ci: Major refactor of release-workflows by ko3n1g · Pull Request #176 · NVIDIA-NeMo/Emerging-Optimizers

ko3n1g · 2026-05-07T08:48:30Z

Why

See the design discussion in NVIDIA-NeMo/FW-CI-templates#466.

What

Delete .github/workflows/build-test-publish-wheel.yml.
Rewrite .github/workflows/release.yaml as the single caller for both push and workflow_dispatch.

Test plan

PR validate-only (push (mirror), sha 66de54d, 2026-05-07T11:27:35Z, success): https://github.com/NVIDIA-NeMo/Emerging-Optimizers/actions/runs/25492996303
workflow_dispatch dry-run=true (sha 66de54d, 2026-05-07T11:28:32Z, success): https://github.com/NVIDIA-NeMo/Emerging-Optimizers/actions/runs/25493037909
Real release via workflow_dispatch dry-run=false on the next planned RC.

Rollout

Land FW-CI-templates#466.
Cut FW-CI-templates v1.0.0.
Bump the SHA pin in this PR → tag.

Adopts the FW-CI-templates v1.0.0 pattern (NVIDIA-NeMo/FW-CI-templates#466): - single release.yaml caller for both push (validate-only) and workflow_dispatch (real release / dry-run) - no PyPI wheel publish (skip-wheel-build: true) — same pattern as RL - App-only auth (drops PAT/SSH_KEY/SSH_PWD) - pre-flight gate skips heavy work on deploy-release/* + docs_only - Slack webhook resolves at env scope (public for inert; main for real) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> Signed-off-by: oliver könig <okoenig@nvidia.com>

greptile-apps · 2026-05-07T08:51:34Z

Greptile Summary

This PR refactors the CI release pipeline for Emerging-Optimizers, replacing the standalone build-test-publish-wheel.yml with a unified release.yaml that handles both push-triggered validation and workflow_dispatch-triggered releases through a single caller workflow referencing _release_library.yml@v1.1.0.

New trigger model: push to main, r**, pull-request/**, and deploy-release/* branches runs a pre-flight check and validation-only release pass; workflow_dispatch skips pre-flight and runs the full release pipeline.
release-summary job: A new terminal job queries the GitHub API with gh run view to surface any failed jobs and fail the workflow run accordingly.
Open issues from existing review threads: create-gh-release: ${{ inputs.create-gh-release || true }} always evaluates to true regardless of user input; publish-docs: true is hardcoded and ignores the publish-docs workflow input; and build-test-publish-wheel.yml was not deleted despite being listed in the PR description, leaving the old workflow active on the same branch triggers.

Confidence Score: 4/5

The workflow logic for push and workflow_dispatch paths is sound, but several open review threads flag real defects in how user-provided boolean inputs are forwarded to the reusable workflow.

The create-gh-release: ${{ inputs.create-gh-release || true }} expression always produces true, making it impossible to suppress GH release creation via workflow_dispatch. Similarly, publish-docs is hardcoded to true and cannot be overridden by callers. Both inputs are surfaced to operators in the dispatch UI but silently ignored at runtime. Additionally, build-test-publish-wheel.yml was not removed, leaving the old workflow active on the same branch triggers and risking duplicate release-pipeline executions.

.github/workflows/release.yaml — specifically the boolean input forwarding expressions and the hardcoded publish-docs value. The undeleted .github/workflows/build-test-publish-wheel.yml also warrants attention.

Important Files Changed

Filename	Overview
.github/workflows/release.yaml	Complete rewrite of the release workflow to unify push-triggered validation and workflow_dispatch-triggered releases; contains open issues around boolean input overrides, hardcoded publish-docs, and a missing file deletion.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A([Trigger]) --> B{Event type?}
    B -->|push to main or r-branches| C[pre-flight job]
    B -->|workflow_dispatch| D[pre-flight SKIPPED]

    C --> E{docs_only or deployment?}
    E -->|yes| F[release SKIPPED]
    E -->|no| G[release job\nvalidate-only=true]

    D --> H[release job\nvalidate-only=false]

    F --> I[release-summary job]
    G --> I
    H --> I

    I --> J{Failed jobs?}
    J -->|none| K([exit 0])
    J -->|found| L([exit 1])

_{Reviews (12): Last reviewed commit: "ci: pin FW-CI templates to v1.1.0" | Re-trigger Greptile}

greptile-apps · 2026-05-07T08:51:37Z

+      validate-only: ${{ github.event_name != 'workflow_dispatch' }}
+      dry-run: ${{ inputs.dry-run || false }}
+      version-bump-branch: ${{ inputs.version-bump-branch || github.ref_name }}
+      create-gh-release: ${{ inputs.create-gh-release || true }}


When workflow_dispatch fires and the user explicitly sets create-gh-release: false, the expression inputs.create-gh-release || true evaluates to false || true = true, silently overriding the user's intent and always creating a GH release. This is the only boolean input that uses a truthy fallback (|| true); the others use || false which is safe.

Suggested change

create-gh-release: ${{ inputs.create-gh-release || true }}

create-gh-release: ${{ inputs.create-gh-release != '' && inputs.create-gh-release || true }}

greptile-apps · 2026-05-07T08:51:38Z

+    if: |
+      (
+        needs.pre-flight.outputs.docs_only == 'true'
+        || needs.pre-flight.outputs.is_deployment_workflow == 'true'
+        || always()
+      )
+      && !cancelled()


always() placed inside the OR group short-circuits the entire parenthesised expression to true regardless of the docs_only and is_deployment_workflow checks — they are never evaluated. The effective condition is just !cancelled(). The dead checks add noise and may mislead future maintainers about when this job is supposed to be skipped.

Suggested change

if: |

(

needs.pre-flight.outputs.docs_only == 'true'

|| needs.pre-flight.outputs.is_deployment_workflow == 'true'

|| always()

)

&& !cancelled()

if: always() && !cancelled()

github-actions · 2026-05-07T08:56:05Z

Test Results

50 files ±0 110 suites ±0 1m 47s ⏱️ +27s
1 031 tests ±0 1 031 ✅ ±0 0 💤 ±0 0 ❌ ±0
2 313 runs ±0 2 313 ✅ ±0 0 💤 ±0 0 ❌ ±0

Results for commit fc0e75b. ± Comparison against base commit 04ee932.

♻️ This comment has been updated with latest results.

codecov · 2026-05-07T08:58:04Z

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Lets env-scoped SLACK_WEBHOOK reach the notify job in the called workflow. Signed-off-by: oliver könig <okoenig@nvidia.com>

ko3n1g · 2026-05-07T10:56:03Z

/ok to test eb723b5