docs: drop stale "marker probe" wording in lib_bin_collision test - #63
cargo-affected-bot wants to merge 2 commits into
Conversation
Resolution moved to NEXTEST_BINARY_ID; the shim no longer probes binaries (src/shim.rs: "no marker probe needed"). Reword the assert message and the second-collect comment to match, consistent with the file's own header. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
# Conflicts: # tests/functional/lib_bin_collision.rs
cargo-affected-bot
left a comment
cargo-affected-bot
left a comment
There was a problem hiding this comment.
This PR no longer carries a diff — gh pr diff 63 and git diff origin/main...HEAD are both empty, and the merged tree at 1685b80 is byte-identical to main.
Both strings the branch reworded were deleted upstream before the merge. The basename fallback ambiguous assertion was replaced by the produced no coverage check in #77, and the second-collect comment was rewritten in the same sweep — it now reads "Neither half of that attribution goes through a path probe any more", which already says what this branch set out to say. The conflict-resolution merge taking main's side was the right call: nothing the branch held is still missing from main.
The two surviving marker probe mentions are accurate and should stay. tests/functional/lib_bin_collision.rs's header describes the pre-fix bug in past tense ("used to bail with basename fallback ambiguous — marker probe matched 0 of them"), and src/shim.rs says "no marker probe needed" — both are statements about the mechanism that was removed, not claims that it still exists.
Closing as superseded: there is nothing left here to merge, and the branch keeps drawing a full CI matrix on every base merge.
1 participant
The top-of-file doc comment already records that lib/bin basename collisions are resolved by reading
NEXTEST_BINARY_IDfrom the env — "the shim no longer needs to map paths or probe binaries" (andsrc/shim.rsstates "no marker probe needed"). Two later references in the same test still describe the removed mechanism:basename fallback ambiguousassertion message said "marker probe must disambiguate lib+bin" — there is no marker probe anymore; the disambiguation isNEXTEST_BINARY_ID.collectcomment said it drives "the same probe path again" — the stable path isbinary_idresolution, not a probe.This aligns the wording with the current code and the file's own header. The assertions are unchanged: they still guard against the legacy
failed to resolve binary_id/basename fallback ambiguouserror strings ever reappearing, so the regression coverage is intact.Comment/message-only change — no behavior change, so no new test. Verified with
cargo check --tests.