docs: reconcile desktop and remote substrate evidence #21
Workflow file for this run
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
| name: docker-rdp-vision-ladder | |
| # Real-RDP vision-ladder qualification. | |
| # | |
| # Runs the record -> compile -> replay contract through the vision-only resolver | |
| # ladder over a genuine FreeRDP client/server round-trip (benchmark/rdp_ladder), | |
| # on a real RDP pixel surface with NO structural backend. This is the CI-viable | |
| # Linux analog of the aardwolf-over-Windows transport proof in benchmark/rdp. | |
| # | |
| # Pull requests that touch this harness or its exact core dependencies run the | |
| # qualification before merge. workflow_dispatch remains available for later | |
| # default-branch reruns. There is deliberately no schedule until a rebased | |
| # candidate produces a fresh accepted 3x healthy + 3x refusal batch, avoiding | |
| # recurring runner spend and red-main noise from an intentionally pending | |
| # qualification. The fixture image is amd64+arm64; GitHub's runner is amd64. | |
| on: | |
| pull_request: | |
| paths: | |
| - ".github/workflows/docker-rdp-vision-ladder.yml" | |
| - "benchmark/rdp_ladder/**" | |
| - "pyproject.toml" | |
| - "uv.lock" | |
| - "openadapt_flow/backends/rdp_backend.py" | |
| - "openadapt_flow/compiler/compile.py" | |
| - "openadapt_flow/deployment.py" | |
| - "openadapt_flow/ir.py" | |
| - "openadapt_flow/policy.py" | |
| - "openadapt_flow/recorder.py" | |
| - "openadapt_flow/run_gate.py" | |
| - "openadapt_flow/runtime/identity.py" | |
| - "openadapt_flow/runtime/identity_template.py" | |
| - "openadapt_flow/runtime/identity_vlm.py" | |
| - "openadapt_flow/runtime/replayer.py" | |
| - "openadapt_flow/runtime/resolver.py" | |
| - "openadapt_flow/runtime/effects/**" | |
| workflow_dispatch: {} | |
| permissions: | |
| contents: read | |
| jobs: | |
| docker-rdp-vision-ladder: | |
| runs-on: ubuntu-24.04 | |
| timeout-minutes: 25 | |
| steps: | |
| - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1 | |
| with: | |
| fetch-depth: 0 | |
| # pull_request normally checks out a synthetic merge ref. Qualifying | |
| # the exact PR head keeps the recorded candidate SHA reproducible; | |
| # workflow_dispatch falls back to its explicitly selected ref. | |
| ref: ${{ github.event.pull_request.head.sha || github.sha }} | |
| - uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0 | |
| with: | |
| python-version: "3.12" | |
| - name: Install flow with the vision + rdp stack | |
| run: | | |
| python -m pip install --upgrade pip | |
| # Core owns cv2/RapidOCR/Pillow/NumPy hashing; the rdp extra installs the | |
| # product transport dependency even though this Linux fixture injects | |
| # the swappable DockerX11RdpTransport into the same backend. | |
| pip install -e ".[rdp]" | |
| pip check | |
| - name: Build the real-RDP fixture image | |
| run: docker build -t oaflow-rdp-fixture:latest benchmark/rdp_ladder/fixture | |
| - name: Start the RDP round-trip | |
| run: | | |
| mkdir -p "${RUNNER_TEMP}/rdp-ladder-oracle" | |
| docker run -d --name oaflow-rdp-ladder --shm-size=1g \ | |
| -e RDP_FIXTURE_ORACLE_ROOT=/oracle \ | |
| -v "${RUNNER_TEMP}/rdp-ladder-oracle:/oracle" \ | |
| oaflow-rdp-fixture:latest | |
| sleep 22 | |
| - name: Run the vision-ladder qualification | |
| run: | | |
| python benchmark/rdp_ladder/run_rdp_ladder_qualification.py \ | |
| --container oaflow-rdp-ladder \ | |
| --oracle-root "${RUNNER_TEMP}/rdp-ladder-oracle" \ | |
| --output runs/rdp-ladder/results.json \ | |
| --candidate-commit "$(git rev-parse HEAD)" \ | |
| --base-commit "$(git merge-base HEAD origin/main)" | |
| - name: Upload evidence | |
| if: always() | |
| uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1 | |
| with: | |
| name: rdp-ladder-qualification | |
| path: runs/rdp-ladder/results.json | |
| if-no-files-found: error | |
| - name: Tear down | |
| if: always() | |
| run: docker rm -f oaflow-rdp-ladder || true |