perf-bench.yml

name: Performance Benchmarks

on:
  pull_request:
    branches: [main]
    paths:
      - 'apps/core/**'
      - 'apps/prime-mcp/**'
      - 'apps/chronis/**'
      - 'Cargo.lock'

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

env:
  CARGO_TERM_COLOR: always

jobs:
  criterion:
    name: Criterion wall-clock (throughput tracker, non-blocking)
    runs-on: ubuntu-latest
    defaults:
      run:
        working-directory: apps/core

    steps:
      - name: Checkout PR
        uses: actions/checkout@v6

      - name: Install Rust toolchain
        uses: dtolnay/rust-toolchain@nightly

      - name: Cache Rust dependencies
        uses: Swatinem/rust-cache@v2
        with:
          workspaces: "apps/core -> target"
          cache-on-failure: true

      - name: Run benchmarks (PR)
        run: cargo bench --bench performance_benchmarks -- --output-format bencher | tee /tmp/pr-bench.txt

      - name: Checkout base branch
        uses: actions/checkout@v6
        with:
          ref: ${{ github.event.pull_request.base.sha }}
          clean: false
          path: base

      - name: Run benchmarks (base)
        working-directory: base/apps/core
        run: cargo bench --bench performance_benchmarks -- --output-format bencher | tee /tmp/base-bench.txt

      - name: Compare benchmarks
        uses: benchmark-action/github-action-benchmark@v1
        with:
          tool: cargo
          output-file-path: /tmp/pr-bench.txt
          external-data-json-path: /tmp/base-bench.txt
          comment-on-alert: true
          alert-threshold: "115%"
          fail-on-alert: false
          summary-always: true
          github-token: ${{ secrets.GITHUB_TOKEN }}
          comment-always: true

  iai-callgrind:
    # Deterministic instruction-count regression gate. Runs alongside the
    # criterion wall-clock job; unlike criterion's 115% advisory threshold,
    # this job fails the build on >3% instruction-count regression.
    # See ~/.claude/skills/rust-perf/SKILL.md Phase 3.
    name: iai-callgrind (${{ matrix.package }})
    runs-on: ubuntu-latest
    strategy:
      fail-fast: false
      matrix:
        include:
          - package: allsource-core
            working-directory: apps/core
          - package: allsource-prime
            working-directory: apps/prime-mcp
          - package: chronis
            working-directory: apps/chronis
    permissions:
      contents: read
      pull-requests: write
    defaults:
      run:
        working-directory: ${{ matrix.working-directory }}

    steps:
      - name: Checkout PR
        uses: actions/checkout@v6
        with:
          fetch-depth: 0

      - name: Install valgrind
        run: |
          sudo apt-get update
          sudo apt-get install -y valgrind

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

      - name: Cache Rust dependencies
        uses: Swatinem/rust-cache@v2
        with:
          workspaces: "${{ matrix.working-directory }} -> target"
          shared-key: iai-perf-${{ matrix.package }}
          cache-on-failure: true

      - name: Run iai-callgrind and compare to main baseline
        id: bench
        run: |
          set -o pipefail
          cargo bench --bench iai -- --save-baseline main | tee /tmp/iai-output.txt

      - name: Enforce 3% regression threshold
        run: |
          python3 - <<'PY'
          import pathlib, re, sys
          output = pathlib.Path("/tmp/iai-output.txt").read_text()
          pattern = re.compile(r"Instructions:.*?Change:\s*([+-]?\d+\.?\d*)%", re.DOTALL)
          worst = 0.0
          for m in pattern.finditer(output):
              pct = float(m.group(1))
              if pct > worst:
                  worst = pct
          print(f"worst instruction-count regression: {worst:.2f}% (threshold 3%)")
          if worst > 3.0:
              print(f"::error::Perf regression {worst:.2f}% exceeds 3% threshold")
              sys.exit(1)
          PY

      - name: Comment iai summary on PR
        if: always() && github.event_name == 'pull_request'
        uses: actions/github-script@v7
        with:
          script: |
            const fs = require('fs');
            const body = fs.readFileSync('/tmp/iai-output.txt', 'utf8');
            const truncated = body.length > 60000 ? body.slice(0, 60000) + '\n... (truncated)' : body;
            const marker = '<!-- iai-perf:${{ matrix.package }} -->';
            const commentBody = `${marker}\n### iai-callgrind — \`${{ matrix.package }}/iai\`\n\n\`\`\`\n${truncated}\n\`\`\``;
            const { data: comments } = await github.rest.issues.listComments({
              owner: context.repo.owner,
              repo: context.repo.repo,
              issue_number: context.issue.number,
            });
            const existing = comments.find(c => c.body && c.body.includes(marker));
            if (existing) {
              await github.rest.issues.updateComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                comment_id: existing.id,
                body: commentBody,
              });
            } else {
              await github.rest.issues.createComment({
                owner: context.repo.owner,
                repo: context.repo.repo,
                issue_number: context.issue.number,
                body: commentBody,
              });
            }