hubcio opened a new pull request, #3925:
URL: https://github.com/apache/iggy/pull/3925

   Two classes of warning were accumulating unseen. The doctest
   leg ran cargo doc with no gate and stayed green through 69
   rustdoc diagnostics across 20 crates; broken intra-doc links
   are the costly ones, since they render as plain text and the
   docs quietly lose their navigation. Separately, the clippy
   leg denies warnings but runs with --all-features while the
   test legs build at default features, so a warning that only
   shows up when a feature is off was invisible - that is how
   four dead_code findings in the integration test binary sat
   in CI output.
   
   RUSTDOCFLAGS closes the first and is cache-safe: it enters
   the fingerprint of doc units only, never lib or bin units.
   
   For the second the Cargo lints table beat RUSTFLAGS=-D
   warnings. RUSTFLAGS enters the fingerprint of every unit
   including registry dependencies, so it forces a full rebuild
   of the dep tree and would have to be mirrored into
   coverage-baseline, the sole writer of the shared cargo
   cache. Measured both ways here: the lints table rebuilt 50
   workspace crates and zero registry crates.
   
   Eleven crates keep a local [lints.rust] block because cargo
   refuses to merge an inherited table with the [lints.clippy]
   they already carry. That makes their nursery = "warn"
   effectively deny, since rustc resolves the warnings group at
   emission time and catches anything left at warn whatever the
   flag order. CI behaviour is unchanged: the clippy leg
   already appended -D warnings.
   
   Most rustdoc fixes drop the link brackets where the target
   is private and rustdoc has nothing to point at under
   --no-deps. Widening visibility to satisfy a doc comment is
   backwards, and --document-private-items would publish the
   internals of every crate.
   
   The dead ci-qemu feature goes with it; nothing has
   referenced it since the monorepo restructure in 8db64db94.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to