dangernoodle-io / breadboard / 26001579243

100%

push

github

feat(bb_event_ring): retained-flag + bb_update_check publishes initial snapshot at init (#255)

Closes the cold-boot subscriber gap for the update.available SSE topic.

Problem: when a device cold-boots, bb_update_check's first manifest fetch
fires after CONFIG_BB_UPDATE_CHECK_INTERVAL_S (300s default). Until then
no one has posted to update.available, so its ring is empty. SSE clients
connecting in that window receive no replay and see no state.

Three coupled changes in one concern:

1. bb_event_ring_attach_ex(topic, capacity, max_entry, retained, out):
   new preferred form of bb_event_ring_attach. The retained flag marks a
   topic as "state, not events" — documents intent at the call site and
   reserves API surface for future drain semantics. bb_event_ring_attach
   becomes a thin wrapper calling _ex with retained=false (no behavioral
   change for existing callers).

2. bb_event_routes_attach_ex(topic_name, retained): mirrors the ring _ex
   variant at the routes layer, passing retained through to bb_event_ring_attach_ex.
   bb_event_routes_attach wraps _ex(name, false).

3. bb_update_check_init now unconditionally publishes an initial snapshot
   (available=false, current=<running version>, latest="", last_check_ok=false)
   immediately after registering the update.available topic. This ensures
   the ring is non-empty from T=0 so any SSE client connecting before the
   first periodic check receives the last known value rather than empty state.

The auto-attach call in bb_update_check_espidf.c is updated to use
bb_event_routes_attach_ex("update.available", true). The explicit error log
added in #254 is preserved.

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

956 of 956 branches covered (100.0%)

Branch coverage included in aggregate %.

11 of 11 new or added lines in 3 files covered. (100.0%)

1709 of 1709 relevant lines covered (100.0%)

1021.99 hits per line