Penny Wood

Hi, sorry I haven't looked at this for a very long time. I seem to remember providing a test case somewhere that demonstrated the issue when I submitted it as...

I haven't thought about this for well over a year, and I really can't remember much about it as I'm doing other things. It would seem from what you said...

[Test case here](https://github.com/sffjunkie/astral/blob/cecc6cce3b382e2e995e3ad34a7a2a31b64ef936/src/test/test_Issues.py#L71)

Compared to the horizon? The Sun rises earlier and sets later at altitude though.

I've come across the same issue as above. The forward and reverse calculations don't quite match up. I had a look at the maths but I don't know enough about...

The MQTT event stream *almost* does this. It sends the state changed events from one HA instance to another and allows these to display on the front-end. What it doesn't...

If we have successful service call transmission, then there's no reason why you can't run remote automations, it's more just that it wouldn't be the best idea. The main issue...

> The problem with contextvars is that we sometimes use create_task, which would cause the context to be lost. Actually `create_task` copies the current context into the new task, so...

Actually.... Looks like someone has already done the backporting [here](https://github.com/fantix/aiocontextvars). That implementation does have some limitations, but they're not enormous. It basically does what I was thinking of doing. Given...

OK I'll have one more crack at convincing you and then give up :grin: > I think that if we make service calls have mandatory context, This will work, but...