Specify sleep timing method for the websocket #7723
Conversation
Specify sleep timing method for the websocket's keep-alive sender, based on whether GCM is disabled. Fixes signalapp#6644 // FREEBIE
|
To make sure it stays linked in the future, the accompanying PR is: |
|
Just a reminder that this PR needs to be merged alongside signalapp/libsignal-service-java#53. |
|
Ping! |
|
@greyson-signal As signalapp/libsignal-service-java@c37cbf5 got merged I think this one should also be added, as I'm suffering from delayed message delivery, too. |
|
@dpapavas Is there anything I could to to improve the laggy message receive situation without this patch? Since upgrading to Oreo / LineageOS 15.1 it seems I have lags as big as 3 to 5 hours sometimes :-/. |
|
Well, you can wake the device from time to time, by turning on the screen. This would allow the various blocking sleeps to time out and then, depending on how sockets are handled by your device, either it will detect that your connection to the server is dead and respawn a socket and get any pending messages/receipts, or it might still take some time for that to happen. In the latter case force quitting and restarting Signal should help. |
|
Ping! |
|
I don't understand what prevents a merge of this pull, if the server-side bits have been mergerd already... |
|
Apologies, this got lost while I was dealing with notification issues. I've made this part of #8204, since I had to upgrade the libsignal-service version for a separate change. Thank you! |
|
@greyson-signal Thank you so much, now I would love to see a new WebSocket release on the website in order to check this out 👍. |
|
Thanks for merging this! |
Specify sleep timing method for the websocket's keep-alive sender,
based on whether GCM is disabled.
Fixes #6644
// FREEBIE
Contributor checklist
Fixes #1234syntaxDescription
See its accompanying PR signalapp/libsignal-service-java#53 for more details. This might be related to and partially or completely fix #7638, but the supplied log doesn't allow me to confirm this and the discussion makes me suspect it might alse be related to a different problem. I'm therefore didn't mention it in the commit message.