[Brooooooklyn]

Requested by Brooklyn · project thread

Before: three compute_globbed_inputs unit tests in crates/vt/src/session/execute/glob.rs could not fail on any regression without a neighbouring test failing on it too.

After: those three tests are gone. Every behaviour they asserted is still asserted by another test in the same module. Only #[cfg(test)] code changes.

How: each removal was defended by mutating glob.rs with the most realistic regression the test could catch, running all vt lib tests, and reverting. The key mutations were re-run after the deletion.

Totals

• Files touched: 1 (test module only).
• Cases deleted: 3 of 79 vt lib tests (76 after). Only the vt lib suite was counted, not the whole workspace.
• Cases folded: 0.
• Net lines: −42 (+0 / −42).
• Runtime saved: about 16 ms, the summed duration of the deleted cases.

Removals

1. test_overlapping_positive_globs_deduplicates: the result is a BTreeMap, so a file matched by two globs can't be counted twice. Dropping the Entry::Vacant check and always inserting passes every test, this one included. The only regression it does catch, ignoring negative globs, also fails 5 other tests. Merging several positive globs stays covered by test_positive_with_negative_exclusion and test_multiple_positive_globs.
2. test_empty_positive_globs_returns_empty: the same call as test_negative_only_returns_empty without the negative glob, with the same assertion. Treating an empty positive set as ** fails both, and fails test_negative_only_returns_empty after the deletion.
3. test_sibling_positive_glob_matches_sibling_package: globs reaching this function are already workspace-root-relative, so "sibling package" means nothing to it. test_sibling_negative_glob_excludes_from_sibling asserts the same file (packages/shared/src/utils.ts) with a wider glob. Stripping a trailing /** fails both, and after the deletion still fails 2 tests.

Mutations (all in glob.rs, each reverted)

Change	Tests that failed
always insert (no Entry::Vacant check)	none
walk only the first positive glob	test_multiple_positive_globs, test_multiple_negative_globs
ignore negative globs	6 before, 5 after
skip patterns without *	test_file_hashes_change_with_content, test_multiple_positive_globs
Break instead of Continue on an already-hashed file	none (see below)
strip a trailing /**	3 before, 2 after
remove the positive_globs.is_empty() early return	none (the loop over an empty set does nothing)
empty positive globs become **	2 before, 1 after

Gap found, not fixed here: returning ControlFlow::Break instead of Continue for a file already hashed by an earlier pattern passes every vt lib test, on main too. With that regression, overlapping globs silently drop later files from the fingerprint. A test with overlapping globs plus a unique file that sorts after them would catch it.

What ran and what did not

• Ran on Linux at c6c91f96 and on this branch: cargo test -p vt --lib (0 failures before and after), the mutations above, cargo fmt --check, cargo clippy -p vt --tests -- -D warnings and git diff --check, all clean.
• Not run: workspace-wide just test and just lint, macOS and Windows, the e2e/snapshot suites, and vt_subprocess_test / vt_pty_terminal.

Open pull requests: none of the 57 open touches glob.rs.

Ages: all three tests were added in #104 on 2026-03-09. None was added for a bug, and none is in a permission or security suite.

To revert: git revert 23630f0c.

---

Generated by Claude Code

[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 2026-11-04T17:50:24.927305Z	23630f0	PR opened

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.

[github-actions]

fspy benchmark

linux

dynamic/launch             change  -0.46%  [ -5.75% ..  +4.11%]  overhead   +49.59%
dynamic/access             change  -0.01%  [ -1.20% ..  +0.65%]  overhead    +8.16%
dynamic/access-relative    change  -0.10%  [ -1.04% ..  +0.79%]  overhead   +48.97%
dynamic/access-contended   change  +0.00%  [ -2.16% ..  +2.45%]  overhead   +11.43%
static/launch              change  +0.12%  [ -6.86% ..  +7.27%]  overhead  +150.76%
static/access              change  +0.17%  [ -1.21% ..  +1.64%]  overhead  +815.84%
static/access-relative     change  -0.02%  [ -1.14% ..  +1.16%]  overhead +1329.30%
static/access-contended    change  -0.26%  [ -1.37% ..  +1.21%]  overhead +3484.40%

macos

dynamic/launch             change  -0.14%  [ -2.38% ..  +2.60%]  overhead  +222.30%
dynamic/access             change  -0.40%  [ -4.26% ..  +3.37%]  overhead    +2.93%
dynamic/access-relative    change  +0.11%  [ -1.67% ..  +2.48%]  overhead  +276.17%
dynamic/access-contended   change  +0.47%  [ -2.37% ..  +2.53%]  overhead    +3.71%

windows

dynamic/launch             change  -0.65%  [ -7.65% ..  +5.84%]  overhead   +27.97%
dynamic/access             change  +0.00%  [ -1.02% ..  +1.21%]  overhead    +1.64%
dynamic/access-relative    change  -0.20%  [ -1.01% ..  +0.61%]  overhead    +1.61%
dynamic/access-contended   change  +2.08%  [ -2.82% .. +15.12%]  overhead    +2.58%