test: split slow tests, cache CI deps, add pre-push gate

.github/workflows/ci.yml

@@ -1,25 +1,34 @@
 # =============================================================================
 # .github/workflows/ci.yml  —  Continuous Integration
 # =============================================================================
-# Runs on pull requests to main. Fast feedback on tests + lint.
+# Pull requests to main: fast required gate (tests minus @pytest.mark.slow) + lint.
+# The full suite (including slow tests) runs nightly on a schedule and on demand
+# via workflow_dispatch. Dependencies are cached via setup-uv to avoid
+# re-downloading torch & friends on every run.
 # =============================================================================
 
 name: CI
 
 on:
   pull_request:
     branches: [main]
+  workflow_dispatch:
+  schedule:
+    # Nightly full suite at 03:17 UTC (off-peak; avoids the top-of-hour cron herd).
+    - cron: "17 3 * * *"
 
 permissions: read-all
 
 concurrency:
-  group: ci-${{ github.ref }}
+  group: ci-${{ github.workflow }}-${{ github.ref }}
   cancel-in-progress: true
 
 jobs:
+  # ── Fast required gate: everything except @pytest.mark.slow ────────────────
   test:
-    name: Tests
+    name: Tests (fast)
     runs-on: ubuntu-latest
+    if: github.event_name != 'schedule'
     permissions:
       contents: read
 
@@ -31,16 +40,51 @@ jobs:
           python-version: "3.13"
 
       - uses: astral-sh/setup-uv@d4b2f3b6ecc6e67c4457f6d3e41ec42d3d0fcb86 # v5.4.2
+        with:
+          enable-cache: true
+          cache-dependency-glob: |
+            uv.lock
+            pyproject.toml
 
       - name: Install dependencies
-        run: uv pip install --system -e ".[dev]"
+        run: uv sync --extra dev
+
+      - name: Run pytest (fast — excludes @pytest.mark.slow)
+        run: uv run pytest tests/ -m "not slow" -n auto --tb=short
+
+  # ── Full suite: nightly schedule + manual dispatch ─────────────────────────
+  test-full:
+    name: Tests (full suite)
+    runs-on: ubuntu-latest
+    if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
+    permissions:
+      contents: read
 
-      - name: Run pytest
+    steps:
+      - uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1
+
+      - uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5.6.0
+        with:
+          python-version: "3.13"
+
+      - uses: astral-sh/setup-uv@d4b2f3b6ecc6e67c4457f6d3e41ec42d3d0fcb86 # v5.4.2
+        with:
+          enable-cache: true
+          cache-dependency-glob: |
+            uv.lock
+            pyproject.toml
+
+      - name: Install dependencies
+        run: uv sync --extra dev
+
+      - name: Run pytest (full suite, including slow)
         run: uv run pytest tests/ -v -n auto --tb=short
 
+  # ── Lint ───────────────────────────────────────────────────────────────────
   lint:
     name: Lint
     runs-on: ubuntu-latest
+    if: github.event_name != 'schedule'
     permissions:
       contents: read
 
@@ -52,10 +96,15 @@ jobs:
           python-version: "3.13"
 
       - uses: astral-sh/setup-uv@d4b2f3b6ecc6e67c4457f6d3e41ec42d3d0fcb86 # v5.4.2
+        with:
+          enable-cache: true
+          cache-dependency-glob: |
+            uv.lock
+            pyproject.toml
 
       - name: Install dependencies
-        run: uv pip install --system -e ".[dev]"
+        run: uv sync --extra dev
 
       - name: Ruff
-        run: ruff check src/ tests/
+        run: uv run ruff check src/ tests/
         continue-on-error: true

.github/workflows/release.yml

@@ -44,13 +44,20 @@ jobs:
           python-version: "3.13"
 
       - uses: astral-sh/setup-uv@d4b2f3b6ecc6e67c4457f6d3e41ec42d3d0fcb86 # v5.4.2
+        with:
+          enable-cache: true
+          cache-dependency-glob: |
+            uv.lock
+            pyproject.toml
 
       - name: Install dependencies
-        run: uv pip install --system -e ".[dev,docs]"
+        run: uv sync --extra dev --extra docs
 
       # ── Tests ─────────────────────────────────────────────────────────
-      - name: Run tests
-        run: uv run pytest tests/ -v -n auto --tb=short
+      # Fast gate only: the full suite (incl. slow) already ran on the PR's
+      # required check and nightly. This is a quick pre-publish sanity check.
+      - name: Run tests (fast — excludes slow)
+        run: uv run pytest tests/ -m "not slow" -n auto --tb=short
 
       # ── Semantic Release ──────────────────────────────────────────────
       - uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4.4.0

