Repository navigation
Explore self-contained Java classifier JARs with a statically linked libcuopt #1818
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -64,6 +64,104 @@ jobs: | |
| date: ${{ inputs.date }} | ||
| sha: ${{ inputs.sha }} | ||
| script: ci/build_cpp.sh | ||
| java-static-build-matrix: | ||
| permissions: | ||
| contents: read | ||
| uses: rapidsai/shared-workflows/.github/workflows/compute-matrix.yaml@main | ||
| with: | ||
| build_type: ${{ inputs.build_type || 'branch' }} | ||
| matrix_name: conda-cpp-build | ||
| matrix_filter: 'map(. + {CUDA_MAJOR: (.CUDA_VER | split(".") | .[0])}) | unique_by([.ARCH, .CUDA_MAJOR])' | ||
|
|
||
| java-static-build: | ||
| needs: [build-details, java-static-build-matrix] | ||
| permissions: | ||
| actions: read | ||
| contents: read | ||
| id-token: write | ||
| packages: read | ||
| pull-requests: read | ||
| secrets: inherit # zizmor: ignore[secrets-inherit] | ||
| uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@main | ||
| strategy: | ||
| fail-fast: false | ||
| matrix: ${{ fromJSON(needs.java-static-build-matrix.outputs.matrix) }} | ||
| with: | ||
| build_type: ${{ inputs.build_type || 'branch' }} | ||
| branch: ${{ inputs.branch }} | ||
| date: ${{ inputs.date }} | ||
| sha: ${{ inputs.sha }} | ||
| node_type: "cpu16" | ||
| arch: ${{ matrix.ARCH }} | ||
| # Must be CUDA-version-pinned, not the bare "26.10-latest" tag: custom-job.yaml has no | ||
| # per-matrix env passthrough, so an untagged image would leave RAPIDS_CUDA_VERSION at | ||
| # that image's single baked-in default for every matrix entry. | ||
| # ci/build_java_static.sh reads it to pick the conda CUDA toolkit, so every "cuN" classifier | ||
| # JAR would actually be built against the same CUDA version and just mislabeled. Matches | ||
| # the pattern conda-cpp-build.yaml itself uses for this image family. | ||
| container_image: "rapidsai/ci-conda:26.10-cuda${{ matrix.CUDA_VER }}-${{ matrix.LINUX_VER }}-py${{ matrix.PY_VER }}" | ||
|
There was a problem hiding this comment. I think we should be using the wheel/Rocky 8 image here in order to avoid issues like this: NVIDIA/cudf#23753
Collaborator
Author
There was a problem hiding this comment. Done — switched to |
||
| script: "ci/build_java_static.sh" | ||
| artifact-name: "cuopt_java_${{ matrix.ARCH }}_cu${{ matrix.CUDA_MAJOR }}" | ||
| file_to_upload: "java/cuopt/classifier-jars/" | ||
|
|
||
| # Runs the full Java suite against each classifier JAR on a GPU, with no libcuopt installed, | ||
| # so a JAR that loads but computes wrong answers fails here rather than at a user. | ||
| java-static-test: | ||
|
There was a problem hiding this comment. I believe the general pattern is to have the (nightly) test job be defined in the test.yaml file.
Collaborator
Author
There was a problem hiding this comment. Done — java-static-test now lives in test.yaml, dispatched from build.yaml's |
||
| needs: [java-static-build, java-static-build-matrix] | ||
| permissions: | ||
| actions: read | ||
| contents: read | ||
| id-token: write | ||
| packages: read | ||
| pull-requests: read | ||
| secrets: inherit # zizmor: ignore[secrets-inherit] | ||
| uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@main | ||
| strategy: | ||
| fail-fast: false | ||
| matrix: ${{ fromJSON(needs.java-static-build-matrix.outputs.matrix) }} | ||
| with: | ||
| build_type: ${{ inputs.build_type || 'branch' }} | ||
| branch: ${{ inputs.branch }} | ||
| date: ${{ inputs.date }} | ||
| sha: ${{ inputs.sha }} | ||
| node_type: "gpu-l4-latest-1" | ||
| arch: ${{ matrix.ARCH }} | ||
| # A plain CUDA-devel + dnf environment rather than rapidsai/ci-conda: see the matching | ||
| # comment in pr.yaml. | ||
| container_image: "rapidsai/ci-wheel:26.10-cuda${{ matrix.CUDA_VER }}-${{ matrix.LINUX_VER }}-py${{ matrix.PY_VER }}" | ||
| script: "ci/test_java_static.sh" | ||
|
|
||
| # Combines every classifier into one Maven-repository-layout artifact, which is the form a | ||
| # publishing workflow consumes. See rapidsai/build-infra#379. | ||
| java-static-gather: | ||
|
There was a problem hiding this comment. The artifact from the last PR job run: https://github.com/NVIDIA/cuopt/actions/runs/35367985039/job/105683600034 has the following directory layout:
Collaborator
Author
There was a problem hiding this comment. Two separate things here:
There was a problem hiding this comment. Regarding 1 above, the path is valid for
Collaborator
Author
There was a problem hiding this comment. Think we might have multiple packages under cuopt namespace, so I think other option is better. Is there a particular reason for RAPIDS to adhere to that standard ? |
