Vincent Petry

make sure to shift+refresh the page / conversations in question in case it's a caching issue please also check if the moment where this happens, whether there was a hash...

I'm currently busy with other stuff but will keep an eye. I think I also managed to reproduce this a few times locally.

I haven't seen it happening locally. Usually this happens when an important server call returned 404, which is interpreted as "not found". From your log above it seems the signaling...

In my local setup, when my computer is being slow, if I aggressively switch between two conversations it also ends up in a 404.

I'm now able to reproduce this from time to times using a JS script that randomly switches between two conversations while bots are posting messages there: ```javascript function simulateSwitching(tokens) {...

note: my test is with internal signaling, not HPB. with the simulator I was able to capture a 404. What I see is that during the switch a DELETE is...

As far as I can see and observe, there is already logic for detecting this and cancelling: https://github.com/nextcloud/spreed/blob/master/src/utils/signaling.js#L438 I added more logging locally and indeed I see that this works...

Some things that come to mind: - [ ] if calling joinRoom(token1) on a room that is pending leaveRoom(token1) leaving due to slowness, then cancel the leaveRoom call? (or provide...

A catch: As we can see, when `leaveRoom` is called first with the token, there's some async stuff happening until it can say "ok, we're done leaving and clearing the...

4. check if the token in the URL (or in the $store) is different than the one at the moment of the 404, then assume we already switched to the...