[PR #2635] [MERGED] test(javascript): verify Windows path case and load scope #5298

New issue

Closed

opened 2026-10-31 01:05:32 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-31 01:05:32 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/2635
Author: @N0zoM1z0
Created: 10/13/2026
Status: ✅ Merged
Merged: 10/13/2026
Merged by: @N0zoM1z0

Base: main ← Head: fix/windows-posix-runtime-case-2422

📝 Commits (4)

6b301b7 fix(javascript): preserve recorded POSIX path case on Windows
0748c99 docs(javascript): record Windows path integration verification
52e9af7 fix(ci): run Windows regressions for matcher source edits
e6e8694 docs(javascript): record merged matcher ownership

📊 Changes

5 files changed (+224 additions, -65 deletions)

View changed files

📝 .github/workflows/ci.yml (+1 -0)
➕ experiments/analysis-efficiency/windows-posix-reconciliation.md (+128 -0)
📝 scripts/ci/scopes.mjs (+2 -0)
📝 tests/boundary/filesystem/ciChangeScope.test.ts (+4 -0)
📝 tests/boundary/filesystem/javascriptApplicationWorkflows.test.ts (+89 -65)

📄 Description

Windows path regression coverage

A matcher-only edit previously selected JavaScript recovery CI but skipped the Windows lane, where a host-default relative-path operation could treat /runtime/App and /runtime/app as the same recorded POSIX root. This follow-up to merged #2422 makes both the matcher and its workflow tests select the Windows lane. Thanks to @Aurantium-47 for identifying the path-grammar bug.

The four existing controls now assert matching basis, coherent target/script coordinates, retained caller path spelling and static load state. Rejected POSIX membership must remain unknown and must not mark a static asset loaded; exact POSIX and case-insensitive Windows mappings continue to match. The final diff contains workflow tests, CI scope ownership and the verification record.

Validation

Rebased onto main fb28ec153: 230 focused tests pass, including the updated JavaScript workflows, CI scope/required-lane contracts, semantic structure, release-checkpoint and real x64 dyld CLI admission. Complete check:ci passes all eight tasks and architecture guards; docs checks pass. The unrelated release-checkpoint and dyld fixture failures were corrected upstream; this PR adds no patches to those files.
Actual Windows 11 x64 CLI/MCP verification uses npm 11.16.0's published validate-lockfile.js, independently matched to the official registry artifact. A controlled Node process exercised 587 lock records and passive REA Inspector Evidence recorded 114 loaded scripts.
The final Windows MCP replay at e6e86946d passes exact/wrong-case implicit and explicit POSIX cases. All four normalized hashes match the previously verified CLI/MCP results, preserving original Evidence and caller-selected mappings. No owned Node processes remained after the replay.
Electron Evidence controls are service regressions. The real provider used for runtime capture was Node Inspector; this verification does not establish Electron GUI, IDA or Ghidra behavior.

Probes run serially at BelowNormal priority, on one CPU, with 512 MiB Node heaps. Final same-connection calls took 251–620 ms; sampled process-family working-set sum peaked at 499,380,224 bytes (shared pages may be counted twice; exited tails may be missed). Discovery plus request/reply payloads totalled 6,975,765 bytes. Only compact receipts were presented for inspection; duplicate successful output files are removed after verification. These are resource and delivery observations, not model token counts or a paired speedup claim. The change adds no runtime I/O or cache.

Source pins, public-artifact provenance and measurement limits: Windows verification record.

_{🔄 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/2635 **Author:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Created:** 10/13/2026 **Status:** ✅ Merged **Merged:** 10/13/2026 **Merged by:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Base:** `main` ← **Head:** `fix/windows-posix-runtime-case-2422` --- ### 📝 Commits (4) - [`6b301b7`](https://github.com/morluto/rea/commit/6b301b7d24b8ba60f53d404ac007c77829cafac5) fix(javascript): preserve recorded POSIX path case on Windows - [`0748c99`](https://github.com/morluto/rea/commit/0748c9962a9d6ba0c7ede8164ff7c3ee7ea0d039) docs(javascript): record Windows path integration verification - [`52e9af7`](https://github.com/morluto/rea/commit/52e9af73b1c87eaa83bc18cb91cada6ee2cdf91a) fix(ci): run Windows regressions for matcher source edits - [`e6e8694`](https://github.com/morluto/rea/commit/e6e86946dd28f6927ebefc075e68414a97e04f6d) docs(javascript): record merged matcher ownership ### 📊 Changes **5 files changed** (+224 additions, -65 deletions) <details> <summary>View changed files</summary> 📝 `.github/workflows/ci.yml` (+1 -0) ➕ `experiments/analysis-efficiency/windows-posix-reconciliation.md` (+128 -0) 📝 `scripts/ci/scopes.mjs` (+2 -0) 📝 `tests/boundary/filesystem/ciChangeScope.test.ts` (+4 -0) 📝 `tests/boundary/filesystem/javascriptApplicationWorkflows.test.ts` (+89 -65) </details> ### 📄 Description ## Windows path regression coverage A matcher-only edit previously selected JavaScript recovery CI but skipped the Windows lane, where a host-default relative-path operation could treat `/runtime/App` and `/runtime/app` as the same recorded POSIX root. This follow-up to merged #2422 makes both the matcher and its workflow tests select the Windows lane. Thanks to @Aurantium-47 for identifying the path-grammar bug. The four existing controls now assert matching basis, coherent target/script coordinates, retained caller path spelling and static load state. Rejected POSIX membership must remain unknown and must not mark a static asset loaded; exact POSIX and case-insensitive Windows mappings continue to match. The final diff contains workflow tests, CI scope ownership and the verification record. ## Validation - Rebased onto main `fb28ec153`: 230 focused tests pass, including the updated JavaScript workflows, CI scope/required-lane contracts, semantic structure, release-checkpoint and real x64 dyld CLI admission. Complete `check:ci` passes all eight tasks and architecture guards; docs checks pass. The unrelated release-checkpoint and dyld fixture failures were corrected upstream; this PR adds no patches to those files. - Actual Windows 11 x64 CLI/MCP verification uses npm 11.16.0's published `validate-lockfile.js`, independently matched to the official registry artifact. A controlled Node process exercised 587 lock records and passive REA Inspector Evidence recorded 114 loaded scripts. - The final Windows MCP replay at `e6e86946d` passes exact/wrong-case implicit and explicit POSIX cases. All four normalized hashes match the previously verified CLI/MCP results, preserving original Evidence and caller-selected mappings. No owned Node processes remained after the replay. - Electron Evidence controls are service regressions. The real provider used for runtime capture was Node Inspector; this verification does not establish Electron GUI, IDA or Ghidra behavior. Probes run serially at BelowNormal priority, on one CPU, with 512 MiB Node heaps. Final same-connection calls took 251–620 ms; sampled process-family working-set sum peaked at 499,380,224 bytes (shared pages may be counted twice; exited tails may be missed). Discovery plus request/reply payloads totalled 6,975,765 bytes. Only compact receipts were presented for inspection; duplicate successful output files are removed after verification. These are resource and delivery observations, not model token counts or a paired speedup claim. The change adds no runtime I/O or cache. Source pins, public-artifact provenance and measurement limits: [Windows verification record](https://github.com/morluto/rea/blob/e6e86946dd28f6927ebefc075e68414a97e04f6d/experiments/analysis-efficiency/windows-posix-reconciliation.md). --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-31 01:05:32 +00:00