Run CI checks only for affected files

.github/scripts/ci_changes.py

@@ -6,7 +6,17 @@
 import tomllib
 from pathlib import Path
 
-GROUPS = ("cpp", "python_quality", "python_build", "shell", "wheels")
+GROUPS = (
+    "cpp",
+    "cpp_format",
+    "cpp_tidy",
+    "python_quality",
+    "python_build",
+    "shell",
+    "wheels",
+)
+
+CPP_SUFFIXES = (".c", ".cc", ".cpp", ".h", ".hpp")
 
 
 def classify(paths):
@@ -23,19 +33,24 @@ def classify(paths):
             ".editorconfig",
         ):
             return set(GROUPS)
-        if path in ("CMakeLists.txt", "pyproject.toml") or path.startswith("cmake/"):
-            selected.update(("cpp", "python_quality", "python_build", "wheels"))
+        if path == "pyproject.toml":
+            selected.update(("python_quality", "python_build", "wheels"))
+        elif path == "CMakeLists.txt" or path.startswith("cmake/"):
+            selected.update(("cpp", "cpp_tidy", "python_build", "wheels"))
         elif path.startswith(("include/", "src/")):
-            selected.update(("cpp", "python_build", "wheels"))
+            selected.update(("cpp", "cpp_format", "cpp_tidy", "python_build", "wheels"))
         elif path.startswith("python_bindings/"):
-            selected.update(("python_quality", "python_build", "wheels"))
-            if not path.endswith(".py"):
-                selected.add("cpp")
+            selected.update(("python_build", "wheels"))
+            if path.endswith(".py"):
+                selected.add("python_quality")
+            elif path.endswith(CPP_SUFFIXES):
+                selected.update(("cpp_format", "cpp_tidy"))
         elif (
             path.startswith(
                 (
                     "tests/unit/",
                     "tests/integration/",
+                    "tests/dispatch/",
                     "tests/common/",
                     "tests/consumer/",
                     "sample/cpp/",
@@ -44,24 +59,34 @@ def classify(paths):
             or path == "tests/CMakeLists.txt"
         ):
             selected.add("cpp")
-        elif path.startswith(("tests/python/", "sample/python/")):
+            if path.endswith(CPP_SUFFIXES):
+                selected.add("cpp_format")
+        elif path.startswith("tests/python/"):
+            selected.update(("python_build", "wheels"))
+            if path.endswith(".py"):
+                selected.add("python_quality")
+        elif path.startswith("sample/python/"):
             selected.update(("python_quality", "python_build", "wheels"))
         elif path.startswith(("python/", ".github/scripts/")):
             selected.add("python_quality")
         elif path == ".github/workflows/test.yaml":
-            selected.update(("cpp", "shell"))
+            selected.update(("cpp", "cpp_format", "cpp_tidy"))
         elif path == ".github/workflows/python.yml":
             selected.update(("python_quality", "python_build"))
         elif path == ".github/workflows/release.yml":
             selected.update(("python_quality", "wheels"))
-        elif path in (".clang-format", ".clang-tidy"):
-            selected.add("cpp")
+        elif path == ".clang-format":
+            selected.add("cpp_format")
+        elif path == ".clang-tidy":
+            selected.add("cpp_tidy")
         elif path.startswith("scripts/") and path.endswith(".sh"):
             selected.add("shell")
             if path == "scripts/check-python.sh":
                 selected.add("python_quality")
-            else:
-                selected.add("cpp")
+            elif path == "scripts/check-tidy.sh":
+                selected.add("cpp_tidy")
+            elif path != "scripts/check-includes.sh":
+                selected.add("cpp_format")
         else:
             # Unknown inputs may affect builds; never silently skip their checks.
             return set(GROUPS)

.github/scripts/test_ci_changes.py

@@ -26,22 +26,26 @@ def test_docs_only_skip_builds(self):
     def test_shared_code_runs_cpp_and_python(self):
         for path in ("include/rabitqlib/index/ivf/ivf.hpp", "src/simd/dispatch.cpp"):
             with self.subTest(path=path):
-                self.assertEqual(classify([path]), {"cpp", "python_build", "wheels"})
+                self.assertEqual(
+                    classify([path]),
+                    {"cpp", "cpp_format", "cpp_tidy", "python_build", "wheels"},
+                )
 
