Merge main into 13.4.x

.github/actions/griffe-api-check/action.yml

@@ -0,0 +1,44 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# SPDX-License-Identifier: Apache-2.0
+
+name: griffe API check
+
+description: >-
+  Check a package's public API (as defined by `__all__`) for changes using
+  griffe.
+
+inputs:
+  package-name:
+    description: "Importable package name to check, e.g. cuda.core"
+    required: true
+  package-dir:
+    description: "Directory to search for the package sources, e.g. cuda_core"
+    required: true
+  merge-base:
+    description: >-
+      Git ref/sha to compare the current code against, typically the PR's
+      merge-base with its target branch.
+    required: true
+
+runs:
+  using: composite
+  steps:
+    - name: Install uv
+      uses: astral-sh/setup-uv@fac544c07dec837d0ccb6301d7b5580bf5edae39  # v8.2.0
+      with:
+        enable-cache: false
+
+    - name: Check API
+      shell: bash --noprofile --norc -euo pipefail {0}
+      env:
+        PACKAGE_NAME: ${{ inputs.package-name }}
+        PACKAGE_DIR: ${{ inputs.package-dir }}
+        MERGE_BASE: ${{ inputs.merge-base }}
+      run: |
+        uvx griffe check "$PACKAGE_NAME" \
+          --search "$PACKAGE_DIR" \
+          --find-stubs-packages \
+          --against "$MERGE_BASE" \
+          --format github \
+          2>&1

.github/workflows/build-wheel.yml

@@ -165,6 +165,19 @@ jobs:
         run: |
           twine check --strict cuda_pathfinder/*.whl
 
+      - name: Constrain builds to the local cuda.pathfinder wheel
+        run: |
+          pathfinder_wheels=(cuda_pathfinder/cuda_pathfinder-*.whl)
+          test "${#pathfinder_wheels[@]}" -eq 1
+          test -f "${pathfinder_wheels[0]}"
+          mkdir -p wheel-constraints
+          if [[ "${{ inputs.host-platform }}" == win* ]]; then
+            pathfinder_uri="file:///$(cygpath -am "${pathfinder_wheels[0]}")"
+          else
+            pathfinder_uri="file:///host$(realpath "${pathfinder_wheels[0]}")"
+          fi
+          printf 'cuda-pathfinder @ %s\n' "${pathfinder_uri}" | tee wheel-constraints/cuda-bindings.txt
+
       - name: Upload cuda.pathfinder build artifacts
         if: ${{ strategy.job-index == 0 && inputs.host-platform == 'linux-64' }}
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a  # v7.0.1
@@ -182,13 +195,15 @@ jobs:
           cuda-channel: ${{ inputs.cuda-channel }}
 
       - name: Build cuda.bindings wheel
-        uses: pypa/cibuildwheel@294735312765b09d24a2fbec22660ce817587d55  # v4.1.0
+        uses: pypa/cibuildwheel@4726cd35bb13f7bde50cf2761f2499ac7b3aa32c  # v4.1.1
         with:
           package-dir: ./cuda_bindings/
           output-dir: ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}
         env:
           CIBW_BUILD: ${{ env.CIBW_BUILD }}
           CIBW_ARCHS_WINDOWS: ${{ (inputs.host-platform == 'win-arm64' && 'ARM64') || 'AMD64' }}
+          CIBW_BEFORE_BUILD_LINUX: 'python -m pip install --upgrade "pip>=25.3"'
+          CIBW_BEFORE_BUILD_WINDOWS: 'python -m pip install --upgrade "pip>=25.3" delvewheel'
           # TODO: remove cpython-prerelease once 3.15 is officially supported
           # Allow CPython pre-release builds (currently 3.15 / 3.15t). This is a
           # no-op for stable Python versions because CIBW_BUILD still filters
@@ -198,6 +213,8 @@ jobs:
           CIBW_ENVIRONMENT_LINUX: >
             CUDA_PATH=/host/${{ env.CUDA_PATH }}
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
+            PIP_BUILD_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-bindings.txt
+            PIP_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-bindings.txt
             CC="/host/${{ env.SCCACHE_PATH }} cc"
             CXX="/host/${{ env.SCCACHE_PATH }} c++"
             SCCACHE_GHA_ENABLED=true
@@ -211,6 +228,8 @@ jobs:
           CIBW_ENVIRONMENT_WINDOWS: >
             CUDA_PATH="$(cygpath -w ${{ env.CUDA_PATH }})"
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
+            PIP_BUILD_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-bindings.txt)"
+            PIP_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-bindings.txt)"
           # check cache stats before leaving cibuildwheel
           CIBW_BEFORE_TEST_LINUX: >
             "/host/${{ env.SCCACHE_PATH }}" --show-adv-stats &&
@@ -243,6 +262,27 @@ jobs:
         run: |
           twine check --strict ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}/*.whl
 
