From: Thibaut M. <thi...@gm...> - 2005-04-21 09:01:56
|
Hi Michael, On 4/19/05, Michael Roitzsch <mr...@us...> wrote: > Hi Thibaut, >=20 > > yes, this one is ok. > > Anyway, i think that everything in HEAD is ok, except the ffmpeg merge > > which has not been tested yet. >=20 > I will take care of this. take a look here: http://www.xinehq.de/~tmattern/report/play_5_secs.test/play_5_secs.test.htm= l The merge has introduced new segfaults. I take this one : avi/dv/voxnews.dv, i think i've finally found what's the problem, it's easy to fix. > > IMHO, a good way to fix the End Of Stream detection problem (see hacks > > in video_decoder.c) would be add a new concept of "control frame" for > > the output loops (audio and video) the same way we have control > > buffers for decoder loops. This would allow to do some cleanup and > > really fix the problem in all cases including multi steams (which is > > not handled correctly). > > This scheme could even be extended (more control frames/buffers) to > > serialize calls to the video or audio driver. >=20 > I think this is exactly the way to go. It unifies the communication paths > between the components, so it should make things conceptually simpler. yes but hmm... this requires some work. Do you have time to think about tha= t ? > Michael >=20 > -- > The only difference between a car salesman and a computer salesman is > that the car salesman knows he's lying. Thibaut |