hmm, on Sat, Feb 08, 2014 at 01:02:36PM -0500, Brad Smith said that > On Sat, Feb 08, 2014 at 01:05:40AM -0700, Anthony J. Bentley wrote: > > On Fri, Jan 31, 2014 at 5:16 AM, frantisek holop <[email protected]> wrote: > > > it is me again, with a new release of mpv. > > > > Were the recent mplayer changes pushed upstream? Do they need to be > > pushed to mpv as well? > > http://www.openbsd.org/cgi-bin/cvsweb/ports/x11/mplayer/patches/patch-libao2_ao_sndio_c > > It'll require more than that. mpv is using the old sndio backend > before it was pushed upstream.
i have extracted the current mplayer source, applied the patches from ports and compared the sndio audio out part (libao2/ao_sndio.c) vs mpv's (audio/out/ao_sndio.c) i am a bit out of my depth here, but apart init() and some small bits i dont see a big difference in the API. Alexander's copyright is also from 2008 in both cases. if the demi-gods that did the big sndio migration back then for all the players are listening in, i would be most grateful if they had a tiny look what does mpv need to be a fully supported port... there is definitely at least one bug i am struggling with right now: https://github.com/mpv-player/mpv/issues/531 where according to the author of mpv the sndio backend is not reporting the position correctly: "... when the AO keeps returning a high value with get_delay(), the playloop will not think EOF is reached, even if the demuxer, decoder and audio filter are all out of data. get_delay is like the audio playback position of the AO, and if that is stuck, we can't know when audio actually ends." -f -- perfection is the road, not the destination.