||
| needs: [java-static-build] | ||
| runs-on: linux-amd64-cpu4 | ||
| permissions: | ||
| contents: read | ||
| steps: | ||
| - name: Checkout code repo | ||
| uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1 | ||
| with: | ||
| ref: ${{ inputs.sha }} | ||
| persist-credentials: false | ||
| - name: Download per-classifier JAR artifacts | ||
| uses: actions/download-artifact@d3f86a106a0bac45b974a628896c90dbdf5c8093 # v4.3.0 | ||
| with: | ||
| pattern: cuopt_java_* | ||
| path: ${{ runner.temp }}/jars | ||
| merge-multiple: true | ||
| - name: Assemble Maven repository layout | ||
| run: | | ||
| ./java/cuopt/ci/assemble_maven_repo.sh \ | ||
| --jars-dir "${RUNNER_TEMP}/jars" \ | ||
| --output-dir "${RUNNER_TEMP}/maven-repo" | ||
| - name: Upload combined Maven repository artifact | ||
| uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2 | ||
| with: | ||
| name: cuopt_java_maven_repo | ||
| path: ${{ runner.temp }}/maven-repo | ||
| if-no-files-found: error | ||
|
|
||
| java-build: | ||
| needs: cpp-build | ||
| permissions: | ||
|
|
||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,70 @@ | ||
| #!/bin/bash | ||
|
|
||
| # SPDX-FileCopyrightText: Copyright (c) 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved. | ||
| # SPDX-License-Identifier: Apache-2.0 | ||
|
|
||
| # Builds a self-contained Java classifier JAR and checks that it is actually self-contained. | ||
| # | ||
| # Unlike ci/build_java.sh, which installs a prebuilt libcuopt and links it as a shared library, | ||
| # this compiles libcuopt from source as a static archive and embeds it, so the JAR is the only | ||
| # thing a consumer installs. See #1817. | ||
|
|
||
| set -euo pipefail | ||
|
|
||
| if [[ -e /opt/conda/etc/profile.d/conda.sh ]]; then | ||
| . /opt/conda/etc/profile.d/conda.sh | ||
| fi | ||
|
|
||
| rapids-logger "Configuring conda strict channel priority" | ||
| conda config --set channel_priority strict | ||
|
|
||
| rapids-logger "Generating Java static build dependencies" | ||
| ENV_YAML_DIR=$(mktemp -d) | ||
| rapids-dependency-file-generator \ | ||
| --output conda \ | ||
| --file-key java_static \ | ||
| --matrix "cuda=${RAPIDS_CUDA_VERSION%.*};arch=$(arch)" | tee "${ENV_YAML_DIR}/env.yaml" | ||
|
|
||
| rapids-mamba-retry env create --yes -f "${ENV_YAML_DIR}/env.yaml" -n java_static | ||
|
|
||
| # Temporarily allow unbound variables for conda activation. | ||
| set +u | ||
| conda activate java_static | ||
| set -u | ||
|
|
||
| rapids-print-env | ||
|
|
||
| export CUOPT_PREFIX="${CONDA_PREFIX}" | ||
| STATIC_BUILD_DIR="${PWD}/cpp/build-static" | ||
| JNI_BUILD_DIR="${PWD}/java/cuopt/build/native-static" | ||
| JAR_OUTPUT_DIR="${PWD}/java/cuopt/classifier-jars" | ||
|
|
||
| rapids-logger "Building the scoped static libcuopt" | ||
| BUILD_DIR="${STATIC_BUILD_DIR}" bash java/cuopt/ci/build_static_libcuopt.sh | ||
|
|
||
| rapids-logger "Linking libcuopt into cuopt_jni" | ||
| cmake -S java/cuopt -B "${JNI_BUILD_DIR}" -GNinja \ | ||
| -DCMAKE_BUILD_TYPE=Release \ | ||
| -DCUOPT_PREFIX="${CUOPT_PREFIX}" \ | ||
| -DCUOPT_STATIC_BUILD_DIR="${STATIC_BUILD_DIR}" \ | ||
| -DCUOPT_EXTRA_INCLUDE_DIRS="${PWD}/cpp/include;${STATIC_BUILD_DIR}/include" | ||
| cmake --build "${JNI_BUILD_DIR}" --parallel "${PARALLEL_LEVEL:-$(nproc)}" | ||
|
|
||
| rapids-logger "Packaging the classifier JAR" | ||
| bash java/cuopt/ci/build_cuopt_java_jar.sh \ | ||
| --native-lib "${JNI_BUILD_DIR}/libcuopt_jni.so" \ | ||
| --cuda-version "${RAPIDS_CUDA_VERSION}" \ | ||
| --output-dir "${JAR_OUTPUT_DIR}" | ||
|
|
||
| # The JAR looking fine on this machine proves nothing: the build environment supplies every | ||
| # dependency by construction. This resolves them the way a consumer's machine would. | ||
| rapids-logger "Verifying the JAR is self-contained" | ||
| CLASSIFIER_JAR=$(find "${JAR_OUTPUT_DIR}" -name 'cuopt-*.jar' \ | ||
| ! -name '*-sources.jar' ! -name '*-javadoc.jar' -print -quit) | ||
| bash java/cuopt/ci/verify_jar_dependencies.sh --jar "${CLASSIFIER_JAR}" | ||
|
|
||
| # The gather job combines the classifier directories from every matrix entry into one Maven | ||
| # repository layout; this job uploads its own directory as-is. | ||
| rapids-logger "Result" | ||
| du -h "${CLASSIFIER_JAR}" | sed 's/^/ /' | ||
| find "${JAR_OUTPUT_DIR}" -type f | sed "s|^${JAR_OUTPUT_DIR}/| |" | sort |
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Uh oh!
There was an error while loading. Please reload this page.