.pre-commit-config.yaml

@@ -0,0 +1,42 @@
+# Pre-commit / pre-push hooks for spotoptim.
+#
+# Install once per clone (hooks live in .git/ and are never pushed):
+#   uv run pre-commit install --hook-type pre-push --hook-type pre-commit
+#
+# - on `git push`   : runs the FAST test suite and blocks the push if it fails
+# - on `git commit` : runs ruff + black --check on changed Python files
+#
+# Emergency bypass (skips all hooks): git push --no-verify
+#
+# Requires pre-commit >= 3.2.0 (stage was renamed `push` -> `pre-push`).
+minimum_pre_commit_version: "3.2.0"
+default_stages: [pre-commit]
+
+repos:
+  - repo: local
+    hooks:
+      # Core gate: block `git push` when the fast suite fails.
+      - id: pytest-fast
+        name: pytest (fast, excludes slow)
+        entry: uv run pytest -m "not slow" -n auto
+        language: system
+        pass_filenames: false
+        always_run: true
+        stages: [pre-push]
+
+      # Lightweight, non-mutating commit-time checks (project-pinned tools via uv).
+      - id: ruff
+        name: ruff check
+        entry: uv run ruff check
+        language: system
+        types_or: [python, pyi]
+        require_serial: true
+        stages: [pre-commit]
+
+      - id: black
+        name: black --check
+        entry: uv run black --check
+        language: system
+        types_or: [python, pyi]
+        require_serial: true
+        stages: [pre-commit]

README.md

@@ -238,16 +238,42 @@ See docs/examples.md for more details and additional examples.
 git clone https://github.com/sequential-parameter-optimization/spotoptim.git
 cd spotoptim
 
-# Install with uv
-uv pip install -e .
+# Install with uv (editable, with dev dependencies)
+uv sync --extra dev
 
-# Run tests
-uv run pytest tests/
+# Run the fast suite (excludes slow end-to-end tests) — what CI/PRs gate on
+uv run pytest tests/ -m "not slow" -n auto
+
+# Run the full suite (everything, including slow tests)
+uv run pytest tests/ -n auto
 
