morluto/rea

Fork 0

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

[PR #2506] [MERGED] fix(native): name the ownership helper prerequisite in command failures #5283

New issue

Closed

opened 2026-10-31 01:05:26 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-31 01:05:26 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/2506
Author: @kaoru0822-kitauji
Created: 10/13/2026
Status: ✅ Merged
Merged: 10/13/2026
Merged by: @kaoru0822-kitauji

Base: main ← Head: fix/native-ownership-helper-detail

📝 Commits (3)

050bac6 fix(native): name the ownership helper prerequisite in command failures
f402f78 test(native): cover the provider's ownership helper diagnostics
160ef99 Merge remote-tracking branch 'origin/main' into fix/native-ownership-helper-detail

📊 Changes

6 files changed (+90 additions, -2 deletions)

View changed files

📝 src/artifacts/MachOSliceArtifactReader.ts (+3 -1)
📝 src/native/CommandRunner.ts (+18 -1)
📝 src/native/NativeMacOSProvider.ts (+1 -0)
📝 tests/boundary/providers/native/machoSliceReader.test.ts (+22 -0)
📝 tests/boundary/providers/native/nativeCommandTermination.test.ts (+21 -0)
📝 tests/boundary/providers/native/nativeProvider.test.ts (+25 -0)

📄 Description

Problem

Every supervised native command launch on macOS needs the Darwin process-ownership helper. When that helper can't be compiled, XcrunCommandRunner turned the typed DarwinProcessOwnershipInspectionError into a bare io failure and dropped its message. Causes include a relative REA_DARWIN_SWIFTC_PATH, REA_DARWIN_SWIFTC_PATH without an absolute SDKROOT, a missing compiler, or an SDK/compiler mismatch after a toolchain update. Users then saw messages that pointed nowhere near the cause:

$ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-artifact /bin/ls --format json
… "code":"artifact_operation_failed", … "detail":"lipo could not enumerate universal Mach-O slices"

$ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-macho /bin/ls --format json
… "code":"execution_failure", "message":"… Run `rea doctor` …",
  "diagnostics":{"tool":"file","reason":"io","exit_code":null,…}

rea doctor doesn't report it either. The cause chain was lipo: io → xcrun: io → DarwinProcessOwnershipInspectionError("Direct Darwin helper compilation requires absolute REA_DARWIN_SWIFTC_PATH and SDKROOT …"). The SQLite and web-capture providers already show this message in diagnostics.reason. Only the native path lost it.

Fix

CommandRunner classifies the typed inspection error once, at the launch boundary. It stores the message as NativeCommandFailure.detail and carries it through the xcrun --find resolution wrapper.
NativeMacOSProvider adds detail to the ProviderAdapterError diagnostics.
MachOSliceArtifactReader appends the detail to its "lipo could not enumerate" message.

Failure reasons and categories are unchanged. When the helper is healthy, the output is byte-identical: inspect-artifact /bin/ls returns the same evidence ID before and after.

After:

$ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-artifact /bin/ls --format json
… "detail":"lipo could not enumerate universal Mach-O slices: Direct Darwin helper compilation requires absolute REA_DARWIN_SWIFTC_PATH and SDKROOT paths selecting a compatible installed Swift compiler and macOS SDK."

$ REA_DARWIN_SWIFTC_PATH=/nonexistent SDKROOT=/x rea inspect-macho /bin/ls --format json
… "diagnostics":{…,"reason":"io",…,"detail":"macOS process ownership inspection requires the Apple Swift compiler via the selected Swift compiler (REA_DARWIN_SWIFTC_PATH) (compiler result: ENOENT): no compiler diagnostics were captured"}

Validation

Two new boundary tests use the existing NativeCommandLauncher seam: one on the runner (nativeCommandTermination.test.ts) and one on the slice reader (machoSliceReader.test.ts). Both fail on main and pass with this change. The production ownership host is a process-wide singleton built from process.env at import time, so it can't be injected per test.
npm run test:focused passed for those two files plus nativeProvider.test.ts and MachOSliceArtifactReader.test.ts (45 tests). npm run check:fast and format:check also passed.
I ran the real CLI on macOS with Command Line Tools and covered three cases: a relative compiler path, a missing SDKROOT, and a nonexistent compiler.

Remaining

rea doctor still doesn't check whether the ownership helper compiles, so its remediation can't point at this. That needs its own doctor check, which I left out of this PR.

🤖 Generated with Claude Code

Summary by CodeRabbit

Bug Fixes
- Improved diagnostics for Mach-O slice enumeration and native command failures by including relevant system error details when available.
- Clarified when a required macOS process-inspection helper is unavailable.

_{🔄 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/2506 **Author:** [@kaoru0822-kitauji](https://github.com/kaoru0822-kitauji) **Created:** 10/13/2026 **Status:** ✅ Merged **Merged:** 10/13/2026 **Merged by:** [@kaoru0822-kitauji](https://github.com/kaoru0822-kitauji) **Base:** `main` ← **Head:** `fix/native-ownership-helper-detail` --- ### 📝 Commits (3) - [`050bac6`](https://github.com/morluto/rea/commit/050bac62ca25b7c99f8c2f049d822476069a8cbb) fix(native): name the ownership helper prerequisite in command failures - [`f402f78`](https://github.com/morluto/rea/commit/f402f78894b7abed11450419d8cbeb6f364011ff) test(native): cover the provider's ownership helper diagnostics - [`160ef99`](https://github.com/morluto/rea/commit/160ef99fa224f099ca0921dd6c51b8bb04ea64f5) Merge remote-tracking branch 'origin/main' into fix/native-ownership-helper-detail ### 📊 Changes **6 files changed** (+90 additions, -2 deletions) <details> <summary>View changed files</summary> 📝 `src/artifacts/MachOSliceArtifactReader.ts` (+3 -1) 📝 `src/native/CommandRunner.ts` (+18 -1) 📝 `src/native/NativeMacOSProvider.ts` (+1 -0) 📝 `tests/boundary/providers/native/machoSliceReader.test.ts` (+22 -0) 📝 `tests/boundary/providers/native/nativeCommandTermination.test.ts` (+21 -0) 📝 `tests/boundary/providers/native/nativeProvider.test.ts` (+25 -0) </details> ### 📄 Description ## Problem Every supervised native command launch on macOS needs the Darwin process-ownership helper. When that helper can't be compiled, `XcrunCommandRunner` turned the typed `DarwinProcessOwnershipInspectionError` into a bare `io` failure and dropped its message. Causes include a relative `REA_DARWIN_SWIFTC_PATH`, `REA_DARWIN_SWIFTC_PATH` without an absolute `SDKROOT`, a missing compiler, or an SDK/compiler mismatch after a toolchain update. Users then saw messages that pointed nowhere near the cause: ```console $ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-artifact /bin/ls --format json … "code":"artifact_operation_failed", … "detail":"lipo could not enumerate universal Mach-O slices" $ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-macho /bin/ls --format json … "code":"execution_failure", "message":"… Run `rea doctor` …", "diagnostics":{"tool":"file","reason":"io","exit_code":null,…} ``` `rea doctor` doesn't report it either. The cause chain was `lipo: io` → `xcrun: io` → `DarwinProcessOwnershipInspectionError("Direct Darwin helper compilation requires absolute REA_DARWIN_SWIFTC_PATH and SDKROOT …")`. The SQLite and web-capture providers already show this message in `diagnostics.reason`. Only the native path lost it. ## Fix - `CommandRunner` classifies the typed inspection error once, at the launch boundary. It stores the message as `NativeCommandFailure.detail` and carries it through the `xcrun --find` resolution wrapper. - `NativeMacOSProvider` adds `detail` to the `ProviderAdapterError` diagnostics. - `MachOSliceArtifactReader` appends the detail to its "lipo could not enumerate" message. Failure reasons and categories are unchanged. When the helper is healthy, the output is byte-identical: `inspect-artifact /bin/ls` returns the same evidence ID before and after. After: ```console $ REA_DARWIN_SWIFTC_PATH=/usr/bin/swiftc rea inspect-artifact /bin/ls --format json … "detail":"lipo could not enumerate universal Mach-O slices: Direct Darwin helper compilation requires absolute REA_DARWIN_SWIFTC_PATH and SDKROOT paths selecting a compatible installed Swift compiler and macOS SDK." $ REA_DARWIN_SWIFTC_PATH=/nonexistent SDKROOT=/x rea inspect-macho /bin/ls --format json … "diagnostics":{…,"reason":"io",…,"detail":"macOS process ownership inspection requires the Apple Swift compiler via the selected Swift compiler (REA_DARWIN_SWIFTC_PATH) (compiler result: ENOENT): no compiler diagnostics were captured"} ``` ## Validation - Two new boundary tests use the existing `NativeCommandLauncher` seam: one on the runner (`nativeCommandTermination.test.ts`) and one on the slice reader (`machoSliceReader.test.ts`). Both fail on main and pass with this change. The production ownership host is a process-wide singleton built from `process.env` at import time, so it can't be injected per test. - `npm run test:focused` passed for those two files plus `nativeProvider.test.ts` and `MachOSliceArtifactReader.test.ts` (45 tests). `npm run check:fast` and `format:check` also passed. - I ran the real CLI on macOS with Command Line Tools and covered three cases: a relative compiler path, a missing `SDKROOT`, and a nonexistent compiler. ## Remaining `rea doctor` still doesn't check whether the ownership helper compiles, so its remediation can't point at this. That needs its own doctor check, which I left out of this PR. 🤖 Generated with [Claude Code](https://claude.com/claude-code)  ## Summary by CodeRabbit * **Bug Fixes** * Improved diagnostics for Mach-O slice enumeration and native command failures by including relevant system error details when available. * Clarified when a required macOS process-inspection helper is unavailable.  --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-31 01:05:26 +00:00

closed this issue
added the
pull-request
label

bsakaguchi referenced this issue

2026-11-01 21:43:34 +00:00

[PR #5283] feat(mcp): hold several analysis targets open concurrently #6855

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#5283

No description provided.

Rows
Columns