     def test_language_specific_changes(self):
         cases = {
-            "tests/unit/rabitqlib/index/ivf_test.cpp": {"cpp"},
-            "sample/cpp/ivf_rabitq_querying.cpp": {"cpp"},
+            "tests/unit/rabitqlib/index/ivf_test.cpp": {"cpp", "cpp_format"},
+            "tests/dispatch/fastscan_no_simd_test.cpp": {"cpp", "cpp_format"},
+            "sample/cpp/ivf_rabitq_querying.cpp": {"cpp", "cpp_format"},
             "python_bindings/ivf_bindings.cpp": {
-                "cpp",
-                "python_quality",
+                "cpp_format",
+                "cpp_tidy",
                 "python_build",
                 "wheels",
             },
             "python_bindings/__init__.py": {"python_quality", "python_build", "wheels"},
+            "python_bindings/licenses/LLVM-OpenMP.txt": {"python_build", "wheels"},
             "tests/python/test_ivf.py": {"python_quality", "python_build", "wheels"},
             "tests/python/fixtures/ivf_legacy_4bit.index": {
-                "python_quality",
                 "python_build",
                 "wheels",
             },
@@ -52,27 +56,33 @@ def test_language_specific_changes(self):
             },
             ".github/scripts/prepare_release.py": {"python_quality"},
             "scripts/check-python.sh": {"shell", "python_quality"},
-            "scripts/check-tidy.sh": {"shell", "cpp"},
+            "scripts/check-tidy.sh": {"shell", "cpp_tidy"},
+            "scripts/check-format.sh": {"shell", "cpp_format"},
+            "scripts/check-includes.sh": {"shell"},
+            ".clang-format": {"cpp_format"},
+            ".clang-tidy": {"cpp_tidy"},
+            ".github/workflows/test.yaml": {"cpp", "cpp_format", "cpp_tidy"},
         }
         for path, expected in cases.items():
             with self.subTest(path=path):
                 self.assertEqual(classify([path]), expected)
 
-    def test_build_configuration_checks_both_languages(self):
-        for path in (
-            "pyproject.toml",
-            "CMakeLists.txt",
-            "cmake/RaBitQLibConfig.cmake.in",
-        ):
-            self.assertEqual(
-                classify([path]), {"cpp", "python_quality", "python_build", "wheels"}
-            )
+    def test_build_configuration_selects_affected_checks(self):
+        self.assertEqual(
+            classify(["pyproject.toml"]),
+            {"python_quality", "python_build", "wheels"},
+        )
+        for path in ("CMakeLists.txt", "cmake/RaBitQLibConfig.cmake.in"):
+            with self.subTest(path=path):
+                self.assertEqual(
+                    classify([path]), {"cpp", "cpp_tidy", "python_build", "wheels"}
+                )
 
     def test_mixed_changes_and_renamed_source(self):
         self.assertIn("cpp", classify(["docs/new.md", "src/deleted.cpp"]))
         self.assertEqual(
             classify(["sample/python/tool.py", "tests/unit/example.cpp"]),
-            {"cpp", "python_quality", "python_build", "wheels"},
+            {"cpp", "cpp_format", "python_quality", "python_build", "wheels"},
         )
 
     def test_unknown_inputs_and_filter_edits_run_everything(self):

.github/workflows/changes.yml

@@ -5,6 +5,10 @@ on:
     outputs:
       cpp:
         value: ${{ jobs.detect.outputs.cpp }}
+      cpp_format:
+        value: ${{ jobs.detect.outputs.cpp_format }}
+      cpp_tidy:
+        value: ${{ jobs.detect.outputs.cpp_tidy }}
       python_quality:
         value: ${{ jobs.detect.outputs.python_quality }}
       python_build:
@@ -23,6 +27,8 @@ jobs:
     timeout-minutes: 5
     outputs:
       cpp: ${{ steps.paths.outputs.cpp }}
+      cpp_format: ${{ steps.paths.outputs.cpp_format }}
+      cpp_tidy: ${{ steps.paths.outputs.cpp_tidy }}
       python_quality: ${{ steps.paths.outputs.python_quality }}
       python_build: ${{ steps.paths.outputs.python_build }}
       shell: ${{ steps.paths.outputs.shell }}

.github/workflows/test.yaml

@@ -13,7 +13,7 @@ jobs:
 
   format:
     needs: changes
-    if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.cpp == 'true') }}
+    if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.cpp_format == 'true') }}
     name: C++ formatting
     runs-on: ubuntu-latest
     timeout-minutes: 10
@@ -28,7 +28,7 @@ jobs:
 
   clang-tidy:
     needs: changes
-    if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.cpp == 'true') }}
+    if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.cpp_tidy == 'true') }}
     name: C++ static analysis
     runs-on: ubuntu-latest
     timeout-minutes: 30
@@ -55,44 +55,6 @@ jobs:
       - name: Run clang-tidy
         run: ./scripts/check-tidy.sh build-tidy
 