+      - name: Constrain cuda.core to the local cuda.bindings wheel
+        run: |
+          pathfinder_wheels=(cuda_pathfinder/cuda_pathfinder-*.whl)
+          bindings_wheels=("${CUDA_BINDINGS_ARTIFACTS_DIR}"/cuda_bindings-"${BUILD_CUDA_MAJOR}".*.whl)
+          test "${#pathfinder_wheels[@]}" -eq 1
+          test "${#bindings_wheels[@]}" -eq 1
+          test -f "${pathfinder_wheels[0]}"
+          test -f "${bindings_wheels[0]}"
+          mkdir -p wheel-constraints
+          if [[ "${{ inputs.host-platform }}" == win* ]]; then
+            pathfinder_uri="file:///$(cygpath -am "${pathfinder_wheels[0]}")"
+            bindings_uri="file:///$(cygpath -am "${bindings_wheels[0]}")"
+          else
+            pathfinder_uri="file:///host$(realpath "${pathfinder_wheels[0]}")"
+            bindings_uri="file:///host$(realpath "${bindings_wheels[0]}")"
+          fi
+          {
+            printf 'cuda-pathfinder @ %s\n' "${pathfinder_uri}"
+            printf 'cuda-bindings @ %s\n' "${bindings_uri}"
+          } | tee wheel-constraints/cuda-core.txt
+
       - name: Upload cuda.bindings build artifacts
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a  # v7.0.1
         with:
@@ -251,13 +291,15 @@ jobs:
           if-no-files-found: error
 
