Repository navigation
Submit Maven dependency snapshots; drop counterproductive dependency-review retry - #213
Conversation
…cy-review retry (java-helpers#160) Co-Authored-By: Andreas Igel <andreas.igel@computacenter.com>
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
|
Dependency Review✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.Snapshot WarningsEnsure that dependencies are being submitted on PR branches. Re-running this action after a short time may resolve the issue. See the documentation for more information and troubleshooting advice. OpenSSF Scorecard
Scanned Files
|
|
Superseded by #216. The Maven dependency-submission approach here doesn't fix the reported problem — submitting on |
|
Correction: this PR was already merged (2026-07-20), so it's already on |



Summary
Fixes the confusing "No snapshots were found for the head SHA" warning that
dependency-reviewposts in PR comments. The check is not actually failing — it's green — but the warning is prominent and looks like a failure. Fixes #212 (follow-up to #160 / #182).Root cause: this Maven repo's dependency graph is populated only by GitHub's static
pom.xmlparsing. No dependency submission snapshots exist, sodependency-review-actionemits the snapshot warning while still passing via the manifest-derived graph. Theretry-on-snapshot-warnings: trueadded in #182 is counterproductive here — with no submission workflow it retries ~2 min for a snapshot that never arrives, then proceeds with the warning anyway.Changes:
New
dependency-submission.yml— on push tomain(+ manual dispatch), submits the full transitive Maven dependency snapshot to the dependency graph viaadvanced-security/maven-dependency-submission-action(SHA-pinned@v5.0.0,contents: write). This improves vulnerability coverage (transitive, not just declared deps) and givesmainreal snapshot data for reviews to compare against.dependency-review.yml— remove the counterproductiveretry-on-snapshot-warnings, and document that the head-SHA warning on fork PRs is expected/benign:Why the head-SHA warning can't be fully removed for fork PRs: submitting a snapshot for a fork PR's head commit would require building untrusted fork code with a
contents: writetoken — the exact pattern the fork-CI design deliberately avoids. So the base side (main) gets snapshots; the fork PR head warning remains but is harmless and now documented.Related