test(server): pin git config for fixtures and compare native realpaths

[juliusmarminge]

Git for Windows installs with core.autocrlf=true, so every repository the
suite creates checked committed LF files out as CRLF and byte-exact content
assertions in GitManager, CheckpointReactor, and the orchestration engine
integration test compared 'v2\r\n' against 'v2\n'. A vitest setup file
now pins autocrlf, filemode, longpaths, gpgsign, and the default branch
through GIT_CONFIG_* so every git child the driver spawns sees them, ahead
of whatever the host's ~/.gitconfig says.

Two tests compared an Effect realPath of os.tmpdir() against git's own
report of the same directory; on GitHub's Windows runners the temp dir is
spelled with its 8.3 short name (RUNNER~1) and only realpathSync.native
expands it. The newline-in-worktree-path test is skipped on Windows, where
NTFS rejects the name outright.

Part of the Windows test-suite stack (#9577); see #9538 for the Windows lane these are verified against.

Model: Claude Fable 5 in Claude Code.

🤖 Generated with Claude Code

Note

Pin Git config for test fixtures and switch path comparisons to native realpath

• Adds gitConfig.setup.ts to inject fixed Git defaults (no autocrlf, no signing, no filemode, longpaths, main branch) into the test environment, and loads it via vite.config.ts before server tests run
• Replaces Effect FileSystem realpath calls with Node's native realpath in RepositoryIdentityResolver and GitVcsDriverCore tests so path comparisons match platform behavior, including Windows short-name canonicalization
• Skips the newline-path worktree listing test on win32 because NTFS disallows newline characters in file names
• Behavioral Change: Git subprocesses spawned by server tests now inherit the pinned config entries; path equality assertions use native realpath rather than Effect's implementation

^{Macroscope summarized 217c4da.}

---

Note

Low Risk
Test-only changes (Vitest setup, assertion helpers, and a platform skip); no production server or Git driver behavior changes.

Overview
Stabilizes the server test suite on Windows and against host-specific Git settings.

Pinned Git for all server tests: A Vitest setup file sets GIT_CONFIG_* environment variables so every spawned git process gets fixed defaults (core.autocrlf=false, no GPG signing, init.defaultBranch=main, etc.), overriding ~/.gitconfig and Git for Windows defaults that were breaking byte-exact assertions.

Path assertions: Git-related tests now compare paths with NodeFS.realpathSync.native instead of Effect FileSystem.realPath, so temp directories on Windows match Git’s long-path reporting when runners use 8.3 short names.

Windows-only skip: The worktree test that uses a newline in the directory name is skipped on win32 because NTFS cannot create that path.

^{Reviewed by Cursor Bugbot for commit 217c4da. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

ℹ️ No successful main baseline artifact is available yet. This run establishes the initial measurement.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	—	13.7 KiB	—	15.1 KiB	✅
Codex	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Codex	Live turn WebSocket wire	—	6.7 KiB	—	7.8 KiB	✅
Codex	Live turn WebSocket decoded	—	58.5 KiB	—	66.4 KiB	✅
Codex	Live turn messages	—	10	—	21	✅
Claude	Total thread wire	—	13.6 KiB	—	15.1 KiB	✅
Claude	Thread snapshot wire	—	7.0 KiB	—	7.3 KiB	✅
Claude	Live turn WebSocket wire	—	6.6 KiB	—	7.8 KiB	✅
Claude	Live turn WebSocket decoded	—	57.9 KiB	—	66.4 KiB	✅
Claude	Live turn messages	—	10	—	21	✅

Baseline: unavailable · PR result: 217c4da · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 113.8 KiB
• Claude decoded thread snapshot: 114.5 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.

[macroscopeapp]

Approvability

Verdict: Not approved

Macroscope's review found this PR not approvable — The changes are limited to test portability and deterministic Git fixture setup, with no product-runtime behavior change. However, both test files add file-level directives disabling the nodeBuiltinImport static-analysis diagnostic, which requires human review.

^{You can add or adjust custom eligibility rules. Learn more.}