On 29 Feb 2008, nix_at_esperi.org.uk said:
> I just rolled back a short distance (to 16542, hardly any distance at
> all, and surely not covering anything that should affect this), and...
> now the problem's not happening.
>
> I hate multithreading bugs.

It's still happening.

I suspect it happens if you're within one buffer-length of the end of
the playlist. Instead of reading in all remaining tracks, it reads in
just the track curently playing... I can make it happen reproducibly if
the buffer empties when only two tracks are left in the playlist (the
one currently being played and thus partially buffered, and one more).

-- 
`The rest is a tale of post and counter-post.' --- Ian Rawlings
                                                   describes USENET