chore: temporarily run knowledge-graph workflow on pull requests

Lets the graph be built and fetched before any release carries it, so the
self-update path can be exercised end to end. Also stamps the PR head commit
rather than the merge commit, which exists in no local clone and would make
every staleness check report an invalid revision range.

Dry-run only — drop this commit before merging.
This commit is contained in:
2026-08-14 20:54:53 +02:00
parent 51131e377e
commit 563a190ce0
3 changed files with 35 additions and 1 deletions
+8 -1
View File
@@ -27,6 +27,8 @@ on:
workflows: ["Continuous Delivery"]
types: [completed]
workflow_dispatch: {}
# TEMPORARY dry-run trigger — this commit is meant to be dropped before merge.
pull_request: {}
permissions:
contents: read
@@ -51,7 +53,12 @@ jobs:
id: gate
run: |
TAG="$(git tag --points-at HEAD | grep '^v' | head -n 1 || true)"
echo "sha=$(git rev-parse HEAD)" >> "$GITHUB_OUTPUT"
# TEMPORARY: on pull_request the checkout is the merge commit, which exists in
# no local clone - stamping it would make every staleness check report
# "Invalid revision range". Stamp the PR head, which developers do have.
SHA="${{ github.event.pull_request.head.sha }}"
[ -n "$SHA" ] || SHA="$(git rev-parse HEAD)"
echo "sha=$SHA" >> "$GITHUB_OUTPUT"
echo "tag=$TAG" >> "$GITHUB_OUTPUT"
if [ "${{ github.event_name }}" != "workflow_run" ]; then
# dry-run contexts: build, but there is no release to attach to