Alan-Jowett / sonde / 23456196941

82%

push

github

test(modem): add RESET/advertising state and boundary preservation tests (#407)

* test(modem): add RESET/advertising and boundary preservation tests

Cover three validation gaps from issue #360:

1. MD-0300: RESET clears peer table — sends frames to register peers,
   then verifies \eset_state()\ empties them. MockRadio now tracks
   peers in \send()\ and clears them in \eset_state()\.

2. MD-0407/MD-0412: RESET tears down active BLE session — enables BLE,
   establishes a connection with pending indications and stale events,
   sends RESET, then asserts: advertising off, no stale \BLE_RECV\ or
   \BLE_PAIRING_CONFIRM\ leaks to USB, indication queue cleared.
   MockBle \disable()\ now clears event queue and indicated list to
   match the real \EspBleDriver::disable()\ contract.

3. MD-0401: BLE relay 1:1 message boundary preservation — injects 5
   rapid GATT writes in one poll cycle and asserts exactly 5 separate
   \BLE_RECV\ messages with correct payloads.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: address review — correct test refs, clear BLE events on disable

- Update test references: T-0628 (peer table), T-0633 (BLE RESET),
  T-0630 (boundary preservation) per modem-validation.md
- Document \Ble::disable()\ contract: must clear pending events and
  indication queue so stale data does not leak after RESET
- Fix \EspBleDriver::disable()\ to also clear \s.events\ (MD-0412)
- Align MockBle comment with trait contract

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: prevent BLE_DISCONNECTED event loss on disable and guard against advertise-on-disconnect

- Move stale BLE event draining from Ble::disable() to bridge RESET
  handler so BLE_DISABLE can still deliver BLE_DISCONNECTED per
  modem-protocol.md §4.14
- Add second advertising stop after disconnect in EspBleDriver::disable()
  to counter NimBLE advertise_on_disconnect auto-restart (MD-0407/MD-0412)
- Update Ble trait doc to cl... (continued)

107 of 117 new or added lines in 1 file covered. (91.45%)

18511 of 21938 relevant lines covered (84.38%)

48.52 hits per line