prisma-risk / tsoracle / 26213504134

95%

push

github

chore(lints): warn on unwrap/expect in non-test code (#28)

Each library/binary crate root now carries

    #![cfg_attr(not(test), warn(clippy::unwrap_used, clippy::expect_used))]

so CI's `cargo clippy ... -- -D warnings` blocks any new `.unwrap()` /
`.expect()` in `src/`. The `cfg_attr(not(test), ...)` form scopes the lint to
non-test builds of each crate, and inner attributes don't propagate to
`tests/`, `examples/`, or `benches/` — those are separate compilation units.
This sidesteps two upstream clippy limitations
(rust-lang/rust-clippy#9612 and rust-lang/rust-clippy#13981) where the
`clippy.toml` `allow-{unwrap,expect}-in-tests` setting fails to cover helper
fns inside test modules and code in the `tests/` directory.

The six pre-existing runtime `.expect(...)` callsites get
`#[expect(clippy::expect_used, reason = "...")]` markers explaining the
invariant and linking to the relevant tracking issue:

  - allocator convenience wrappers (#4): on_leadership_gained,
    prepare_window_extension, commit_window_extension
  - client driver_task first_arrival (#8)
  - server leader-hint metadata key (#5)
  - server reflection FILE_DESCRIPTOR_SET (#9)

The two `openraft-toolkit::log_store` cf_handle helpers carry inline
invariant rationale since the column families are validated at open.

CONTRIBUTING.md gets a new "Panic policy" section documenting the mechanism,
the upstream clippy bugs we sidestep, and the convention for adding new
`#[expect]` exemptions (explain invariant + link tracking issue).

Closes #6.

4 of 4 new or added lines in 1 file covered. (100.0%)

2429 of 2598 relevant lines covered (93.49%)

201361.19 hits per line