fix: Race condition in the NapariLiveWidget for the new alignment feature

[hongquanli]

Summary

• Fixes race condition where NapariLiveWidget sends duplicate MCU illumination commands during acquisition mode switches, causing command timeouts

Problem

During multi-point acquisition, when the worker thread switches microscope modes, a race condition occurs:

1. Worker emits signal_current_configuration(config) (Qt signal to GUI)
2. Worker calls set_microscope_mode(config) → sends SET_ILLUMINATION cmd
3. Worker calls turn_on_illumination() → sends TURN_ON_ILLUMINATION cmd (e.g., cmd 32)
4. GUI thread (delayed) processes signal → napariLiveWidget.update_ui_for_mode(config)
5. GUI updates slider → triggers valueChanged → update_config_illumination_intensity()
6. BUG: GUI calls update_illumination() → sends duplicate SET_ILLUMINATION cmd (e.g., cmd 33)

With two commands sent nearly simultaneously:

• _cmd_id becomes 33 (the last one assigned)
• MCU sends ACKs for cmd 32, but system expects cmd 33
• Cmd 33's ACK is never matched → timeout after 3.5 seconds

From the log:

18:10:50.645 - Worker: >>> sending command 32, type=TURN_ON_ILLUMINATION
18:10:50.645 - GUI:    >>> sending command 33, type=SET_ILLUMINATION  ← DUPLICATE!
18:10:50.649 - [MCU] !!! received ack for command 32, but waiting for command 33
... (repeated for 3.5 seconds) ...
18:10:54.182 - [MCU] !!! Command 33 ABORTED: Command timed out

Solution

LiveControlWidget already has an is_switching_mode guard to prevent this. NapariLiveWidget was missing it.

Added the same guard pattern to NapariLiveWidget:

• Added is_switching_mode flag to constructor
• Wrapped update_ui_for_mode() with try/finally to set/clear the flag
• Added early return in update_config_illumination_intensity(), update_config_exposure_time(), and update_config_analog_gain() when flag is set

🤖 Generated with Claude Code