[Refactor] Сократить workload SQLite runner regression без потери safety proof #72

Closed
opened 2026-10-11 21:47:21 +03:00 by codex-bot · 1 comment
Owner

Problem

Opt-in regression выполняет 3 sequential +2 parallel full --all: 95 functional suite invocations ради directory/process/sentinel/cleanup guarantees. Это configured count, не measured CI duration; ordinary CI этим command не пользуется.

Expected Behavior

Runner mechanics проверяются меньшим existing real SQLite payload; full product e2e остаётся самостоятельным gate.

Evidence

Audit date 2026-10-11; 09. IDs TI-005, CA-H02. Revision 0ad84ff4f0b4712abc7e69dc299856f14af6eadd; Confidence High configured workload; Verification status static.
Affected: test/sqlite-regression.cjs:20–91; sqlite.cjs:7–17,58–89; README:250–254.

Scope

In scope

Existing small-suite selection, прежние failure/signal/sentinel assertions.

Out of scope

Новый fake suite/framework, removal runner env guard, пользовательская DB, уменьшение canonical functional coverage.

Proposed Approach

Минимум две real suites на normal pass и два одновременных runner processes; migrations/Jest и failure branches настоящие.

Acceptance Criteria

  • Уникальность DB между suites/processes, sentinel unchanged и cleanup доказаны теми же observations.
  • Migration failure, SIGINT/SIGTERM дают nonzero и не оставляют работающих children/temp directories.
  • Полный product --all selection неизменён; before/after одного opt-in запуска записан без выдуманного speedup.

Verification

Existing sqlite-regression command на disposable workload; static assertion parity плюс один measured run каждого варианта при implementation.

Dependencies

Нет.

Traceability

Campaign WealthLedger/main#23 . Findings TI-005, CA-H02. WealthLedger/backend; T19.

<!-- wealthledger-audit:2026-10-11:sqlite-runner-workload --> ## Problem Opt-in regression выполняет 3 sequential +2 parallel full --all: 95 functional suite invocations ради directory/process/sentinel/cleanup guarantees. Это configured count, не measured CI duration; ordinary CI этим command не пользуется. ## Expected Behavior Runner mechanics проверяются меньшим existing real SQLite payload; full product e2e остаётся самостоятельным gate. ## Evidence Audit date `2026-10-11`; [09](https://git.personal.ksv741.ru/WealthLedger/main/src/commit/d07fbfabb1295780952a554185989d2d5b5c8349/audits/2026-10-11/reports/09-test-architecture-and-infrastructure-report.md). IDs TI-005, CA-H02. Revision `0ad84ff4f0b4712abc7e69dc299856f14af6eadd`; Confidence High configured workload; Verification status static. Affected: [test/sqlite-regression.cjs:20–91](https://git.personal.ksv741.ru/WealthLedger/backend/src/commit/0ad84ff4f0b4712abc7e69dc299856f14af6eadd/test/sqlite-regression.cjs#L20-L91); sqlite.cjs:7–17,58–89; README:250–254. ## Scope ### In scope Existing small-suite selection, прежние failure/signal/sentinel assertions. ### Out of scope Новый fake suite/framework, removal runner env guard, пользовательская DB, уменьшение canonical functional coverage. ## Proposed Approach Минимум две real suites на normal pass и два одновременных runner processes; migrations/Jest и failure branches настоящие. ## Acceptance Criteria - [ ] Уникальность DB между suites/processes, sentinel unchanged и cleanup доказаны теми же observations. - [ ] Migration failure, SIGINT/SIGTERM дают nonzero и не оставляют работающих children/temp directories. - [ ] Полный product --all selection неизменён; before/after одного opt-in запуска записан без выдуманного speedup. ## Verification Existing sqlite-regression command на disposable workload; static assertion parity плюс один measured run каждого варианта при implementation. ## Dependencies Нет. ## Traceability Campaign https://git.personal.ksv741.ru/WealthLedger/main/issues/23 . Findings TI-005, CA-H02. WealthLedger/backend; T19.
codex-bot added this to the Technical Audit Remediation — 2026-10-11 milestone 2026-10-11 21:47:21 +03:00
codex-bot added the Kind/Enhancement
Priority
Low
4
audit/2026-10-11
labels 2026-10-11 21:47:21 +03:00
codex-bot was assigned by hermes-agent 2026-10-12 00:04:20 +03:00
hermes-agent added this to the WealthLedger — Delivery Board project 2026-10-12 00:04:20 +03:00
hermes-agent added Kind/Refactor and removed Kind/Enhancement labels 2026-10-12 00:04:20 +03:00
Author
Owner

Implemented and merged via #88 (1033d21).

The runner regression uses the existing two-suite real SQLite payload, retaining sequential/parallel isolation, sentinel, migration failure, and SIGINT/SIGTERM cleanup checks. The complete suite remains npm run test:e2e. Verified with node test/sqlite-regression.cjs.

Implemented and merged via #88 (`1033d21`). The runner regression uses the existing two-suite real SQLite payload, retaining sequential/parallel isolation, sentinel, migration failure, and SIGINT/SIGTERM cleanup checks. The complete suite remains `npm run test:e2e`. Verified with `node test/sqlite-regression.cjs`.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: WealthLedger/backend#72