Victor Gaydov

OK, so I was able to fix the problem, but I still don't fully understand it. Log reports were done each 5 seconds, and they were clocked by CPU clock....

Probably it is important that log reports were clocked by timer on same CPU that clocked sender (it was sending a file via localhost). So basically we were sampling latency...

More thoughts about this case. - receiver creates session when it receives first packet - on session creation, receiver also starts log reporting timer, based on cpu clock - sender...

There are no visible bursts in packet receiving time. (X is packet number, Y is recv time) 

There are bursts in frame read time, triggered by PulseAudio. X is frame number, Y is read time.  i.e., PulseAudio blocks us for a while, then unblocks and allows...

Hi, thanks for report! Is it possible to get output of `roc-test-sndio -v -t` ? This would give more verbose logs of the test. If not, do you have any...

Thanks for PR and for catching the bug! I suggest a bit different fix for the issue. We can make that extra chunk an implementation detail - if user requested...

> Sounds good, made these changes in [da43de9](https://github.com/roc-streaming/roc-toolkit/pull/743/commits/da43de9c7cae6fe0f06a6e364920271e478ffbe8). Also, do we want EmbeddedCapacity to represent (a), the max queue capacity that we allow for using data embedding (in this case...

I guess we should do something like: ``` AlignedStorage embedded_data_; ``` ?

Thank you! Small improvement: 81b2235b8b1971885947310ce7abf052ae78bc35