morluto/rea

Fork 0

mirror of https://github.com/morluto/rea.git synced 2026-11-05 06:45:22 +00:00

[PR #5627] [MERGED] test(evaluation): include native preparation in test deadlines #7003

New issue

Closed

opened 2026-11-01 21:44:38 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-11-01 21:44:38 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/5627
Author: @gensokyo-dev
Created: 10/20/2026
Status: ✅ Merged
Merged: 10/20/2026
Merged by: @gensokyo-dev

Base: main ← Head: test/5094-native-preparation-budgets

📝 Commits (1)

e37289c test(evaluation): include native preparation in test deadlines

📊 Changes

2 files changed (+26 additions, -4 deletions)

View changed files

📝 tests/boundary/process/agentEvaluationCleanup.test.ts (+7 -1)
📝 tests/boundary/process/agentEvaluationCommands.test.ts (+19 -3)

📄 Description

Summary

Refs #5094 and #5615. Native ownership preparation now has its own bounded deadline, but the evaluation subprocess tests still use 15/35-second parent deadlines. On a cold Darwin SDK cache those can kill the probe before the independently budgeted compiler finishes; the owning preparation module records a measured cold compilation of about 88 seconds. This is a source-derived harness mismatch, not a new macOS runtime reproduction.

Derive the real-subprocess test deadlines from resolveSwiftcCompileTimeoutMs() on Darwin, plus their existing command/test allowances. Linux/Windows budgets and the short command deadlines being tested remain unchanged. The permission/fixture cleanup matrix inherits its existing non-Darwin test-runner timeout. Totals are clamped to Node's representable timer maximum. Controlled preparation-injection cases keep their existing short bounds because they run no compiler.

The real-child driver also joins its retained command scope in finally through the existing verifier cleanup helper, preserving a primary assertion/command failure if cleanup also fails. Only two test files change. No production behavior, provider authority, CLI/MCP contract, Evidence, dependencies, cache, agent context or analysis cost changes. Additional maintainer scope decision: not applicable.

Executed validation

Pinned main 7d6f90485, Linux x64, Node 24.18.0 / npm 11.16.0:

npm run check:fast passes, including another run on final head e37289c63: typecheck, lint and resolved module-boundary checks.
12 focused tests pass against the already built unchanged runtime: real child/output, exact-owner retry, preparation transfer and permission/fixture cleanup. The exact-file Vitest runner avoids rebuilding unchanged production code after the harness-only edit.
One earlier attempt failed conservatively on a transient live process whose ownership token was unreadable (EACCES); its receipt is retained. Ownership uncertainty remains a failure; no production ownership check is weakened.
Formatting and diff whitespace checks pass. No redundant arithmetic-only regression tests are added.

Native macOS cold compilation remains unverified on this Linux host. The timeout calculation follows the existing preparation owner/configuration rather than introducing a second guessed compile limit. Actual cold/warm Darwin test latency is unmeasured. Existing Linux fixtures exercise real Node processes with inert data; no vendor model, authentication copy or target execution runs.

Checks run sequentially with one CPU, 8-GiB hard / 7-GiB early family bounds, no swap, and a host-pressure guard. Resource receipts remain locally under /tmp/rea-5094-test-budget-proof; only observed results are claimed. The completed focused task costs 57.582 s elapsed, 43.430 s family CPU, 477.9 MiB peak; payload-exit counters confirm cleanup and no OOM (not final whole-scope counters). This test-only adjustment has no production token/performance comparison.

CI/review assessment and authorized bypass

Latest head e37289c63: Codex completes with no findings, Greptile completes with 5/5, no unresolved review thread. Classification/title pass; readiness/static/source/build remain queued. No unfinished CI is claimed as passing.

The task owner explicitly authorized bypass after independent qualification. This test-only change is qualified by final-head checks, 12 real child/cleanup tests and a source-derived match to the existing preparation owner's budget. Native macOS verification remains a stated gap. A fresh head/unresolved-review check gates merge; pending CI results remain visible.

Author readiness

Human owner: @gensokyo-dev

The task owner explicitly delegated validation, readiness declarations and merge to this agent. These declarations are supplied under that authorization; no separate new manual human review by the original author is asserted.

I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work.
I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps.
I declare this PR ready for initial maintainer review.

