[xine-devel] dxr3 support design issues

[Eduard H. <ed...@ao...>]

Hi.

Today I start a new thread, since I want to concentrate on implementing
the dxr3 plugin.

The good message:
*) I get dxr3 output with voice and video in sync.

But there are also bad ones:

*) Some changes in the dxr3 kernel drivers are needed
	Ok, this is a non-issue for this ML. But now I must
	also communicate with dxr3. Good that the guy developing
	this already knows me :)
	
*) I can only watch the movie from the beginning.
	Setting a new position with the slider does not work.
	Every time the the position is set, the metronom time begins
	with 0. The result is, that my plugin hangs.

The fact is, that I still get the true PTS values from the decoder
buf->PTS. I tried to convert this PTS with the
metronom->got_video_frame function, but this function does something
different.

A first analysis shows that the metronom->video_wrap_offset gets
the wrong value. Maybe due to the different call sequence.

Using software decoders:
1. metronom_got_audio_samples
2. metronom_got_video_frame

Now with dxr3_decoder:
1. metronom_got_video_frame
2. metronom_got_audio_samples

Some hints?

-- 
Eduard Hasenleithner
student of
Salzburg University of Applied Sciences and Technologies