Files
bambuddy/backend/tests/unit
maziggy 13183e365a fix(ams): clear empty AMS-HT slot via its real tray_exist_bits position (#2670)
After #2594 every empty-slot clearing path skipped AMS-HT units, so a removed
HT spool never cleared on the printer card while Bambu Studio correctly showed
Empty. Root cause: the HT presence bit is packed as a single consecutive bit in
tray_exist_bits at 16 + (ams_id - 128) (HT-A=16, HT-B=17, ...), not the regular
ams_id*4 position -- so the bitmask cleanup, which skipped id>=128, never
touched it. The HT's state field is firmware-variant (loaded reports 11 on H2D,
9 on the #2594 firmware) and it keeps echoing stale tray_type after removal, so
the bitmask is the only reliable, firmware-independent signal. Confirmed against
a live H2D capture (loaded=0x10f7f, empty=0xf7f) and the OrcaSlicer
DevFilaSystem.cpp reference (is_exists = tray_exist_bits >> (16 + (ams_id-128))).

- apply_tray_exist_bits: handle AMS-HT (128-135) at 16+(ams_id-128) instead of
  skipping it; clears an empty HT and, because a loaded HT keeps its bit set,
  never wrongly clears one (keeps the #2594 fix intact). Ids outside the known
  regular (0-15) and HT (128-135) ranges are left untouched rather than guessed.
- Build the AMS change-hash from the merged state, not the raw payload, so a
  removal signalled only by the bitmask (firmware still echoing tray_type) still
  flips the hash and fires on_ams_change to unbind the spool_assignment row.
- Emit the exists presence bit in the websocket status serializer (REST already
  did) so the card renders "Empty" instead of "?" where state is ambiguous.
2026-07-27 10:34:56 +02:00
..