It looks like the ffmpeg resampler implementation doesn't suffer from bug 47156, but the solution for leftover data doesn't seem entirely correct: if there is leftover data, it's not taken into account in pa_resampler_result(), pa_resampler_max_block_size() and pa_resampler_reset(). I originally started to write this bug report, because there seemed to be unnecessary copying of data going on in ffmpeg_resample(), but that's probably not the case, because the ffmpeg resampler handles only mono audio, so multichannel data has to be deinterleaved for processing and again interleaved after processing. If the input is mono to begin with, then there would be room for optimizing for that case still, though.
I'll include the leftover related fixes in the fix for bug 47156. I'm setting that bug as a dependency for this one. The mono optimization is still something that could be done.
-- GitLab Migration Automatic Message -- This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/issues/192.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.