       - name: Build cuda.core wheel
-        uses: pypa/cibuildwheel@294735312765b09d24a2fbec22660ce817587d55  # v4.1.0
+        uses: pypa/cibuildwheel@4726cd35bb13f7bde50cf2761f2499ac7b3aa32c  # v4.1.1
         with:
           package-dir: ./cuda_core/
           output-dir: ${{ env.CUDA_CORE_ARTIFACTS_DIR }}
         env:
           CIBW_BUILD: ${{ env.CIBW_BUILD }}
           CIBW_ARCHS_WINDOWS: ${{ (inputs.host-platform == 'win-arm64' && 'ARM64') || 'AMD64' }}
+          CIBW_BEFORE_BUILD_LINUX: 'python -m pip install --upgrade "pip>=25.3"'
+          CIBW_BEFORE_BUILD_WINDOWS: 'python -m pip install --upgrade "pip>=25.3" delvewheel'
           # TODO: remove cpython-prerelease once 3.15 is officially supported
           # Allow CPython pre-release builds (currently 3.15 / 3.15t). This is a
           # no-op for stable Python versions because CIBW_BUILD still filters
@@ -268,7 +310,8 @@ jobs:
             CUDA_PATH=/host/${{ env.CUDA_PATH }}
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
             CUDA_CORE_BUILD_MAJOR=${{ env.BUILD_CUDA_MAJOR }}
-            PIP_FIND_LINKS=/host/${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}
+            PIP_BUILD_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-core.txt
+            PIP_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-core.txt
             CC="/host/${{ env.SCCACHE_PATH }} cc"
             CXX="/host/${{ env.SCCACHE_PATH }} c++"
             SCCACHE_GHA_ENABLED=true
@@ -283,7 +326,8 @@ jobs:
             CUDA_PATH="$(cygpath -w ${{ env.CUDA_PATH }})"
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
             CUDA_CORE_BUILD_MAJOR=${{ env.BUILD_CUDA_MAJOR }}
-            PIP_FIND_LINKS="$(cygpath -w ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }})"
+            PIP_BUILD_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-core.txt)"
+            PIP_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-core.txt)"
           # check cache stats before leaving cibuildwheel
           CIBW_BEFORE_TEST_LINUX: >
             "/host${{ env.SCCACHE_PATH }}" --show-adv-stats &&
@@ -366,14 +410,17 @@ jobs:
         id: setup-python2
         uses: actions/setup-python@ece7cb06caefa5fff74198d8649806c4678c61a1  # v6.3.0
         with:
-          # TODO: Pin beta.2 precisely until cibuildwheel catches up (b4 broke ABI for Cython);
-          # this precise pin requires the explicit `freethreaded`.
-          # When 3.15 is officially supported we can also remove the `allow-prereleases` override.
-          python-version: ${{ startsWith(matrix.python-version, '3.15') && '3.15.0-beta.2' || matrix.python-version }}
-          freethreaded: ${{ endsWith(matrix.python-version, 't') }}
+          python-version: ${{ matrix.python-version }}
           architecture: ${{ ((inputs.host-platform == 'linux-aarch64' || inputs.host-platform == 'win-arm64') && 'arm64') || 'x64' }}
+          # TODO: remove allow-prereleases once 3.15 is officially supported
           allow-prereleases: ${{ startsWith(matrix.python-version, '3.15') }}
 
+      - name: Enable Scientific Python Nightly Wheels for Python 3.15
+        if: ${{ !inputs.single-cuda-major && startsWith(matrix.python-version, '3.15') }}
+        run: |
+          echo "PIP_EXTRA_INDEX_URL=https://pypi.anaconda.org/scientific-python-nightly-wheels/simple" >> "$GITHUB_ENV"
+          echo "PIP_ONLY_BINARY=numpy" >> "$GITHUB_ENV"
+
       - name: verify free-threaded build
         if: ${{ !inputs.single-cuda-major && endsWith(matrix.python-version, 't') }}
         run: python -c 'import sys; assert not sys._is_gil_enabled()'
@@ -401,65 +448,6 @@ jobs:
             mv "/c/Program Files/Git/usr/bin/link.exe" "/c/Program Files/Git/usr/bin/link.exe.bak"
           fi
 
