Reaction to errors in entity service calls

[xrad]

While developing an integration I have noticed that it is not trivial to properly communicate an issue in a service call to the frontend. Suppose there is switch entity in the server. The async_turn_on() method talks to some device using some protocol. On this level you may run into issues preventing you from completing the call.

1. Following the dev docs, I tried raising an HomeAssistantError. This causes the frontend to display a nice toast whose content I can control (good!) - but at the same time the service call is aborted. The frontend's optimistic flick of the switch sticks! (bad!)

2. If I do not raise an exception, and of course don't modify the state in the entity, the frontend's optimistic flick of the switch is reverted after about 2 seconds. This is the well known flip-flop handling.

This discussion is decidedly NOT about that flip-flop handling. There are other places where this was discussed. What I would like to see is the combination of 1 and 2. If I got a toast AND the switch flipping back it would be perfect!

There are of course also persistent_notifications, but they seem to be bit heavy-weighted for my use case.

So - is there a reason for the current behavior? Or is there a better way to implement this for my entities?

[xrad]

I noticed when I do this in my entity:

async def async_turn_on(self, **kwargs: Any) -> None:
   try:
      # talk to device
   except HomeAssistantError:
      self._attr_force_update = True
      self.async_write_ha_state()
      self._attr_force_update = False
      raise

then I get actually my intended result of having a snackbar and a synced UI - with the added benefit of an immediate sync instead of the 2 seconds.

So this seems to work fine, but I wonder if this is the intended operation and if this could affect automations in any way.

Any further insight will be highly appreciated!