[gst-devel] DVD Navigation progress

[Jan Schmidt]

On Sat, 2004-01-24 at 07:15, David Schleef wrote:
> On Fri, Jan 23, 2004 at 02:20:31PM +0100, Ronald Bultje wrote:
> > On Fri, 2004-01-23 at 12:44, Benjamin Otte wrote:
> > > > demux.audio_00 \
> > > > demux.audio_00 - switch.sink
> > > > demux.audio_00 /
> > > Pads have to have different names or some get_by_name functions will
> > > break.
> > [..]
> > > I think we shouldn't try to do it implicitly by doing something to
> > > padnames or whatever, but do it explicitly by adding a GstPadGroup or I
> > > don't know what.
> > > Everytime you make an implicit assumption, god kills a kitten.
> > 
> > You're right. We'll make a padgroup then, sounds equally fine for me.
> 
> What about adding a ghost pad to mpegdemux?  The ghost pad
> "mpegdemux.default_audio" points to the audio pad chosen by
> upstream events (and/or property settings).

This is very close to what I mentioned in IRC the other day - except
that I had it as a real 'ALWAYS' pad, not a ghost pad, and I had the
changeover being done by adding a 'streamid' property to the pad's caps.
When mpegdemux switched the stream, it would change that property on the
caps, possibly send a DISCONT, and start sending a copy of the data from
new stream instead. 

> Right now, ghost pads don't exactly have the best behavior for
> this.  Ghost pads currently can't be pointed to a different
> target pad while they're linked.

How do buffers work on a pad that is connected both as a ghost and as a
real pad? I can imagine a case where someone would want to follow the
'default' stream by connecting to the ghost pad, as well as connecting
to (for example) the audio_01 to save off the spanish soundtrack. 

J.

-- 
Jan Schmidt                                  thaytan at mad.scientist.com

"Computer games don't affect kids; I mean if Pac-Man affected us as
kids, we'd all be running around in darkened rooms, munching magic pills
and listening to repetitive electronic music." - Unknown