lugia19

While making the smaller code sample, I noticed a couple things. 1) Whether or not the bug occurs (and where in the file) is entirely random (my favorite). Sometimes it...

Okay, I did some more checking and logging. Of note, the file is seekable, according to SoundFile.seekable() In short, there are two cases where the bug happens (when it does...

I think it's definitely an issue with the fact that the SoundFile is created without having the entire file written to disk, which might be why no one has encountered...

That's probably a good idea. As for the bug where the read number of frames is zero (despite _check_frames reporting that there should be more frames to read), I suppose...

Sure, I'll make a pull request for the change later. As for the bug, I have done more testing and (after fixing a mistake I made when making the cut...

Now that I've actually narrowed it down to the real bug, here's a minimal code sample showcasing the issue. It'll happen with either a .wav file or a .mp3 without...

> As outlined above, we can handle it better by replacing the assert with a warning The assert issue is mostly secondary - the issue I'm running into now is...

The way I've worked around for now is by checking if the data I recieve is the expected length, and if it isn't, I recreate the SoundFile. By checking if...

Maybe I'm the one misunderstanding, I'll check again and let you know, I might've just gotten it wrong

So, after some more testing, I have the following to report. Note that I switched to read instead of buffer_read for these tests. To make a concrete example, I logged...