dangernoodle-io / breadboard / 30923224342

97%

push

github

refactor: replace fd/is_ws with a client handle in bb_data_http's send_fn seam (B1-1123 PR-1) (#1223)

Internal cleanup, not a transport-neutral seam yet: bb_data_http_client_t
is still a component-private HTTP struct (fd, is_ws, topic_filter,
outbound queue) and the typedef stays bb_data_http_send_fn. A future
MQTT/UDP backend still could not implement this signature -- it has no
fd/is_ws and no business depending on bb_data_http's private struct.
The HTTP-specificity moved from the parameter list into the type name;
it did not go away.

What this PR does: drops the socket-flavored int fd, bool is_ws pair
from the seam and replaces it with the real client handle, pushing fd
resolution into the ESP-IDF backend's own slot table (which already
stores the client pointer per slot, so the lookup swap is a straight
pointer-equality scan) and reading is_ws directly off the handle for
the WS-rejection check.

key was investigated per KB 1464's sketch but dropped for this PR: the
outbound queue's single send_fn call site can't distinguish a real
per-key frame's id from a synthetic dropped-marker frame's id=0
without threading new state through, so inventing/guessing a key would
be worse than omitting it.

Whether bb_data_http_client_t * (versus an opaque consumer handle) is
the right long-term seam type is an open question, deferred to PR-2 --
this shape is not settled.

8356 of 8918 branches covered (93.7%)

Branch coverage included in aggregate %.

4 of 4 new or added lines in 2 files covered. (100.0%)

14063 of 14218 relevant lines covered (98.91%)

2660.88 hits per line