Before merge

Directly related test-harness cleanup; production deadlines and authority unchanged.
Final tests and available latest-head CI/review assessed; unfinished CI and authorized bypass recorded above.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/5627 **Author:** [@gensokyo-dev](https://github.com/gensokyo-dev) **Created:** 10/20/2026 **Status:** ✅ Merged **Merged:** 10/20/2026 **Merged by:** [@gensokyo-dev](https://github.com/gensokyo-dev) **Base:** `main` ← **Head:** `test/5094-native-preparation-budgets` --- ### 📝 Commits (1) - [`e37289c`](https://github.com/morluto/rea/commit/e37289c636c94604c1bed12d204ee3bb4f6aaf68) test(evaluation): include native preparation in test deadlines ### 📊 Changes **2 files changed** (+26 additions, -4 deletions) <details> <summary>View changed files</summary> 📝 `tests/boundary/process/agentEvaluationCleanup.test.ts` (+7 -1) 📝 `tests/boundary/process/agentEvaluationCommands.test.ts` (+19 -3) </details> ### 📄 Description ## Summary Refs #5094 and #5615. Native ownership preparation now has its own bounded deadline, but the evaluation subprocess tests still use 15/35-second parent deadlines. On a cold Darwin SDK cache those can kill the probe before the independently budgeted compiler finishes; the owning preparation module records a measured cold compilation of about 88 seconds. This is a source-derived harness mismatch, not a new macOS runtime reproduction. Derive the real-subprocess test deadlines from `resolveSwiftcCompileTimeoutMs()` on Darwin, plus their existing command/test allowances. Linux/Windows budgets and the short command deadlines being tested remain unchanged. The permission/fixture cleanup matrix inherits its existing non-Darwin test-runner timeout. Totals are clamped to Node's representable timer maximum. Controlled preparation-injection cases keep their existing short bounds because they run no compiler. The real-child driver also joins its retained command scope in `finally` through the existing verifier cleanup helper, preserving a primary assertion/command failure if cleanup also fails. Only two test files change. No production behavior, provider authority, CLI/MCP contract, Evidence, dependencies, cache, agent context or analysis cost changes. Additional maintainer scope decision: not applicable. ## Executed validation Pinned main `7d6f90485`, Linux x64, Node 24.18.0 / npm 11.16.0: - `npm run check:fast` passes, including another run on final head **e37289c63**: typecheck, lint and resolved module-boundary checks. - **12 focused tests pass** against the already built unchanged runtime: real child/output, exact-owner retry, preparation transfer and permission/fixture cleanup. The exact-file Vitest runner avoids rebuilding unchanged production code after the harness-only edit. - One earlier attempt failed conservatively on a transient live process whose ownership token was unreadable (`EACCES`); its receipt is retained. Ownership uncertainty remains a failure; no production ownership check is weakened. - Formatting and diff whitespace checks pass. No redundant arithmetic-only regression tests are added. Native macOS cold compilation remains unverified on this Linux host. The timeout calculation follows the existing preparation owner/configuration rather than introducing a second guessed compile limit. Actual cold/warm Darwin test latency is unmeasured. Existing Linux fixtures exercise real Node processes with inert data; no vendor model, authentication copy or target execution runs. Checks run sequentially with one CPU, 8-GiB hard / 7-GiB early family bounds, no swap, and a host-pressure guard. Resource receipts remain locally under `/tmp/rea-5094-test-budget-proof`; only observed results are claimed. The completed focused task costs **57.582 s elapsed**, **43.430 s family CPU**, **477.9 MiB peak**; payload-exit counters confirm cleanup and no OOM (not final whole-scope counters). This test-only adjustment has no production token/performance comparison. ## CI/review assessment and authorized bypass Latest head `e37289c63`: Codex completes with **no findings**, Greptile completes with **5/5**, no unresolved review thread. Classification/title pass; readiness/static/source/build remain queued. No unfinished CI is claimed as passing. The task owner explicitly authorized bypass after independent qualification. This test-only change is qualified by final-head checks, 12 real child/cleanup tests and a source-derived match to the existing preparation owner's budget. Native macOS verification remains a stated gap. A fresh head/unresolved-review check gates merge; pending CI results remain visible. ## Author readiness Human owner: @gensokyo-dev The task owner explicitly delegated validation, readiness declarations and merge to this agent. These declarations are supplied under that authorization; no separate new manual human review by the original author is asserted. - [x] I reviewed the final diff and relevant existing code, understand the change and its edge cases, and take responsibility for the complete PR, including AI-assisted work. - [x] I checked the evidence above, confirmed it supports the current change, and recorded meaningful validation gaps. - [x] I declare this PR ready for initial maintainer review. ## Before merge - [x] Directly related test-harness cleanup; production deadlines and authority unchanged. - [x] Final tests and available latest-head CI/review assessed; unfinished CI and authorized bypass recorded above. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-11-01 21:44:38 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-11-03 13:38:45 +00:00

[PR #7003] [MERGED] fix(process): preserve caller-selected working directory in scenario validation errors #7564

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

morluto/rea#7003

No description provided.

Rows
Columns