glm-5.2
adbd57da46
- R-01: remove `readme = "README.md"` from Cargo.toml (no README yet;
the field will be restored when the README is written as the
second-to-last step before publishing). `cargo publish --dry-run
--allow-dirty` now succeeds.
- R-02: replace the three `expect` calls in non-test library code per
AGENTS.md §2:
- `channels/operations.rs` `register_openable`: collapse the
`is_none()` early-return + `expect("checked above")` into a
single `ok_or_else(...)?` (same error message, no expect).
- `core/types.rs` `StreamBidiStreamSource::accept_bi` and
`close`: use `unwrap_or_else(|e| e.into_inner())` for the
poisoned `std::sync::Mutex` (per the convention, a panic in one
operation must not cascade).
- R-03: remove the unnecessary `#[allow(dead_code)]` from
`registry::context::generate_request_id` — it is called from
`registry::env`, `protocol::connection` (3 sites), and tests.
The review's R-02 acceptance-gate grep is line-level
(`grep -v 'mod tests'` filters by line, not scope), so the ~269
`.expect(` calls inside `#[cfg(test)] mod tests` blocks still match
it. The intent of R-02 — no `expect` in library (non-test) code — is
met: the three non-test sites identified in the review are removed, and
a targeted grep for the removed strings returns no matches.
Verification:
cargo test → 483 passed, 0 failed
cargo clippy --all-targets -- -D warnings → clean
cargo fmt --check → clean
cargo doc --no-deps → clean (0 warnings)
cargo publish --dry-run --allow-dirty → succeeds