 # Build package
 uv build
 ```
 
+### Fast vs. full tests
+
+Heavy end-to-end optimization tests are tagged `@pytest.mark.slow` (registered
+centrally in `tests/conftest.py`). The fast path skips them with `-m "not slow"`;
+the full suite and the nightly CI job run everything.
+
+### Git hooks (pre-push test gate)
+
+A `.pre-commit-config.yaml` runs the **fast** test suite before every `git push`
+and **blocks the push if any test fails**. Hooks are per-clone and are *not*
+installed automatically — install them once after cloning:
+
+```bash
+uv run pre-commit install --hook-type pre-push --hook-type pre-commit
+```
+
+- on `git push` → `uv run pytest -m "not slow" -n auto`
+- on `git commit` → `ruff check` + `black --check` on changed files
+
+Emergency bypass (use sparingly): `git push --no-verify`. Note that the bare
+`pre-commit install` only wires the commit stage — the `--hook-type pre-push`
+flag is required to enable the test gate.
+
 ## Release Troubleshooting
 
 If a release fails (for example with `semantic-release` push/tag permission errors), use this checklist.

pyproject.toml

@@ -49,18 +49,21 @@ dev = [
     "ruff>=0.3.0",
     "safety>=3.0.0",
     "bandit>=1.8.0",
+    "pre-commit>=3.7.0",
 ]
 
 [project.optional-dependencies]
 dev = [
     "pytest>=7.4.0",
     "pytest-cov>=4.1.0",
     "pytest-xdist>=3.5.0",
+    "pytest-timeout>=2.3.0",
     "black>=24.1.0",
     "isort>=5.13.0",
     "ruff>=0.3.0",
     "safety>=3.0.0",
     "bandit>=1.8.0",
+    "pre-commit>=3.7.0",
 ]
 docs = [
     # Quarto API documentation generator (replaces mkdocstrings)
@@ -79,6 +82,11 @@ testpaths = ["tests"]
 python_files = ["test_*.py"]
 python_classes = ["Test*"]
 python_functions = ["test_*"]
+# Keep slow-test timings visible on every run; the registry lives in tests/conftest.py.
+addopts = ["--durations=15"]
+markers = [
+    "slow: heavy end-to-end test; deselect with -m \"not slow\". Runs in nightly/full CI.",
+]
 filterwarnings = [
     "ignore::UserWarning:matplotlib",
 ]

tests/conftest.py

@@ -0,0 +1,85 @@
+"""Shared pytest configuration for the spotoptim test suite.
+
+Slow-test registry
+------------------
+The suite contains ~1.7k tests; a few dozen end-to-end optimization runs
+dominate the wall-clock. Those are tagged ``@pytest.mark.slow`` here, in one
+central place, so that:
+
+* the fast path (PR gate, pre-push hook) runs ``pytest -m "not slow"`` and
+  skips them, and
+* the nightly / full CI job runs everything.
+
+Entries in :data:`SLOW_NODEIDS` are matched against each test's node id with
+the parametrization suffix (``[...]``) stripped. An entry may be:
+
+* a file            ``tests/test_x0_starting_point.py``
+* a class           ``tests/test_batch_eval.py::TestBatchEvalEndToEnd``
+* a single test     ``tests/test_save_load.py::TestEdgeCases::test_reproducibility_after_load``
+
+This list was derived from ``pytest --durations`` (tests taking >~8 s).
+To refresh it, run ``uv run pytest tests/ -n auto --durations=50`` and update
+the entries below. Stale entries that no longer match anything are harmless.
+New slow tests may instead be decorated directly with ``@pytest.mark.slow``.
+"""
+
+import pytest
+
+# Heaviest tests (>~8 s each), grouped by file. Class/file entries also cover
+# their faster sibling integration tests, which belong in the full suite too.
+SLOW_NODEIDS = {
+    # --- whole files (every test is a full optimization run) ---
+    "tests/test_x0_starting_point.py",
+    # --- integration test classes ---
+    "tests/test_batch_eval.py::TestBatchEvalEndToEnd",
+    "tests/test_termination_criteria.py::TestTerminationCriteria",
+    "tests/test_thread_pool_search.py::TestHybridExecutorEndToEnd",
+    "tests/test_no_gil_awareness.py::TestSimulatedNoGilPath",
+    "tests/test_no_gil_awareness.py::TestGilBuildEndToEnd",
+    "tests/test_multiobjective.py::TestMultiObjectiveOptimization",
+    "tests/test_transformations.py::TestTransformationOptimization",
+    "tests/test_reproducibility_comprehensive.py::TestSpotOptimReproducibility",
+    "tests/test_optimize_refactored_methods.py::TestOptimizeIntegration",
+    # --- individual heavy tests ---
+    "tests/test_early_stopping.py::test_max_restarts_does_not_trigger_with_improvement",
+    "tests/test_parallel_optimization.py::TestParallelOptimization::test_parallel_execution_basic",
+    "tests/test_multiobjective.py::TestMultiObjectiveEdgeCases::test_many_objectives",
+    "tests/test_initial_design_nan_handling.py::test_initial_design_with_mixed_nan_inf",
+    "tests/test_cookbook_examples.py::test_example_4_nelder_mead",
+    "tests/test_spotoptim_mlp_surrogate.py::test_mlp_surrogate_uncertainty_in_loop",
+    "tests/test_save_load.py::TestEdgeCases::test_reproducibility_after_load",
+    "tests/test_run_sequential_loop_example.py::test_run_sequential_loop_example",
+    "tests/test_factor_variables.py::TestFactorVariables::test_many_factor_levels",
+    "tests/test_spotoptim_deep.py::TestConvergenceQuality::test_sphere_2d_converges_near_origin",
+    "tests/test_spotoptim_deep.py::TestSeedReproducibility::test_same_seed_same_results",
+    "tests/test_max_iter_validation.py::TestMaxIterValidation::test_max_iter_greater_than_n_initial_works",
+    "tests/test_termination.py::TestBackwardCompatibility::test_old_behavior_without_max_time",
+    "tests/test_termination.py::TestMaxIterTermination::test_max_iter_with_custom_initial_design",
+    "tests/test_tolerance_x.py::TestToleranceXFloatVariables::test_no_duplicate_evaluations_float",
+    "tests/test_tolerance_x.py::TestToleranceXFactorVariables::test_no_duplicate_evaluations_factors",
+    "tests/test_spotoptim.py::TestSpotOptimOptimize::test_optimize_with_seed_reproducibility",
+    "tests/test_parallel_reporting.py::test_parallel_reporting",
+    "tests/test_parallel_merging.py::test_parallel_merging",
+    "tests/test_deterministic.py::TestDeterministicBehavior::test_deterministic_with_provided_initial_design",
+    "tests/test_acquisition_failure.py::TestAcquisitionFailureWithVariableTypes::test_acquisition_failure_with_mixed_variables",
+    "tests/test_spot_optim_args.py::test_kwargs_only",
+    "tests/test_tensorboard.py::TestTensorBoardIntegration::test_tensorboard_with_custom_var_names",
+    "tests/test_tensorboard_clean.py::TestTensorBoardClean::test_clean_with_optimization_run",
+}
+
+
+def _is_slow(nodeid: str) -> bool:
+    """Return True if ``nodeid`` is covered by a SLOW_NODEIDS entry."""
+    base = nodeid.split("[", 1)[0]  # drop parametrization suffix
+    return any(base == entry or base.startswith(entry + "::") for entry in SLOW_NODEIDS)
+
+
+def pytest_collection_modifyitems(config, items):
+    """Tag registered heavy tests with the ``slow`` marker during collection.
+
+    Runs before ``-m`` deselection, so ``pytest -m "not slow"`` skips them.
+    """
+    slow = pytest.mark.slow
+    for item in items:
+        if _is_slow(item.nodeid):
+            item.add_marker(slow)