Skip to content

ci: re-pin acceptance-spec to 380780e (base cache now works); convention gate waits for acceptance-spec#5 - #2249

Merged
buggtb merged 4 commits into
developmentfrom
ci/repin-acceptance-spec
Oct 12, 2026
Merged

buggtb merged 4 commits into
developmentfrom
ci/repin-acceptance-spec

Conversation

@buggtb

@buggtb buggtb commented Oct 12, 2026 •

Copy link
Copy Markdown
Contributor

What

Re-pins every call of spiculedata/acceptance-spec (run-specs.yml, base-check.yml in acceptance.yml) from 65c4575 to 380780e255019826deb3901987f4004a454a7659 (main: includes #3 waivable-paths, #4 base-cache fix). Docs updated (docs/acceptance-specs.md: pin, cache now works, why the gate is still local). That doc held the only stated pin; ci-overview / acceptance README / AGENTS.md name no SHA.

The convention gate (ci / acceptance convention) is deliberately NOT moved in this PR. It is built and proven on branch ci/convention-shared (no PR opened), blocked on a one-line upstream input: spiculedata/acceptance-spec#5 (not merged).

Why

  • Cache: at the old pin the base JAR cache saved nothing (Connections to non-default MySQL ports fails - Upgrade MySQL Connector/J #4). Now it works.
  • Gate: Building queries with dimensions with multiple hierarchies #3 lets convention.yml take Saiku's exact waiver list, and a called job renders as ci / acceptance convention if the calling job is named ci (verified: that exact check name, and it is inside the ci rollup and passes/fails it). But ci.yml also runs on push and merge_group, where the gate must not run. The only way today is if: github.event_name == 'pull_request' on the calling job, and a skipped calling job reports a bare check named after the caller job: here a skipped check named ci, the same name as the single required aggregate, measured on a push to a phase-* branch (ci | completed skipped appeared immediately beside the real ci). I cannot prove the merge queue ignores it without merging, so I did not ship it. Distributed Cache #5 adds skip-outside-pull-request, so the called job skips itself and no bare check exists; measured on a push: only ci / acceptance convention | skipped, no bare ci.

Compatibility

Enforced behaviour unchanged: the gate, its scripts and tests, acceptance/2172 are byte-identical to development. Check names unchanged (acceptance / specs alias, acceptance / acceptance specs, detect changed specs, specs on head|base, spec fails on base, ci / acceptance convention).

Semantics proven for the follow-up (done offline with the real scripts at the new pin, before the branch was split): over 2713 paths (every tracked file plus edge cases) the local waiver list and waivable-paths (9 patterns; .md was case-insensitive, so \.[mM][dD]$) classify identically, 0 differences. saiku-core/e2e/x.js and test/x.java are waivable in the old shared built-in list and NOT with Saiku's list. CLI outputs (cutoff 2181 / unset / no closing keyword / missing / waived / rejected) are identical except one cosmetic line: the closing hint now points at acceptance-spec's docs/spec-format.md. Cutoff stays vars.ACCEPTANCE_ENFORCE_FROM_ISSUE (with: enforce-from-issue: ${{ vars... }} works), same acceptance-waived label.

Verification (GitHub; throwaway PRs #2250-#2255 closed, branches deleted, never merged; they ran the moved-gate branch state, stacked on this PR)

Risk

Low: pin bump of the two workflows only. The hive watcher's acceptance / specs alias is untouched.

Not verified

  • Merge-queue behaviour with the bare skipped ci (why the move waits). The Distributed Cache #5 fix was measured on a push only, not on merge_group.
  • Fork PRs (not-run path); spec-already-green.
  • After Distributed Cache #5 merges: replace the pin in ci.yml and acceptance.yml on ci/convention-shared with the merge commit, delete the local scripts as that branch does (acceptance/2172 ported to assert the ci.yml wiring), merge.

