diff options
| author | Yuriy Andamasov <yuriy@vyos.io> | 2026-05-11 08:53:28 +0300 |
|---|---|---|
| committer | Yuriy Andamasov <yuriy@vyos.io> | 2026-05-11 08:53:28 +0300 |
| commit | 55b5db4feb3e0d8e59e6a9580d952f462d173f1e (patch) | |
| tree | 63dc406f09f4f46caf141717e5b227986ba33887 /docs/_static/css/code-snippets.css | |
| parent | 07bc829929fc9e8873b26b1b3ecebc1020d7dfda (diff) | |
| download | vyos-documentation-55b5db4feb3e0d8e59e6a9580d952f462d173f1e.tar.gz vyos-documentation-55b5db4feb3e0d8e59e6a9580d952f462d173f1e.zip | |
ci(ai-validation): revert DB-download tag pin to latest:true
Backport of #1971 to this release branch — keeps the workflow file
byte-identical across rolling/circinus/sagitta after the wave of
#1968 / #1969 / #1959 / #1960 / #1970.
Reason (full text in the rolling-side PR description):
PR #1969's `tag: ${{ env.REVIEWER_REF }}` on the reference-DB
download would 404 — the `reviewer-v1.x.x` tag has no GitHub release
with DB assets attached. Reference DBs are published to a separate
release stream (`ref-db-<timestamp>`) by the matrixed rebuild-reference
workflow. Reverting to `latest: true` with an explicit inline comment
block explaining the release-asset topology so this is not re-broken
on a future review pass.
The bug only manifests on a PR with `has_md_changes=true` (i.e. a real
`.md`-changing PR); the recent merges that landed the regression were
all infrastructure-only and never invoked the DB step.
Companion PRs: #1971 (rolling), VyOS-Networks/vyos-docs-opus-reviewer#15
(canonical sync).
Diffstat (limited to 'docs/_static/css/code-snippets.css')
0 files changed, 0 insertions, 0 deletions