dangernoodle-io / breadboard / 31138018399

97%

push

github

refactor: collapse bb_http_register_assets/_with_fallback ladder onto a cfg struct (B1-1458) (#1247)

bb_http_register_assets() now takes a single bb_http_register_assets_cfg_t (assets/n/fallback), same idiom as B1-1438's bb_http_serialize_stream_compose_cfg_t; bb_http_register_assets_with_fallback() is retired, no second entry point survives.

All three platform backends (host, espidf, arduino) migrated to the single signature. The espidf backend's old base-only "assets != NULL always required" check is dropped -- it was a divergence from the host backend (which never had it) and from the wide variant's own already-documented contract (assets == NULL legal iff n == 0); no in-tree caller passes assets == NULL, so this closes a latent host/espidf inconsistency without changing any real caller's behavior.

Both cross-component callers verified: platform/espidf/bb_wifi_prov/bb_wifi_prov.c:389 (real fallback) and examples/smoke/main/smoke_app.c:214 (no fallback -- omitted from cfg, zero-init reproduces the pre-collapse base function's exact behavior) -- the latter wasn't in the ticket's known inventory (only bb_wifi_prov.c was), a repeat of the earlier wide-variant-callers-only undercount pattern this epic keeps hitting.

Host tests renamed off the _with_fallback suffix and extended: fallback omitted (zero-trap pin), NULL-cfg BB_ERR_INVALID_ARG (new), and the closed NULL-assets/n==0/no-fallback divergence now proven OK on host.

No coverage baseline relocation.

Claude-Session: https://claude.ai/code/session_0119nuE2NZME9y2aYEVJTPB5

Co-authored-by: Claude Opus 5 <noreply@anthropic.com>

8542 of 9082 branches covered (94.05%)

Branch coverage included in aggregate %.

8 of 8 new or added lines in 1 file covered. (100.0%)

14278 of 14433 relevant lines covered (98.93%)

2601.37 hits per line