[test-qa-hourly] test: pin deterministic metric snapshot tie-break

[i-xtsu-sixyou-ken-mei]

Intent

Continue continuous QA from current main with one small regression-focused test around deterministic waitlist conversion metrics.

Inspection

• Previous QA PR [test-qa-hourly] test: cover waitlist signup pagination boundary #427 is merged; no open PR matches [test-qa-hourly], label test-qa, or branch prefix test-qa/ongoing-.
• Current main is ea659d8e6f5e3e72e92289c7dd0814be46623d63 (merge of [test-qa-hourly] test: cover waitlist signup pagination boundary #427).
• Main CI #2429 (34313809759) completed successfully.
• No open PRs currently overlap the target path.
• apps/control-center/src/server/services/waitlist-growth.ts reads collected 24h metric snapshots ordered by captured_at ASC, then id ASC; later rows overwrite earlier entries in the per-post views map.
• Existing regression coverage pins latest-captured_at behavior and 500-row pagination, but did not directly pin the id tie-break when two snapshots share an identical timestamp.
• .github/workflows/ci.yml still runs quick gates (format repo contracts), code quality (type-check lint deadcode dup), tests (test analytics), E2E, security, and deploy gates.
• Coverage remains available through scripts/coverage.sh summary -> turbo run test:coverage + scripts/coverage-summary.ts.

QA change

Strengthen waitlist-growth-metrics-pagination.test.ts with one regression test where two collected 24h snapshots for the same post share the exact same captured_at value. The higher ordered metric id must deterministically win, preserving views24h = 250 and signupRate = 1 / 250.

This is test-only. No implementation, workflow, docs, ignore, coverage threshold, or agent-instruction change is included.

Validation / commands and results

• GitHub main CI #2429 — PASS.
• Source review of waitlist-growth.ts — PASS: current query explicitly orders captured_at ASC, then id ASC and overwrites the views map in read order.
• Existing test review — PASS: latest-timestamp and pagination behavior already covered; exact-timestamp id tie-break was not directly asserted.
• .github/workflows/ci.yml inspection — PASS: existing gates unchanged.
• scripts/coverage.sh inspection — PASS: coverage aggregation path remains available.
• Focused local test / broader test, lint, typecheck, coverage — not executable in this GitHub-connector runtime; PR CI is the executable repository validation gate.
• Updated-head PR CI — pending after PR creation.

Scope guard

Changed file only:

• apps/control-center/src/server/services/waitlist-growth-metrics-pagination.test.ts

Auto-merge

Not enabled. Repository setting allow_auto_merge is currently false; never bypass protections or repository policy.

Next QA target

Inspect concurrent mutation during social_post_metrics offset pagination. A delete/insert between pages should not silently turn a partial/stale snapshot into a plausible conversion rate; identify the smallest fail-closed invariant that can be tested without overbuilding the telemetry layer.

[i-xtsu-sixyou-ken-mei]

QA run update — 2026-09-09

Inspection

• Reused this open QA PR ([test-qa-hourly], branch test-qa/ongoing-coverage-20260909).
• Previous head dbbfdce023a7aec390843906e0828747385124a4 CI #2430 (34315477146) — PASS.
• Branch was already synced with main before this run (behind 0).
• Reviewed current PR diff and active PRs Clarify reliability priority scores and check counts #429 / fix: keep visual fallbacks on topic and simplify scene trace #430; neither overlaps this QA target.
• Reviewed apps/control-center/src/server/services/waitlist-growth.ts: metric reads are constrained to collected 24h rows, bounded by captured_at <= now, ordered by captured_at ASC then id ASC, and paged in 500-row windows.
• Reviewed existing metric pagination/tie-break tests and CI/coverage contract already recorded in this PR.

QA change

Added apps/control-center/src/server/services/waitlist-growth-metric-query-contract.test.ts to pin the query-level snapshot contract itself. The regression asserts that metric attribution cannot silently drift by removing any of these safeguards:

• social_post_id batch filter,
• measurement_window = 24h,
• collection_status = collected,
• fixed captured_at upper bound,
• stable captured_at ASC, id ASC ordering,
• first page range 0..499.