-      # TODO: remove the numpy pre-build steps once 3.15 is officially supported
-      # (numpy will publish pre-built 3.15 wheels at that point)
-      - name: Download and patch numpy sdist (pre-release Python)
-        if: ${{ !inputs.single-cuda-major && startsWith(matrix.python-version, '3.15') }}
-        run: |
-          pip download --no-binary numpy --no-deps "numpy>=1.21.1" -d numpy-sdist/
-          cd numpy-sdist && tar xf numpy-*.tar.gz && rm numpy-*.tar.gz
-          # WAR: numpy 2.4.x ships [tool.cibuildwheel] config that is
-          # incompatible with cibuildwheel v4.0 (cpython-freethreading enable
-          # group, OpenBLAS before-build scripts, etc.). Strip the cibuildwheel
-          # sections but preserve [tool.meson-python] (vendored meson path).
-          python -c "
-          import glob
-          for f in glob.glob('numpy-*/pyproject.toml'):
-              lines, skip = open(f).readlines(), False
-              out = []
-              for line in lines:
-                  hdr = line.strip()
-                  if hdr.startswith('[tool.cibuildwheel') or hdr.startswith('[[tool.cibuildwheel'):
-                      skip = True
-                      continue
-                  if skip and hdr.startswith('[') and 'cibuildwheel' not in hdr:
-                      skip = False
-                  if not skip:
-                      out.append(line)
-              open(f, 'w').writelines(out)
-          "
-          echo "NUMPY_SRC_DIR=$(pwd)/$(ls -d numpy-*/)" >> $GITHUB_ENV
-
-      - name: Build numpy wheel (pre-release Python)
-        if: ${{ !inputs.single-cuda-major && startsWith(matrix.python-version, '3.15') }}
-        uses: pypa/cibuildwheel@294735312765b09d24a2fbec22660ce817587d55  # v4.1.0
-        env:
-          CIBW_BUILD: ${{ env.CIBW_BUILD }}
-          CIBW_SKIP: "*-musllinux* *-win32"
-          CIBW_ARCHS_LINUX: "native"
-          CIBW_ARCHS_WINDOWS: ${{ (inputs.host-platform == 'win-arm64' && 'ARM64') || 'AMD64' }}
-          CIBW_BUILD_VERBOSITY: 1
-          CIBW_CONFIG_SETTINGS: "setup-args=-Dallow-noblas=true"
-          CIBW_CONFIG_SETTINGS_WINDOWS: "setup-args=--vsenv setup-args=-Dallow-noblas=true"
-          CIBW_BEFORE_BUILD_WINDOWS: "pip install delvewheel"
-          CIBW_REPAIR_WHEEL_COMMAND_WINDOWS: "delvewheel repair -w {dest_dir} {wheel}"
-          CIBW_ENABLE: "cpython-prerelease"
-        with:
-          package-dir: ${{ env.NUMPY_SRC_DIR }}
-          output-dir: numpy-wheel/
-
-      - name: Upload numpy wheel
-        if: ${{ !inputs.single-cuda-major && startsWith(matrix.python-version, '3.15') }}
-        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a  # v7.0.1
-        with:
-          name: numpy-python${{ env.PYTHON_VERSION_FORMATTED }}-${{ inputs.host-platform }}
-          path: numpy-wheel/*.whl
-          if-no-files-found: error
-
-      - name: Install numpy wheel
-        if: ${{ !inputs.single-cuda-major && startsWith(matrix.python-version, '3.15') }}
-        run: pip install numpy-wheel/*.whl
-
       - name: Build cuda.bindings Cython tests
         if: ${{ !inputs.single-cuda-major }}
         run: |
@@ -539,23 +527,47 @@ jobs:
           OLD_BRANCH=$(yq '.backport_branch' ci/versions.yml)
           OLD_BASENAME="cuda-bindings-python${PYTHON_VERSION_FORMATTED}-cuda*-${{ inputs.host-platform }}*"
           LATEST_PRIOR_RUN_ID=$(./ci/tools/lookup-run-id --branch "${OLD_BRANCH}" NVIDIA/cuda-python "CI")
+          PREV_BINDINGS_DIR="cuda_bindings/dist-prev"
 
           gh run download $LATEST_PRIOR_RUN_ID -p ${OLD_BASENAME} -R NVIDIA/cuda-python
           rm -rf ${OLD_BASENAME}-tests  # exclude cython test artifacts
           ls -al $OLD_BASENAME
-          mkdir -p "${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}"
-          mv $OLD_BASENAME/*.whl "${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}"
+          mkdir -p "${PREV_BINDINGS_DIR}"
+          mv $OLD_BASENAME/*.whl "${PREV_BINDINGS_DIR}"
           rmdir $OLD_BASENAME
 
