From ee336a6dd4e0af9537ca3a0541edb56c629abc89 Mon Sep 17 00:00:00 2001 From: cargo-affected-bot <282014906+cargo-affected-bot@users.noreply.github.com> Date: Fri, 5 Jun 2026 07:53:48 +0000 Subject: [PATCH] docs: fix stale collect.rs module summary for in-shim extraction #46 moved coverage extraction into the runner shim (spawn-and-wait + in-process extract, per-test result files read back by collect) and updated the module's Approach section, but left the top-of-file summary describing the old execvp + post-run walk/merge/export model. Rewrite the summary to match the Approach section and src/shim.rs. Co-Authored-By: Claude Opus 4.8 --- src/collect.rs | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/src/collect.rs b/src/collect.rs index 0913210..21d70dd 100644 --- a/src/collect.rs +++ b/src/collect.rs @@ -3,9 +3,12 @@ //! Delegates build and test execution to `cargo nextest run`. We insert a //! small runner shim (`cargo-affected runner-shim`) via //! `--config target..runner=[…]` that points `LLVM_PROFILE_FILE` at -//! a per-test subdirectory before `exec`ing the real test binary. After -//! nextest finishes we walk those subdirectories, merge profraws, export -//! coverage, and write per-(test, file) line ranges to SQLite. +//! a per-test subdirectory, then spawns and waits for the real test binary +//! and extracts that test's coverage in-process — merging profraws, +//! exporting with `llvm-cov`, and writing a small per-test result file — +//! before exiting. After nextest finishes we read those result files and +//! write per-(test, file) line ranges to SQLite. See the Approach section +//! below and [`crate::shim`] for the in-shim extraction details. //! //! We use the `--config` array form rather than the //! `CARGO_TARGET__RUNNER` env var because cargo only