Originally written by @ara4n at https://github.com/element-hq/element-x-ios/issues/5817
Steps to reproduce
- Send messages at the same time that the timeline is coincidentally getting reset by SSS for whatever reason
- It then only shows the most recent message (and the date separator)
Here seen on macOS on build 1255 but the same applies on iOS:
There's no loading spinner (although overscrolling the screen kicks it into attempting a backpagination, which fixes it).
Outcome
What did you expect?
The timeline should never get stuck showing partial history.
What happened instead?
So, I stumbled across this while smoketesting the Tachyon DMLS work a few weeks ago via Claude: I fixed it on my experimental branch but haven't had a chance to try to upstream it. So before I forget it again, three possible related issues were thrown up. I don't the code well enough to unpick which of them (maybe all?) fix this failure mode - you can find them at https://gist.github.com/ara4n/d9462cca5c758c5b8f6a9536a1278c85.
- One fixes the actual race between timeline reset & echo:
- There's one which stops duplicate messages appearing if timeline backfill races with local echo:
- Finally, another fixes a similar problem where the timeline gets stuck if /messages ever returns empty:
@Hywan i'm hoping something useful can be salvaged from the claude output here if it'd be possible to take a look. I get the local-echo-race bug once every few days.
I can also provide logs if needed.
Application version
1255
Will you send logs?
No
Originally written by @ara4n at https://github.com/element-hq/element-x-ios/issues/5817
Steps to reproduce
Here seen on macOS on build 1255 but the same applies on iOS:
There's no loading spinner (although overscrolling the screen kicks it into attempting a backpagination, which fixes it).
Outcome
What did you expect?
The timeline should never get stuck showing partial history.
What happened instead?
So, I stumbled across this while smoketesting the Tachyon DMLS work a few weeks ago via Claude: I fixed it on my experimental branch but haven't had a chance to try to upstream it. So before I forget it again, three possible related issues were thrown up. I don't the code well enough to unpick which of them (maybe all?) fix this failure mode - you can find them at https://gist.github.com/ara4n/d9462cca5c758c5b8f6a9536a1278c85.
@Hywan i'm hoping something useful can be salvaged from the claude output here if it'd be possible to take a look. I get the local-echo-race bug once every few days.
I can also provide logs if needed.
Application version
1255
Will you send logs?
No