[MPlayer-users] DVD Menu Support

[Rich Felker]

On Wed, Jun 08, 2005 at 05:34:39PM -0400, Jason Tackaberry wrote:
> Clearly I don't know as much about MPlayer as you do, but I keep reading
> on these lists things like "MPlayer's architecture doesn't allow that."
> I think it's good that the developers are realistic about what MPlayer
> can and can't do.  But this also leads me to believe that MPlayer
> development will eventually approach an impasse.

Yes and no. For playing movies, I don't think it will run into
problems. For advanced video processing (editing, temporal filtering,
proper encoding) however I think it will, unless we can manage to
radically redesign key layers. Some people are working on this, or at
least know what needs to be done, but it's a long ways away from
happening..

> G2 was supposed to be
> the solution to world hunger, and there was much excitement about it
> when it was first announced, but there's not been much progress on that
> front (that I've heard about).

IMO it's abandoned. Arpi says he's working on it in private but I
haven't seen any code, and the rest of us dropped working on the
public branch. I've moved my hopes to ffmpeg -- with libavcodec,
libavformat (which is finally transforming into something that doesn't
suck!) and the upcoming libavfilter (to be designed by yours truly :),
I hope we'll have a viable player without the limitations of MPlayer
or the bad performance and bugs of all the other competitors.

> So from where I stand, Gstreamer is looking very promising.  There is
> much work to be done yet, but that work is happening, and there is a
> sense of energy that is important for a project like that.  I think it's
> worth getting excited about.  In the mean time, I will continue to use,
> hack on, and be happy with mplayer.  But I'm not going to discount
> Gstreamer because right now it's slower.

My understanding is that the philosophy of GStreamer has always
suffered from the same flaws as most windows tools. Bad object
oriented overdesign, support for only a limited (and bad) collection
of colorspaces and image formats requiring a lot of unnecessary
conversions, and the inability to pass data efficiently from one step
of processing to the next. Perhaps the people working on it now
understand this, but with all the trouble I've been through trying to
design a proper architecture, I really can't see anyone who hasn't
suffered through the pitfalls of MPlayer being qualified to know the
ingredients needed in a good design...

Rich