🤖 Generated with Claude Code

…ention.yml

Re-pin to 380780e255019826deb3901987f4004a454a7659 (waivable-paths input, base cache fix).
ci / acceptance convention now calls convention.yml with Saiku's exact waiver list; the local
acceptance-check.mjs and the convention logic in acceptance-spec.mjs are removed, acceptance/2172
is ported to assert the ci.yml wiring.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Signed-off-by: Tom Barber <tom@spicule.co.uk>
@github-actions github-actions Bot added risk/high Change risk tier: high (see docs/risk-tiers.md) ci Created by Hive for agent-filed issue provenance docs agent-config and removed risk/high Change risk tier: high (see docs/risk-tiers.md) labels Oct 12, 2026
@github-actions

Copy link
Copy Markdown

Semantic model diff

No semantic model changed in this pull request, so there was nothing to diff.
saiku#1434 · reproduce locally with saiku model diff

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Signed-off-by: Tom Barber <tom@spicule.co.uk>
@github-actions github-actions Bot added the risk/high Change risk tier: high (see docs/risk-tiers.md) label Oct 12, 2026
…alling job shadows the required ci check)

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Signed-off-by: Tom Barber <tom@spicule.co.uk>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Signed-off-by: Tom Barber <tom@spicule.co.uk>
@buggtb buggtb changed the title ci: re-pin acceptance-spec and move the convention gate onto its convention.yml ci: re-pin acceptance-spec to 380780e (base cache now works); convention gate waits for acceptance-spec#5 Oct 12, 2026
@github-actions

github-actions Bot commented Oct 12, 2026 •

Copy link
Copy Markdown

CI feedback

passing · 897d33b
Generated 2026-10-12T05:29:12.994Z. Machine-readable copy: ci-feedback.json (schema v1, see docs/ci-feedback.md) - artifact on this run.

Nothing to fix: every check in this commit concluded successfully.

@github-actions

Copy link
Copy Markdown

CI feedback

failing · b90d169
Generated 2026-10-12T04:46:46.792Z. Machine-readable copy: ci-feedback.json (schema v1, see docs/ci-feedback.md) - artifact on this run.

1. acceptance / specs

Workflow: acceptance (failure)
Job log: https://github.com/spiculedata/saiku/actions/runs/38205107207/job/114666870920
Reproduce: node .github/scripts/acceptance-runner.mjs run --all --base-url <url>

First real error:

acceptance specs concluded cancelled (see the 'acceptance / acceptance specs' check)

2. ci

Workflow: ci (failure) - rollup of the checks above
Job log: https://github.com/spiculedata/saiku/actions/runs/38205106924/job/114666862277
Reproduce: gh run rerun 38205106924 --failed

First real error:

build=cancelled
ui=cancelled
ok: sheets=skipped
dist=cancelled
ok: scripts=success
ok: flake-policy=success
ok: acceptance-convention=success
demo-deploy=cancelled

Artifacts

@buggtb
buggtb enabled auto-merge October 12, 2026 04:55
@buggtb buggtb closed this Oct 12, 2026
auto-merge was automatically disabled October 12, 2026 05:23

Pull request was closed

@buggtb buggtb reopened this Oct 12, 2026
@github-actions

Copy link
Copy Markdown

PR acceptance metric

Acceptance rate: 88.5% — 440 merged / 497 decided PRs (3 still open, excluded from the denominator).

Metric Value
Acceptance rate 88.5%
Accepted / rejected / open 440 / 57 / 3
Median time to first review (merged) 0.5 h
Median time to merge (merged) 0.9 h
Median time open (decided) 1.2 h
Mean review rounds on merged PRs 1.0
Authors with at least one merged PR 6
Author Merged Rejected Acceptance
buggtb 195 10 95.1%
juanresendiz813 134 1 99.3%
app/dependabot 102 41 71.3%
app/spicule-hive 6 1 85.7%
oiacub1 2 0 100.0%
Prithvi1994 1 0 100.0%
laborima 0 3 0.0%
spicule-hive-bot 0 1 0.0%

Generated by scripts/pr-metrics.mjs — see docs/metrics.md.

@buggtb
buggtb added this pull request to the merge queue Oct 12, 2026
Merged via the queue into development with commit e769590 Oct 12, 2026
44 of 46 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ci Created by Hive for agent-filed issue provenance docs risk/high Change risk tier: high (see docs/risk-tiers.md)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant