Fix GenType path resolution for Windows short paths and POSIX symlinks

[fhammerschmidt]

Summary

GenType could hang on Windows when the project root used an 8.3 short path while compiler source locations used its expanded form. The unmatched absolute path reached a drive root whose parent never changed. On POSIX, building a symlinked project from another working directory could also fail to write GenType output.

Pass the already canonical package path as -bs-project-root during builds, matching the parser's working directory without canonicalizing again for each file. Preserve the caller's project path in compiler-args, since editor incremental type checking compiles absolute source paths through that spelling. Stop resolver traversal whenever a path's parent is stable.

Validation

• A Windows integration test obtains an explicit 8.3 alias, skipping when short names are unavailable. A POSIX regression builds a symlinked project from another working directory. Both exercise cross-directory GenType module resolution and assert the exact generated import path.
• Both cases also invoke compiler-args and compile an absolute incremental source with the returned flags, as the editor does. The POSIX editor regression passes with this change and fails on the previous PR head.
• Added direct OCaml coverage for relative paths and Windows drive-root traversal. The test runner now runs the GenType unit cases on Windows, where the rest of the OUnit suite remains skipped.
• Full compiler tests: 268 passing unit cases, 591 runtime cases, 41 build fixtures, and 744 docstring cases. The Windows-specific unit case is skipped locally and runs in Windows CI.
• Both GenType integration suites and TypeScript checks pass.
• Rust unit tests and Clippy with warnings denied pass.
• Repository formatting and changelog validation pass.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-09T13:26:28.105380Z	a5a255d	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: af184c16b5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[codecov]

Codecov Report

❌ Patch coverage is 60.00000% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 80.50%. Comparing base (8d51b9c) to head (23fe47e).
⚠️ Report is 26 commits behind head on master.

Files with missing lines	Patch %	Lines
tests/ounit_tests/ounit_gentype_tests.ml	58.33%	5 Missing ⚠️
compiler/gentype/module_resolver.ml	66.66%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8639      +/-   ##
==========================================
+ Coverage   80.40%   80.50%   +0.09%     
==========================================
  Files         447      447              
  Lines       60990    61071      +81     
==========================================
+ Hits        49042    49167     +125     
+ Misses      11948    11904      -44     

Files with missing lines	Coverage Δ
compiler/gentype/module_resolver.ml	81.60% <66.66%> (-1.35%)	⬇️
tests/ounit_tests/ounit_gentype_tests.ml	82.75% <58.33%> (-17.25%)	⬇️

... and 37 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[pkg-pr-new]

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript@8639

@rescript/belt

npm i https://pkg.pr.new/@rescript/belt@8639

@rescript/darwin-arm64

npm i https://pkg.pr.new/@rescript/darwin-arm64@8639

@rescript/darwin-x64

npm i https://pkg.pr.new/@rescript/darwin-x64@8639

@rescript/linux-arm64

npm i https://pkg.pr.new/@rescript/linux-arm64@8639

@rescript/linux-x64

npm i https://pkg.pr.new/@rescript/linux-x64@8639

@rescript/runtime

npm i https://pkg.pr.new/@rescript/runtime@8639

@rescript/win32-x64

npm i https://pkg.pr.new/@rescript/win32-x64@8639

commit: 23fe47e

[github-actions]

Developer playground preview: https://rescript-lang.github.io/rescript/dev-playground/?version=pr-8639

[fhammerschmidt]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Chef's kiss.

Reviewed commit: 3102f42d77

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[fhammerschmidt]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Hooray!

Reviewed commit: a5a255da21

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[cknitt]

🤖 This review was generated by Claude (Claude Code) on behalf of @cknitt.

Verdict: approve with nits. No correctness bugs found. Making -bs-project-root canonical (matching package.path) is the right fix, and the dirname path = path guard makes path_to_list always terminate.

• Not Windows-only. The same root mismatch breaks rescript build <symlinked-dir> on POSIX. On master, building a symlinked project from another cwd fails with I/O error: /tmp/X-link//tmp/X/src-a/A.gen.tsx: No such file or directory; with this PR it produces from '../../src-a/A.gen'. Worth mentioning in the title/changelog.
• Test gap: on Linux/macOS, rewatch/tests/gentype/project-root.mjs passes with or without the Rust change, because the kernel's getcwd already resolves symlinks (including macOS /var → /private/var). Suggestions:
  • Anchor the assertion: /\.\.\/\.\.\/src-a\/A\.gen/ would also match a wrong ../../../../src-a/A.gen. Assert on from '../../src-a/A.gen' instead.
  • Add a POSIX case that symlinks the project and runs rescript build <symlink> from a different cwd.
  • The path_to_list guard has no test of its own; reverting it keeps everything green. A small OCaml unit test would be cheap.
• Minor:
  • The Windows test fails (notEqual) instead of skipping when 8.3 short names are disabled on the volume.
  • On a hang, the spawnSync 30s timeout kills cmd.exe but can leave rescript.exe/bsc.exe running, which could trip suite.sh's stale-process check.
  • In rescript compiler-args, -bs-project-root is now canonical while the parser args stay relative. On Windows, a consumer that gives bsc a non-canonical (e.g. 8.3) cwd could now mismatch. Probably rare; worth checking how the editor extension spawns bsc.
  • get_project_root_args canonicalizes once per compiled file; it could be computed once.

[fhammerschmidt]

23fe47e59 addresses the remaining substantive points. Builds now pass the already canonical package path as -bs-project-root, avoiding canonicalization per file. compiler-args preserves the caller’s root spelling: I reproduced editor incremental compilation failing through a POSIX symlink with the previous change, and the added regression passes now and fails on the previous PR head.

The title/changelog cover POSIX symlinks, the integration test builds through a symlink from another working directory and checks the exact generated import, and Windows skips the short-path case when 8.3 names are unavailable. Added direct drive-root traversal tests and enabled the GenType unit cases on Windows; the Windows compiler test step has passed.

Windows timeout process-tree cleanup remains deferred as optional test robustness work. Local compiler, GenType, Rust, formatting, and changelog checks pass. DCO now passes; the full CI and coverage runs are still in progress.