ci: Replace pytest-benchmark with pytest-codspeed and improve benchmarks

- Replace pytest-benchmark with pytest-codspeed in test dependencies
- Rewrite benchmark tests to use @pytest.mark.benchmark decorator
- Add full process lifecycle benchmark (init + run + destroy)
- Parameterize benchmarks across AsyncioConnector and ZMQConnector
- Replace GitHub Actions workflow with CodSpeed integration

Co-authored-by: toby-coleman <13170610+toby-coleman@users.noreply.github.com>
Agent-Logs-Url: https://github.com/plugboard-dev/plugboard/sessions/b8d70f1d-7f6f-412b-9452-e70611a349de

Author: Copilot

.github/workflows/benchmarks.yaml

@@ -1,83 +1,45 @@
 name: Benchmarks
 
 on:
+  push:
+    branches:
+      - main
   pull_request:
     types:
       - opened
       - synchronize
+  # `workflow_dispatch` allows CodSpeed to trigger backtest
+  # performance analysis in order to generate initial data.
+  workflow_dispatch:
 
 jobs:
   benchmark:
-    name: Benchmark tests
+    name: Run benchmarks
     runs-on: ubuntu-latest
     permissions:
       contents: read
-      pull-requests: write
-    strategy:
-      matrix:
-        python_version: [3.12]
+      id-token: write
     steps:
-      - name: Checkout branch
+      - name: Checkout
         uses: actions/checkout@v4
-        with:
-          path: pr
-
-      - name: Checkout main
-        uses: actions/checkout@v4
-        with:
-          ref: main
-          path: main
 
       - name: Install python
         uses: actions/setup-python@v5
         with:
-          python-version: ${{matrix.python_version}}
+          python-version: "3.12"
 
       - name: Install uv
         uses: astral-sh/setup-uv@v4
         with:
           enable-cache: true
-          cache-dependency-glob: "main/uv.lock"
-
-      - name: Setup benchmarks
-        run: |
-          echo "BASE_SHA=$(echo ${{ github.event.pull_request.base.sha }} | cut -c1-8)" >> $GITHUB_ENV
-          echo "HEAD_SHA=$(echo ${{ github.event.pull_request.head.sha }} | cut -c1-8)" >> $GITHUB_ENV
-          echo "PR_COMMENT=$(mktemp)" >> $GITHUB_ENV
-
-      - name: Run benchmarks on PR
-        working-directory: ./pr
-        run: |
-          uv sync --group test
-          uv run pytest --benchmark-only --benchmark-save=pr
-
-      - name: Run benchmarks on main
-        working-directory: ./main
-        continue-on-error: true
-        run: |
-          uv sync --group test
-          uv run pytest --benchmark-only --benchmark-save=base
+          cache-dependency-glob: "uv.lock"
 
-      - name: Compare results
-        continue-on-error: false
-        run: |
-          uvx pytest-benchmark compare **/.benchmarks/**/*.json | tee cmp_results
+      - name: Install project
+        run: uv sync --group test
 
-          echo 'Benchmark comparison for [`${{ env.BASE_SHA }}`](${{ github.event.repository.html_url }}/commit/${{ github.event.pull_request.base.sha }}) (base) vs [`${{ env.HEAD_SHA }}`](${{ github.event.repository.html_url }}/commit/${{ github.event.pull_request.head.sha }}) (PR)' >> pr_comment
-          echo '```' >> pr_comment
-          cat cmp_results >> pr_comment
-          echo '```' >> pr_comment
-          cat pr_comment > ${{ env.PR_COMMENT }}
-      
-      - name: Comment on PR
-        uses: actions/github-script@v7
+      - name: Run benchmarks
+        uses: CodSpeedHQ/action@v4
         with:
-          github-token: ${{ secrets.GITHUB_TOKEN }}
-          script: |
-            github.rest.issues.createComment({
-              issue_number: context.issue.number,
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              body: require('fs').readFileSync('${{ env.PR_COMMENT }}').toString()
-            });
+          mode: walltime
+          run: uv run pytest tests/benchmark/ --codspeed
 

