dangernoodle-io / breadboard / 30734322875

97%

push

github

feat: declare binds_data= on the real bb_data_bind() composition markers (#1200)

B1-1355 PR2 -- declares binds_data= on smoke's bb_wire.h manifest
markers for every component whose own tree actually calls
bb_data_bind(), matching the real per-component call sites
(reboot, factory_reset+storage_delete, wifi, fan+power+thermal).
This makes check_binds_data_cap and the binds-data-mismatch lint
rule (both landed inert in prior PRs) live for smoke's codegen
composition only: examples/floor composes via handwire (decision
#724), so bbtool codegen never resolves any binders for it and
these checks never run against floor at all. Floor still consumes
bb_data slots at runtime (factory_reset/storage_delete via
bb_storage_http_routes_init()/
bb_storage_http_factory_reset_routes_init() in
examples/floor/main/floor_app.c, plus floor's own "log" bind) --
that usage shares the same finite 8-slot table with smoke's, with
no codegen-time guard covering it.

bb_log_event is left undeclared -- its only bb_data_bind() call
lives in examples/floor/main/floor_app.c, a composition root outside
the lint rule's per-component reach (verified its zero-call-site
branch fires without erroring). bb_ota_check/bb_diag/bb_display have
real bb_data_bind() call sites (bb_ota_check has two:
bb_ota_check_common.c:868 and bb_ota_check_wire.c:114) but carry no
// bbtool:init marker anywhere in-tree today (unwired/deliberately
out of composition scope per B1-1357/B1-1153), so there is nothing
to declare for them.

8220 of 8784 branches covered (93.58%)

Branch coverage included in aggregate %.

13914 of 14069 relevant lines covered (98.9%)

2663.17 hits per line