into
    # it via PR and it merges to main once. Run CI on pushes to it (post
    # -merge validation) and on PRs into it (every milestone). Remove
    # both entries when that branch merges to main and is deleted.
    branches: [main, feat/expr-compiler]
    tags: ['[0-9]+.[0-9]+.[0-9]+*']
  pull_request:
    branches: [main, feat/expr-compiler]
  workflow_dispatch:
    inputs:
      run_full_tests:
        description: 'Run full model tests (whisper_full, vit_full, distilbert_full, training_full, fast_kernels_full)'
        type: boolean
        default: true

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

jobs:
  precommit:
    name: precommit (${{ matrix.os }}, variant=${{ matrix.variant }})
    if: github.event_name != 'push' || github.ref_type == 'branch'
    # AOT prebuilt is built on macos-14 and is the older-macOS-compatible
    # path. JIT prebuilt is built on macos-26 and isn't portable back to
    # macos-14 (the macOS 26 libSystem grew `__fmaxf16` and other half-
    # float intrinsics that MLX's Metal stack ends up referencing). Each
    # lane therefore runs on the runner whose SDK the prebuilt targets.
    runs-on: ${{ matrix.os }}
    strategy:
      fail-fast: false
      matrix:
        include:
          - os: macos-14
            variant: aot
          - os: macos-26
            variant: jit
    env:
      MIX_ENV: test
      EMILY_MLX_VARIANT: ${{ matrix.variant }}
      # Pin the cache to the path the actions/cache step below
      # populates. Without it, mix.exs's macOS default of
      # DARWIN_USER_CACHE_DIR (`/private/var/folders/...`) misses
      # the restored cache and rebuilds MLX from scratch every run.
      # mix.exs runs `Path.expand` on this, so `~` resolves correctly.
      EMILY_CACHE: ~/Library/Caches/emily
    steps:
      - uses: actions/checkout@v6

      - name: Setup Beam
        uses: erlef/setup-beam@v1
        with:
          version-file: .tool-versions
          version-type: strict

      - name: Cache deps
        uses: actions/cache@v5
        with:
          path: deps
          key: deps-${{ runner.os }}-${{ hashFiles('mix.lock') }}
          restore-keys: deps-${{ runner.os }}-

      - name: Cache build
        uses: actions/cache@v5
        with:
          path: _build
          key: build-${{ runner.os }}-${{ matrix.variant }}-${{ hashFiles('mix.lock', 'c_src/**', 'Makefile', '.tool-versions', 'mix.exs', 'config/config.exs') }}
          restore-keys: build-${{ runner.os }}-${{ matrix.variant }}-

      # MLX prebuilt tarball (~40 MB AOT / ~5 MB JIT) + extracted install.
      # Keyed by the NIF sources and the pinned MLX version in mix.exs;
      # variants are cached separately because the install dir differs.
      - name: Cache MLX and NIF objects
        uses: actions/cache@v5
        with:
          path: ~/Library/Caches/emily
          key: mlx-${{ runner.os }}-${{ matrix.variant }}-${{ hashFiles('c_src/**', 'Makefile', 'mix.exs') }}
          restore-keys: mlx-${{ runner.os }}-${{ matrix.variant }}-

      # The Bumblebee cache holds tiny-random HuggingFace fixtures
      # downloaded by the conformance suite (~3 MB across 7 repos).
      # Cached by conformance test content so new fixtures invalidate;
      # restore-keys lets unrelated changes reuse the existing cache.
      - name: Cache Bumblebee fixtures
        uses: actions/cache@v5
        with:
          path: ~/Library/Caches/bumblebee
          key: bumblebee-${{ runner.os }}-${{ hashFiles('test/emily/conformance/**') }}
          restore-keys: bumblebee-${{ runner.os }}-

      - run: mix deps.get

      - run: mix precommit

      # Conformance tests are excluded from the default suite because
      # they require network access on a cold cache (see
      # test/test_helper.exs). In CI we always want them green — a
      # DistilBERT forward pass is the canonical integration signal
      # that no Nx op on the transformer critical path has regressed.
      - run: mix test --only conformance

  # Manual-only job: exercises the heavy `*_full` conformance and
  # training variants excluded from the default suite (see
  # test/test_helper.exs). These download full-size checkpoints
  # (hundreds of MB) and take multi-minute wall time, so they run
  # on demand via the Actions "Run workflow" button rather than on
  # every push / PR.
  full-tests:
    name: full model tests (${{ matrix.os }}, variant=${{ matrix.variant }})
    if: github.event_name == 'workflow_dispatch' && inputs.run_full_tests
    runs-on: ${{ matrix.os }}
    strategy:
      fail-fast: false
      matrix:
        include:
          - os: macos-14
            variant: aot
          - os: macos-26
            variant: jit
    env:
      MIX_ENV: test
      EMILY_MLX_VARIANT: ${{ matrix.variant }}
      # Pin the cache to the path the actions/cache step below
      # populates. Without it, mix.exs's macOS default of
      # DARWIN_USER_CACHE_DIR (`/private/var/folders/...`) misses
      # the restored cache and rebuilds MLX from scratch every run.
      # mix.exs runs `Path.expand` on this, so `~` resolves correctly.
      EMILY_CACHE: ~/Library/Caches/emily
    steps:
      - uses: actions/checkout@v6

      - name: Setup Beam
        uses: erlef/setup-beam@v1
        with:
          version-file: .tool-versions
          version-type: strict

      - name: Cache deps
        uses: actions/cache@v5
        with:
          path: deps
          key: deps-${{ runner.os }}-${{ hashFiles('mix.lock') }}
          restore-keys: deps-${{ runner.os }}-

      - name: Cache build
        uses: actions/cache@v5
        with:
          path: _build
          key: build-${{ runner.os }}-${{ matrix.variant }}-${{ hashFiles('mix.lock', 'c_src/**', 'Makefile', '.tool-versions', 'mix.exs', 'config/config.exs') }}
          restore-keys: build-${{ runner.os }}-${{ matrix.variant }}-

      - name: Cache MLX and NIF objects
        uses: actions/cache@v5
        with:
          path: ~/Library/Caches/emily
          key: mlx-${{ runner.os }}-${{ matrix.variant }}-${{ hashFiles('c_src/**', 'Makefile', 'mix.exs') }}
          restore-keys: mlx-${{ runner.os }}-${{ matrix.variant }}-

      # Full checkpoints land here (Bumblebee + scidata MNIST).
      - name: Cache Bumblebee fixtures
        uses: actions/cache@v5
        with:
          path: ~/Library/Caches/bumblebee
          key: bumblebee-full-${{ runner.os }}-${{ hashFiles('test/emily/conformance/**', 'test/emily/bumblebee/**') }}
          restore-keys: |
            bumblebee-full-${{ runner.os }}-
            bumblebee-${{ runner.os }}-

      - run: mix deps.get

      - run: mix compile

      - name: whisper_full
        run: mix test --only whisper_full

      - name: vit_full
        run: mix test --only vit_full

      - name: distilbert_full
        run: mix test --only distilbert_full

      - name: training_full
        run: mix test --only training_full

      - name: fast_kernels_full
        run: mix test --only fast_kernels_full

  # ASan CI deferred: requires OTP built with --enable-sanitizers=address
  # (macOS SIP blocks DYLD_INSERT_LIBRARIES, and late-loaded libasan
  # fails). See Makefile and RELEASE.md for details.