T5 code-quality reviewer flagged 3 NITs (severity NIT, non-blocking — green-lit T5 anyway). Filed here for batched cleanup before v0.7.x release.
1. crates/core/src/events.rs:72 — stale module docstring
Docstring says "adds two domain events emitted by the writer / use-case layer" — pre-existing staleness now compounded (5 transcription variants + 2 verify variants in addition to the original two). One-line trim.
2. crates/app/src/container.rs — test comment math is wrong-but-passing
`app_container_shares_events_across_use_cases` asserts `events_strong == 9` with the comment "1 (container) + 8 (one per UseCase)". The TranscriptionUseCase doesn't store events on the struct — the count holds because the spawned worker_loop task captures an Arc clone. Comment should read "1 container + 7 use-case fields + 1 transcription worker task".
3. crates/app/src/container.rs:325-346 — fallback path may not surface BackendUnavailable
The `unwrap_or_else` fallback constructs a TranscriptionUseCase with an empty registry. When `registry.active()` is called by the worker, it returns whatever `TranscriberRegistry::active` errors with on empty (likely a generic `RegistryError`), NOT `TranscriptionError::BackendUnavailable` as the WHY-block claims. Either (a) verify the error variant matches the claim, (b) update the WHY-block, or (c) wrap the fallback's `registry.active()` to lift the error to `BackendUnavailable` on Err.
Filed under ASR v2 milestone for batched cleanup.
T5 code-quality reviewer flagged 3 NITs (severity NIT, non-blocking — green-lit T5 anyway). Filed here for batched cleanup before v0.7.x release.
1. crates/core/src/events.rs:72 — stale module docstring
Docstring says "adds two domain events emitted by the writer / use-case layer" — pre-existing staleness now compounded (5 transcription variants + 2 verify variants in addition to the original two). One-line trim.
2. crates/app/src/container.rs — test comment math is wrong-but-passing
`app_container_shares_events_across_use_cases` asserts `events_strong == 9` with the comment "1 (container) + 8 (one per UseCase)". The TranscriptionUseCase doesn't store events on the struct — the count holds because the spawned worker_loop task captures an Arc clone. Comment should read "1 container + 7 use-case fields + 1 transcription worker task".
3. crates/app/src/container.rs:325-346 — fallback path may not surface BackendUnavailable
The `unwrap_or_else` fallback constructs a TranscriptionUseCase with an empty registry. When `registry.active()` is called by the worker, it returns whatever `TranscriberRegistry::active` errors with on empty (likely a generic `RegistryError`), NOT `TranscriptionError::BackendUnavailable` as the WHY-block claims. Either (a) verify the error variant matches the claim, (b) update the WHY-block, or (c) wrap the fallback's `registry.active()` to lift the error to `BackendUnavailable` on Err.
Filed under ASR v2 milestone for batched cleanup.