Skip to content

test: close bounded-query and integrity-mode acceptance gaps (#218, #219) - #233

Merged
codeasier merged 1 commit into
mainfrom
test/218-219-acceptance
Oct 11, 2026
Merged

codeasier merged 1 commit into
mainfrom
test/218-219-acceptance

Conversation

@codeasier

Copy link
Copy Markdown
Owner

Summary

Finite acceptance follow-up for #218 and #219. Tests and bilingual documentation only; no runtime, dependency, budget, integrity, or publication behavior changes.

  • Add five scratch result-phase failures (row/byte budgets, timeout, cancellation, SQLite I/O) verifying closed handles, removed private storage, unchanged sources, no sidecars, and exact retry recovery.
  • Add 36 compatibility-v1/native-v2 × default-strong/explicit-immutable regressions for actual hash-read bytes and validator calls, replacements, hidden-change-token trust boundaries, cancellation/eviction, connection ownership, and cold new processes.
  • Document measured admission work in English and Chinese. Strong warm validation remains O(bytes + shards); immutable reuse requires the caller's immutable-publication contract. Neither provides concurrent-writer locking or atomic snapshots.

Verification

On the final combined tree:

  • Focused fixture/provenance, integrity, lifecycle scratch, cache/source reuse, realistic differential and governance checks: 246 passed, 2 intentional unsupported slice-leak skips.
  • Whole-repository Ruff and Black passed; basedpyright: 0 errors (informational warnings retained); diff whitespace check passed.
  • Runtime source tree is byte-identical to current-main baseline tree 71e00c3f529b654ad200d13ca998063cc6d79490.

Separate same-baseline evidence: #218 acceptance ran core/query/CLI-API regression checks (1049 passed, 2 intentional skips, 11 slow deselected), plus nine scale cases with default protection; large prior installed-wheel/scale evidence is reused, not represented as a fresh rerun. Final-head CI still needs to run. The separate #219 full-suite run was collected before its final four cancellation tests and before combining the five #218 tests; all nine are included in the focused combined result above.

Acceptance boundaries

  • A SQLite main-database page cap is not a total temporary-disk/RSS cap.
  • Normal exception/timeout/cancellation cleanup is covered; abrupt process-kill cleanup is not promised.
  • New CLI processes may remain cold; no persistent cache or dataset-size-independent strong validation is claimed.
  • This remains draft pending exact-head CI and acceptance review. No issue auto-close or automatic merge is requested.

Copy link
Copy Markdown
Owner Author

最终远端验收通过:head 459e09549bc883e1a015db235af06c947837d398,tree fdd05481d3e467f2546d52bda2dc2b61dabe1d61;4个变更文件blob均与冻结内容一致,运行时src未变。

CI https://github.com/codeasier/pt-snap-cli/actions/runs/38104812415 已全部7个jobs成功。

  • 新增36项完整性模式测试和5项scratch结果阶段清理测试,在Python3.10–3.13日志中均逐条PASSED。
  • Python3.11/3.12/3.13各2311 passed /15 skipped;3.10为2280 passed /46 skipped。普通matrix中的browser skip由独立Chromium job覆盖,不计为浏览器通过依据。
  • 独立Chromium12项交互测试加7项provenance共19 passed、0 skipped。
  • Python3.13 wheel audit和clean-wheel acceptance成功。

关闭范围仍以本PR中的有限验收契约为准:strong默认完整哈希、immutable依赖显式不可变信任、新CLI可cold;scratch页限额不等于总disk/RSS上限,不承诺强杀清理。保持draft,等待合入确认。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant