morluto/rea

Fork 0

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

[PR #2482] [MERGED] fix(cli): normalize empty Windows profile before imports #5261

New issue

Closed

opened 2026-10-31 01:05:15 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-31 01:05:15 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/2482
Author: @N0zoM1z0
Created: 10/13/2026
Status: ✅ Merged
Merged: 10/13/2026
Merged by: @N0zoM1z0

Base: main ← Head: fix/windows-empty-profile-startup-2345

📝 Commits (2)

e3ce11c fix(cli): normalize empty Windows profile before imports
aca75a5 fix(ci): own the Windows home fallback regression scope

📊 Changes

4 files changed (+77 additions, -0 deletions)

View changed files

📝 .github/workflows/ci.yml (+1 -0)
➕ experiments/analysis-efficiency/windows-empty-profile.md (+69 -0)
📝 scripts/ci/scopes.mjs (+1 -0)
📝 scripts/rea.mjs (+6 -0)

📄 Description

Summary

Follow-up to #2345; thanks to @gefrezim for identifying the import-time Windows home lookup failure.

Treat USERPROFILE="" as unset at the executable dispatcher, before dependency imports. Windows Node otherwise throws uv_os_homedir during Incur import, before REA's existing account-home fallback can run.
Preserve every nonempty selection and use the existing home resolver. No new lookup, fallback abstraction, filesystem query, retry or cache.
Reuse tests/acceptance/setup/homeDirectoryFallback.test.ts in the actual Windows curated CI lane. Its public setup dry-run already covers the failure; no duplicate startup test is added.

Real Windows verification

Fresh main baseline 172407787f1842dedbcd27b5385177595f786ce4, Windows 11 10.0.22631.0, official Windows x64 Node 24.18.0, locked production dependencies:

Profile input (`HOME` empty)	Baseline CLI version/help	Corrected CLI version/help
Unset	Both exit 0	Both exit 0
Empty	Both exit 1, `uv_os_homedir`	Both exit 0, no stderr
Nonempty owned profile	Both exit 0	Both exit 0

Real setup dry runs with empty/nonempty selections choose the expected home, preserve existing configuration bytes and create no configuration under the working directory. No installation was performed.
Actual SDK 2.3.1 stdio MCP discovery passes all three inputs before and after: 187 tools, 2,545,790 serialized bytes, identical catalogue digest. MCP was already healthy; no MCP startup repair is claimed.
Nine related tests pass: existing home-fallback acceptance, production MCP doctor and curated CI path checks.
Runtime build, static gate and documentation checks pass.
No IDA or Ghidra engine was used for this CLI startup fix.

Efficiency and limits

Verification and resource record pins the tested dispatcher digest and scope. Version/help replies remain 6 / 15,497 bytes. Corrected single-CLI exit-time resource observations range from 7.968–9.874 seconds CPU and 299,956–301,852 KiB peak working set; these exclude the probe parent. Probes ran serially, BelowNormal, one CPU, 512 MiB child heaps. Baseline resources were not paired and caches were not normalized: no latency, CPU or memory improvement is claimed. Bytes are not model tokens; actual model usage was not measured. Personal paths and ambient environment were not included in the record.

Review / CI follow-up

The original CI shard correctly rejected our missing package-scope registration for the added Windows test entry. Fixed in aca75a5f0 by registering the path in SCOPES.package. All 84 CI scope tests and two curated-path checks now pass. Runtime source and its real-Windows tested dispatcher digest are unchanged.

Final current-head verification: Windows lane executes the home-fallback regression (one Windows test passes; the POSIX lifecycle case is intentionally skipped). The unrelated in-memory server identity 10-second timeout passed on a failed-job rerun without source changes; the current-head CI and coverage gates are green. Current-head Codex review completed with no further findings and the ownership thread is resolved.

_{🔄 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/2482 **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-empty-profile-startup-2345` --- ### 📝 Commits (2) - [`e3ce11c`](https://github.com/morluto/rea/commit/e3ce11c7b756f003c522302e4cf96c70d761dd9c) fix(cli): normalize empty Windows profile before imports - [`aca75a5`](https://github.com/morluto/rea/commit/aca75a5f0cdbfc8cd4ea19e731e58bc716fece55) fix(ci): own the Windows home fallback regression scope ### 📊 Changes **4 files changed** (+77 additions, -0 deletions) <details> <summary>View changed files</summary> 📝 `.github/workflows/ci.yml` (+1 -0) ➕ `experiments/analysis-efficiency/windows-empty-profile.md` (+69 -0) 📝 `scripts/ci/scopes.mjs` (+1 -0) 📝 `scripts/rea.mjs` (+6 -0) </details> ### 📄 Description ## Summary Follow-up to #2345; thanks to @gefrezim for identifying the import-time Windows home lookup failure. - Treat `USERPROFILE=""` as unset at the executable dispatcher, before dependency imports. Windows Node otherwise throws `uv_os_homedir` during Incur import, before REA's existing account-home fallback can run. - Preserve every nonempty selection and use the existing home resolver. No new lookup, fallback abstraction, filesystem query, retry or cache. - Reuse `tests/acceptance/setup/homeDirectoryFallback.test.ts` in the actual Windows curated CI lane. Its public setup dry-run already covers the failure; no duplicate startup test is added. ## Real Windows verification Fresh main baseline `172407787f1842dedbcd27b5385177595f786ce4`, Windows 11 10.0.22631.0, official Windows x64 Node 24.18.0, locked production dependencies: | Profile input (`HOME` empty) | Baseline CLI version/help | Corrected CLI version/help | | --- | --- | --- | | Unset | Both exit 0 | Both exit 0 | | Empty | Both exit 1, `uv_os_homedir` | Both exit 0, no stderr | | Nonempty owned profile | Both exit 0 | Both exit 0 | - Real setup dry runs with empty/nonempty selections choose the expected home, preserve existing configuration bytes and create no configuration under the working directory. No installation was performed. - Actual SDK 2.3.1 stdio MCP discovery passes all three inputs before and after: 187 tools, 2,545,790 serialized bytes, identical catalogue digest. MCP was already healthy; no MCP startup repair is claimed. - Nine related tests pass: existing home-fallback acceptance, production MCP doctor and curated CI path checks. - Runtime build, static gate and documentation checks pass. - **No IDA or Ghidra engine was used** for this CLI startup fix. ## Efficiency and limits [Verification and resource record](https://github.com/morluto/rea/blob/aca75a5f0cdbfc8cd4ea19e731e58bc716fece55/experiments/analysis-efficiency/windows-empty-profile.md) pins the tested dispatcher digest and scope. Version/help replies remain 6 / 15,497 bytes. Corrected single-CLI exit-time resource observations range from 7.968–9.874 seconds CPU and 299,956–301,852 KiB peak working set; these exclude the probe parent. Probes ran serially, BelowNormal, one CPU, 512 MiB child heaps. Baseline resources were not paired and caches were not normalized: no latency, CPU or memory improvement is claimed. Bytes are not model tokens; actual model usage was not measured. Personal paths and ambient environment were not included in the record. ## Review / CI follow-up The original CI shard correctly rejected our missing package-scope registration for the added Windows test entry. Fixed in `aca75a5f0` by registering the path in `SCOPES.package`. All 84 CI scope tests and two curated-path checks now pass. Runtime source and its real-Windows tested dispatcher digest are unchanged. Final current-head verification: Windows lane executes the home-fallback regression (one Windows test passes; the POSIX lifecycle case is intentionally skipped). The unrelated in-memory server identity 10-second timeout passed on a failed-job rerun without source changes; the current-head CI and coverage gates are green. Current-head Codex review completed with no further findings and the ownership thread is resolved. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-31 01:05:15 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-11-01 21:43:27 +00:00

[PR #5261] [CLOSED] fix(javascript): preserve caller-selected input paths in Evidence #6844

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#5261

No description provided.

Rows
Columns