eclipse-sailing-analytics/sailing-analytics
1
0
Fork 0
mirror of https://github.com/eclipse-sailing-analytics/sailing-analytics.git synced 2026-09-16 02:38:44 +00:00
Code Activity

bug53: don't post build-gate on a cancelled trusted build

Browse Source 
A cancelled build is indeterminate — the tests might have passed had the
run not been interrupted — so posting build-gate=failure would be
misleading. Restrict the failure post to build.result==failure (a real
tested failure); a cancelled trusted build now posts NOTHING, leaving the
build-gate context absent so the PR stays blocked at "Expected" with no
red X. Re-run to obtain a real verdict.

Assisted-By: Claude Opus 4.8
This commit is contained in:
Axel Uhl
2026-08-29 00:24:27 +02:00
parent 137ffc0df6
commit 08b7205d6d
1 changed files with 11 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.github/workflows/release.yml
+11 -9
View File
@@ -329,14 +329,16 @@ jobs:
# auto check-run carries a context the ruleset ignores. Its single step POSTs a
# commit status context=build-gate ONLY when: the change was trivial (build
# skipped, no relevant changes) OR a full build ran WITH tests and succeeded
# -> state=success; OR a full build ran WITH tests and FAILED (build.result not
# success/skipped, skip_tests==false) -> state=failure (red X, so a committer's
# own broken build is loud). In every OTHER case — a compile-only run, whether
# it succeeded OR failed (fork PR, *-reviewed-for-compile, workflow_dispatch
# skip_tests) — it posts NOTHING, so no build-gate context exists for the head
# SHA and the PR stays blocked at "Expected" with no red X. (A compile-only run
# ran no tests, so there is nothing meaningful to fail on; it simply hasn't met
# the gate.) Merge qualification for fork code comes only from a maintainer
# -> state=success; OR a full build ran WITH tests and FAILED
# (build.result==failure, skip_tests==false) -> state=failure (red X, so a
# committer's own broken build is loud). In every OTHER case it posts NOTHING,
# so no build-gate context exists for the head SHA and the PR stays blocked at
# "Expected" with no red X. That "nothing" covers: any compile-only run,
# succeeded OR failed (fork PR, *-reviewed-for-compile, workflow_dispatch
# skip_tests) — no tests ran, so there is nothing meaningful to fail on; AND a
# CANCELLED trusted build — cancellation is indeterminate (tests might have
# passed had it finished), so "failure" would be misleading; re-run to get a
# real verdict. Merge qualification for fork code comes only from a maintainer
# pushing a *-reviewed-for-build tag (full trusted build with tests), which
# posts success/failure. compile-gate (advisory) still shows compile pass/fail.
#
@@ -365,7 +367,7 @@ jobs:
STATE=success; REASON="Full build with tests passed."
elif [[ "$BUILD_RESULT" == "skipped" && "$SHOULD_RUN" == "false" ]]; then
STATE=success; REASON="No relevant changes — nothing to build."
elif [[ "$BUILD_RESULT" != "success" && "$BUILD_RESULT" != "skipped" && "$SKIP_TESTS" != "true" ]]; then
elif [[ "$BUILD_RESULT" == "failure" && "$SKIP_TESTS" != "true" ]]; then
STATE=failure; REASON="Full build with tests failed."
fi
if [[ -n "$STATE" ]]; then