dangernoodle-io / breadboard / 27521784682

100%

push

github

refactor(bb_event_ring): use shared bb_ring for storage (B1-285 consolidation) (#436)

- add bb_ring_peek_at(r, index, ...) — non-destructive indexed read (0=oldest) to
  bb_ring.h / bb_ring.c; 10 new unit tests covering null guards, index bounds,
  wrap-around, idempotent multi-read, null-buf metadata probe
- migrate bb_event_ring internal storage (entries[]+payload_buf[], head/tail/count,
  evict-oldest) to bb_ring_t; all entry i/o goes through bb_ring_push / bb_ring_peek_at
- replay (subscribe_with_replay) uses bb_ring_peek_at — non-destructive,
  multiple concurrent subscribers each replay the full ring independently
- bb_event_ring_set_allocator now forwards to bb_ring_set_allocator AND a local
  s_snap_calloc used for snapshot temp buffers, preserving existing alloc-injection
  test contracts (alloc-fail indices 0/1/2 map to struct/ring-struct/entries as before)
- bb_event_ring_spiram.c neutralized to a comment-only stub — SPIRAM preference for
  event-ring storage is now handled entirely by bb_ring_espidf (bb_ring_set_allocator)
  to avoid double-registration
- bb_event_ring CMakeLists: add bb_ring to REQUIRES; bb_event_ring_espidf CMakeLists:
  drop bb_registry PRIV_REQUIRES (no more EARLY registration from spiram.c)
- bb_log_stream NOT migrated: uses FreeRTOS xRingbuffer (RINGBUF_TYPE_NOSPLIT)
  as a raw byte stream drained in arbitrary chunks — no entry boundaries,
  incompatible with bb_ring's fixed max_entry_bytes model
- all 2178 existing tests pass; 100% line + branch coverage maintained

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

1659 of 1659 branches covered (100.0%)

Branch coverage included in aggregate %.

54 of 54 new or added lines in 1 file covered. (100.0%)

2771 of 2771 relevant lines covered (100.0%)

1151.12 hits per line