Follow-up to #3442/PR #3452 (fixed 8 workflows gating on run tests/run analysis). A self-review pass on that PR found the identical anti-pattern in two files that weren't in #3442's own affected-files list:
.github/workflows/cypress.yml — job cypress gated solely by contains(github.event.pull_request.labels.*.name, 'run e2e tests'). Adding any unrelated label to a PR that already carries run e2e tests re-runs the whole Cypress e2e suite..github/workflows/qadeploy.yml — two jobs, both higher-risk than a wasted CI run since they fire real side effects:
web-deploy gated by contains(..., 'deploy') — re-triggers an actual FTP deploy to the PR's staging directory on any unrelated label add.beta-deploy gated by contains(..., 'beta deploy') — same, for the beta deploy target.
Fix: same pattern as #3452 — AND (github.event.action != 'labeled' || github.event.label.name == '<gating label>') onto each job's existing if:.
Not folded into #3452 because the deploy jobs in qadeploy.yml aren't a mechanical CI-only change — a guard on a real deploy trigger deserves its own review, not to ride along on a PR the original issue asked to skip review for.
Follow-up to #3442/PR #3452 (fixed 8 workflows gating on
run tests/run analysis). A self-review pass on that PR found the identical anti-pattern in two files that weren't in #3442's own affected-files list:.github/workflows/cypress.yml— jobcypressgated solely bycontains(github.event.pull_request.labels.*.name, 'run e2e tests'). Adding any unrelated label to a PR that already carriesrun e2e testsre-runs the whole Cypress e2e suite..github/workflows/qadeploy.yml— two jobs, both higher-risk than a wasted CI run since they fire real side effects:web-deploygated bycontains(..., 'deploy')— re-triggers an actual FTP deploy to the PR's staging directory on any unrelated label add.beta-deploygated bycontains(..., 'beta deploy')— same, for the beta deploy target.Fix: same pattern as #3452 — AND
(github.event.action != 'labeled' || github.event.label.name == '<gating label>')onto each job's existingif:.Not folded into #3452 because the deploy jobs in
qadeploy.ymlaren't a mechanical CI-only change — a guard on a real deploy trigger deserves its own review, not to ride along on a PR the original issue asked to skip review for.