+      - name: Constrain previous cuda.core to the downloaded cuda.bindings wheel
+        run: |
+          pathfinder_wheels=(cuda_pathfinder/cuda_pathfinder-*.whl)
+          bindings_wheels=(cuda_bindings/dist-prev/cuda_bindings-"${BUILD_PREV_CUDA_MAJOR}".*.whl)
+          test "${#pathfinder_wheels[@]}" -eq 1
+          test "${#bindings_wheels[@]}" -eq 1
+          test -f "${pathfinder_wheels[0]}"
+          test -f "${bindings_wheels[0]}"
+          mkdir -p wheel-constraints
+          if [[ "${{ inputs.host-platform }}" == win* ]]; then
+            pathfinder_uri="file:///$(cygpath -am "${pathfinder_wheels[0]}")"
+            bindings_uri="file:///$(cygpath -am "${bindings_wheels[0]}")"
+          else
+            pathfinder_uri="file:///host$(realpath "${pathfinder_wheels[0]}")"
+            bindings_uri="file:///host$(realpath "${bindings_wheels[0]}")"
+          fi
+          {
+            printf 'cuda-pathfinder @ %s\n' "${pathfinder_uri}"
+            printf 'cuda-bindings @ %s\n' "${bindings_uri}"
+          } | tee wheel-constraints/cuda-core-prev.txt
+
       - name: Build cuda.core wheel
         if: ${{ !inputs.single-cuda-major }}
-        uses: pypa/cibuildwheel@294735312765b09d24a2fbec22660ce817587d55  # v4.1.0
+        uses: pypa/cibuildwheel@4726cd35bb13f7bde50cf2761f2499ac7b3aa32c  # v4.1.1
         with:
           package-dir: ./cuda_core/
           output-dir: ${{ env.CUDA_CORE_ARTIFACTS_DIR }}
         env:
           CIBW_BUILD: ${{ env.CIBW_BUILD }}
           CIBW_ARCHS_WINDOWS: ${{ (inputs.host-platform == 'win-arm64' && 'ARM64') || 'AMD64' }}
+          CIBW_BEFORE_BUILD_LINUX: 'python -m pip install --upgrade "pip>=25.3"'
+          CIBW_BEFORE_BUILD_WINDOWS: 'python -m pip install --upgrade "pip>=25.3" delvewheel'
           # TODO: remove cpython-prerelease once 3.15 is officially supported
           # Allow CPython pre-release builds (currently 3.15 / 3.15t). This is a
           # no-op for stable Python versions because CIBW_BUILD still filters
@@ -566,7 +578,8 @@ jobs:
             CUDA_PATH=/host/${{ env.CUDA_PATH }}
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
             CUDA_CORE_BUILD_MAJOR=${{ env.BUILD_PREV_CUDA_MAJOR }}
-            PIP_FIND_LINKS=/host/${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }}
+            PIP_BUILD_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-core-prev.txt
+            PIP_CONSTRAINT=/host/${{ github.workspace }}/wheel-constraints/cuda-core-prev.txt
             CC="/host/${{ env.SCCACHE_PATH }} cc"
             CXX="/host/${{ env.SCCACHE_PATH }} c++"
             SCCACHE_GHA_ENABLED=true
@@ -581,7 +594,8 @@ jobs:
             CUDA_PATH="$(cygpath -w ${{ env.CUDA_PATH }})"
             CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
             CUDA_CORE_BUILD_MAJOR=${{ env.BUILD_PREV_CUDA_MAJOR }}
-            PIP_FIND_LINKS="$(cygpath -w ${{ env.CUDA_BINDINGS_ARTIFACTS_DIR }})"
+            PIP_BUILD_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-core-prev.txt)"
+            PIP_CONSTRAINT="$(cygpath -w ./wheel-constraints/cuda-core-prev.txt)"
           # check cache stats before leaving cibuildwheel
           CIBW_BEFORE_TEST_LINUX: >
             "/host${{ env.SCCACHE_PATH }}" --show-adv-stats &&