pyproject.toml

@@ -80,7 +80,7 @@ test = [
     "optuna>=3.0,<5",
     "pytest>=8.3,<10",
     "pytest-asyncio>=1.0,<2",
-    "pytest-benchmark>=5.1.0",
+    "pytest-codspeed>=4.3.0",
     "pytest-cases>=3.8,<4",
     "pytest-env>=1.1,<2",
     "pytest-rerunfailures>=15.0,<17",

tests/benchmark/test_benchmarking.py

@@ -1,35 +1,61 @@
-"""Simple benchmark tests for Plugboard models."""
+"""Benchmark tests for Plugboard processes."""
 
 import asyncio
 
-from pytest_benchmark.fixture import BenchmarkFixture
+import pytest
 
-from plugboard.connector import AsyncioConnector
-from plugboard.process import LocalProcess, Process
+from plugboard.connector import AsyncioConnector, Connector, ZMQConnector
+from plugboard.process import LocalProcess
 from plugboard.schemas import ConnectorSpec
 from tests.integration.test_process_with_components_run import A, B
 
 
-def _setup_process() -> tuple[tuple[Process], dict]:
-    comp_a = A(name="comp_a", iters=1000)
+ITERS = 1000
+
+
+def _build_process(connector_cls: type[Connector]) -> LocalProcess:
+    """Build a process with the given connector class."""
+    comp_a = A(name="comp_a", iters=ITERS)
     comp_b1 = B(name="comp_b1", factor=1)
     comp_b2 = B(name="comp_b2", factor=2)
     components = [comp_a, comp_b1, comp_b2]
     connectors = [
-        AsyncioConnector(spec=ConnectorSpec(source="comp_a.out_1", target="comp_b1.in_1")),
-        AsyncioConnector(spec=ConnectorSpec(source="comp_b1.out_1", target="comp_b2.in_1")),
+        connector_cls(spec=ConnectorSpec(source="comp_a.out_1", target="comp_b1.in_1")),
+        connector_cls(spec=ConnectorSpec(source="comp_b1.out_1", target="comp_b2.in_1")),
     ]
-    process = LocalProcess(components=components, connectors=connectors)
-    # Initialise process so that this is excluded from the benchmark timing
-    asyncio.run(process.init())
-    # Return args and kwargs tuple for benchmark.pedantic
-    return (process,), {}
-
-
-def _run_process(process: Process) -> None:
-    asyncio.run(process.run())
-
-
-def test_benchmark_process_run(benchmark: BenchmarkFixture) -> None:
-    """Benchmark the running of a Plugboard Process."""
-    benchmark.pedantic(_run_process, setup=_setup_process, rounds=5)
+    return LocalProcess(components=components, connectors=connectors)
+
+
+@pytest.mark.benchmark
+@pytest.mark.parametrize(
+    "connector_cls",
+    [AsyncioConnector, ZMQConnector],
+    ids=["asyncio", "zmq"],
+)
+def test_benchmark_process_run(connector_cls: type[Connector]) -> None:
+    """Benchmark the init and run of a Plugboard Process."""
+
+    async def _run() -> None:
+        process = _build_process(connector_cls)
+        await process.init()
+        await process.run()
+
+    asyncio.run(_run())
+
+
+@pytest.mark.benchmark
+@pytest.mark.parametrize(
+    "connector_cls",
+    [AsyncioConnector, ZMQConnector],
+    ids=["asyncio", "zmq"],
+)
+def test_benchmark_process_lifecycle(connector_cls: type[Connector]) -> None:
+    """Benchmark the full lifecycle (init, run, destroy) of a Plugboard Process."""
+
+    async def _lifecycle() -> None:
+        process = _build_process(connector_cls)
+        await process.init()
+        await process.run()
+        await process.destroy()
+
+    asyncio.run(_lifecycle())