-  includes:
-    needs: changes
-    if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.cpp == 'true') }}
-    name: Include cleaner (advisory)
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    permissions:
-      contents: read
-    steps:
-      - uses: actions/checkout@v7
-      - name: Install clang-tidy and build dependencies
-        run: |
-          sudo apt-get update -qq
-          sudo apt-get install -y clang-tidy clang libomp-dev cmake ninja-build
-      - name: Configure compilation database
-        run: >-
-          cmake -S . -B build-includes -G Ninja
-          -DCMAKE_CXX_COMPILER=clang++
-          -DRABITQ_BUILD_SAMPLES=OFF
-          -DRABITQ_BUILD_TESTS=OFF
-          -DRABITQ_BUILD_PYTHON_BINDINGS=OFF
-          -DRABITQ_ENABLE_NATIVE_OPTIMIZATION=OFF
-          -DCMAKE_BUILD_TYPE=Release
-      - name: Analyze library includes
-        id: analysis
-        continue-on-error: true
-        shell: bash
-        run: |
-          set -o pipefail
-          ./scripts/check-includes.sh build-includes 2>&1 | tee include-report.txt
-      - name: Store include report
-        if: ${{ !cancelled() && steps.analysis.outcome != 'skipped' }}
-        uses: actions/upload-artifact@v7
-        with:
-          name: include-report
-          path: include-report.txt
-          if-no-files-found: error
-
   shellcheck:
     needs: changes
     if: ${{ !cancelled() && (needs.changes.result != 'success' || needs.changes.outputs.shell == 'true') }}

DEVELOPMENT.md

@@ -90,9 +90,9 @@ whether validation was focused or complete.
 
 ### Include dependency reports
 
-The **Include cleaner (advisory)** job uses clang-tidy's `misc-include-cleaner`
-check to report missing and unused includes. CI uses `ubuntu-latest` and
-unversioned distribution packages:
+The optional include dependency check uses clang-tidy's
+`misc-include-cleaner` to report missing and unused includes. Run it locally when
+reviewing include changes. On Ubuntu, install unversioned distribution packages:
 
 ```bash
 sudo apt-get update
@@ -135,8 +135,7 @@ Vendored files are excluded. The script also ignores suggestions to include
 Eigen and hnswlib implementation headers behind their existing public headers;
 these vendor snapshots lack the export annotations needed by include-cleaner.
 `INCLUDE_JOBS` controls parallelism (default: 2).
-The script fails for findings or analyzer errors; CI keeps this step advisory
-and uploads the `include-report` artifact without modifying files.
+The script fails for findings or analyzer errors and does not modify files.
 
 Review suggestions before applying them, especially for templates and public
 forwarding headers. For a source file, automatic fixes can be applied with:

MAINTAINING.md

@@ -11,15 +11,18 @@ only the affected job groups:
 | Changes | Checks |
 | --- | --- |
 | Documentation and Markdown only | Documentation workflow for `docs/`; no C++ or Python builds |
-| C++ headers or library sources | C++ formatting, analysis, tests, sanitizers, consumer build, and Python wheels/tests |
-| C++ tests or examples | C++ checks |
-| Python bindings | Python checks and wheel tests; C++ checks for compiled binding files |
+| C++ headers or library sources | C++ formatting, clang-tidy, platform tests, sanitizers, consumer build, and Python builds/wheels |
+| C++ tests or examples | C++ formatting and platform tests; no clang-tidy |
+| Compiled Python bindings | C++ formatting and clang-tidy, plus Python checks and wheel tests |
 | Python tests | Python checks and wheel tests |
 | Python scripts or examples | Python lint |
-| CMake or package configuration | C++ and Python build/test checks |
-| Shell scripts | ShellCheck and the checks driven by those scripts |
+| CMake configuration | C++ tests and clang-tidy, plus Python builds/wheels |
+| `pyproject.toml` | Python lint, builds, and wheels; no C++ checks |
+| Shell scripts | ShellCheck and any check driven by the changed script |
 
-Shared headers still trigger broad regression suites; CI does not infer
+The optional include-cleaner report is available through
+`./scripts/check-includes.sh` and no longer runs on every C++ change. Shared
+headers still trigger broad regression suites; CI does not infer
 individual test dependencies from C++ function changes. Unknown paths run all
 checks. Pushes compare the complete pushed range; pull requests compare against
 their base. Renames and deletions are included. Relevant jobs also run if