Commit: d400fae4fbc845d1c5c84bba47ba4bc774139cea — test(control-center): pin metric snapshot query contract.

Commands / validation

• GitHub fetch_commit_workflow_runs for previous head — PASS, CI #2430 completed successfully.
• GitHub compare main...test-qa/ongoing-coverage-20260909 before change — PASS, ahead 1 / behind 0, test-only diff.
• Source inspection of waitlist-growth.ts — PASS; implementation currently matches the new query assertions.
• PR diff review after change — PASS; current QA scope is two test files only, no implementation/workflow/docs/ignore/threshold changes.
• Updated-head GitHub Actions CI #2435 (34319402683) — QUEUED.
• Local focused/broader test, lint, typecheck, and coverage commands are not executable in this connector-only runtime; repository CI remains the executable validation gate.

Auto-merge

Not enabled. Repository setting allow_auto_merge=false; do not bypass repository policy.

Next QA target

After CI is green, test a minimal fail-closed invariant for concurrent mutation during social_post_metrics offset pagination—specifically a duplicate metric row appearing across adjacent pages due to row displacement.

[i-xtsu-sixyou-ken-mei]

QA run update — CI-first repair

Inspection

• Reused this open QA PR ([test-qa-hourly], branch test-qa/ongoing-coverage-20260909).
• Previous head d400fae4fbc845d1c5c84bba47ba4bc774139cea CI #2435 (34319402683) completed failure.
• CI job inspection: quick-gates failed; e2e, security, check-dead-env, and code-quality passed. Coverage artifacts were produced.
• Downloaded quick-gates-verify-1 artifact and inspected result.json: format=failed, repo=passed, contracts=passed.
• format.log showed exactly one Prettier drift in apps/control-center/src/server/services/waitlist-growth-metric-query-contract.test.ts (toHaveBeenNthCalledWith formatting only).
• Current main advanced to a38cd9e987e88415cd8d4670f8aed21f8378ab4d via generated snapshot commits. Compare showed only analytics/landing-page snapshot data changes, with no overlap with the QA test files.

Action

1. Applied canonical Prettier formatting only to the failing assertion; no behavior change.
2. Safely synchronized the QA branch to current main by replaying the two QA test-file changes on top of a38cd9e... because the branch had diverged only through non-overlapping generated snapshot commits.
3. Restored:
   • deterministic metric-id tie-break regression in waitlist-growth-metrics-pagination.test.ts
   • metric query-contract regression in waitlist-growth-metric-query-contract.test.ts, already in canonical Prettier form.

Current replay commits:

• 09e2390bfacb3bbddf8be5a4944a4420098be224 — restore metric tie-break regression after main sync
• 23632d7a4ccf606d00f353140a0ae60ab9db3067 — restore metric query contract after main sync

Commands / results

• GitHub CI #2435 job inspection — FAIL isolated to quick-gates.
• quick-gates-verify-1/result.json — format FAIL, repo PASS, contracts PASS.
• quick-gates-verify-1/logs/format.log — one Prettier-only diff in the new query-contract test.
• Compare test-qa/ongoing-coverage-20260909...main — branch diverged; main-side files were generated snapshots only, no QA-path overlap.
• Focused local test / broader test / lint / typecheck / coverage — not executable in this connector runtime; updated-head GitHub CI is the executable repository validation gate.

Scope guard

Still test-only: the PR changes only the two Control Center waitlist-growth test files. No production implementation, workflow, docs, ignores, CI gates, or coverage thresholds changed.

Auto-merge

Not enabled in this run. Will only be considered after updated-head CI is green and repository auto-merge policy permits it.

Next QA target

After CI is green, exercise concurrent mutation around offset pagination: duplicate/disappearing metric rows across adjacent pages should not silently yield a plausible stale conversion rate.