Bram Stolk

GPG v2.0 still gives 'null' descriptions for rooms.

This still happens with gpg v 2.3 and com.google.android.gms 11.0.1 From a HUAWEI JDN-W09 device running Android 6.0 and Play Games version 3.9.08 ``` I/swaag ( 6797): Succesfully showed room...

Comparing the logs of a working match versus a disconnected match, I see that the first sign of trouble is actually: **I/RoomServiceStateMachine( 8563): Attempting to reconnect to: p_CL2Ww9b4x7XqKxAB** Is it...

Thanks. Any thoughts on 16 bit integers? Would that be in anyway compatible with the approaches taken in fpzip, or is that fundamentally different?

Thank you. I was under the impression that fpzip was able to cope with the 3D nature of my data, unlike PNG, ZH4, ZSTD, etc? In my volume, there is...

Maybe OP's example was not a good one, but it is a valid point that overestimating the remaining cost can cause suboptimal plans. Personally, I don't worry much about those...

Thanks for the suggestion, I tried doubling ```g_Settings.messageQueueSizeInBytes``` which did not help. Then I tried doublng ```g_Settings.maxNbMessagesPerUpdate``` to 20, which helped a bit. Still missing measurement, but it seems to...

Quadrupling queuesize, quadrupling message-per-update and halving sleep-between-updates, the situation gets better again, but still data is missing. I can see this because I know there should be a 100 render_scanlines...

I think it's possible to avoid losing data without blocking threads. If every thread writes to its own queue, instead of a shared queue, there would be no race conditions,...

Thanks. It indeed shows that rmtMessageQueue_AllocMessage() fails, shortly after start. Sure, you can always overflow the queue. But I think it's possible to handle this better. Once this happens, I...