Join GitHub today
GitHub is home to over 36 million developers working together to host and review code, manage projects, and build software together.Sign up
Have the room member tests retry a few times #631
These tests were pretty flaky in a multi-process homeserver (like Dendrite). The events took a few moments to persist, and thus a race condition was produced where SyTest would request the status right afterwards (before the events finished persisting).
We now retry a few times before calling it failed.
I'm not quite sure what I think about this. Tightlooping on get_room_state doesn't seem optimal, though I guess it's ok.
In other places I think we wait for the change to come down
FWIW it does have a backoff delay, so it doesn't hammer tooooo much
We do tend to use