Daniel Edgy Edgecombe

To inject additional params into the timing middleware, you'll need to customise the timing middleware. I wanted to add a common id across different timing logs. I modified the add...

For clarity, I use `request.state.trace_id` elsewhere in the request lifetime to allow linking other logs to the timing logs.

This wouldn't work so well for modifying it with request based data. If you added the timer onto the request.state (as is done in the fastapi_utils.timing middleware). You could then...

I am seeing similar issues, where I have my db_session fixture that ensures unittest changes are rolled back after every test, this requires the app to use the same session...

I have a feeling that has worked, so I now get the same error in other fixtures that use the db_session, as its now connected to the TestClient loop. ```conftest.py...

With a recent merge, you should be able to do in a future release. ``` watson start --at 09:30 projectB watson start projectA ``` Would be simple enough to replicate...

``` $ watson start projectA --at "16:07" Starting project projectA at 16:07 $ watson interrupt projectB --at "16:10" Stopping project projectA, started 15 minutes ago and stopped 12 minutes ago....

@jmaupetit Yeah, I was still logged in on my current work account when I opened the PR. oops.

I have just tried switching to redis backend (as postgres backend seems to have memory leak issues due to asyncpg somewhere). The issue with the subscribe not closing is due...

It's not nice, but you can work around broadcast and maintain existing websocket behaviour like so. ``` async def ws_sender(websocket): exc = None async with broadcast.subscribe(channel=websocket.channel_name) as subscriber: try: async...