[MPlayer-dev-eng] [RFC] Some modest proposals

[Benjamin Zores]

On Sun, Jun 13, 2010 at 2:17 PM, Reimar Döffinger
<Reimar.Doeffinger at gmx.de> wrote:
> On Sun, Jun 13, 2010 at 02:03:23PM +0200, Dan Oscarsson wrote:
>> Many people think that now and then, you have to scrap most of the old
>> code and start again. mplayer shows it age - a rewrite of the core could
>> give a new start.
>
> There was one try of that (called G2).
> But a rewrite of the core honestly would probably result in a player
> that for one or more years would be broken for quite a few cases.
> Also you can find a "rewritten core" e.g. in ffplay, which uses threads
> etc., and I am sure Michael would be quite happy for it to gain more
> functionality.
> Not saying that's the only way to do it, but I expect it to have a higher
> probability of survival than just throwing away everything there is and
> start from scratch.

I fully agree with that, you want a rewrite of MPlayer (hell yeah it
would be needed), that's "simple":
- port libmpcodecs leftovers to lavc
- port libmpdemux to lavf
- port all af/vf to libavfilter
- create a libvo/libao in ffmpeg (that actually might be the first
point to work on)
- extend ffplay

This way MPlayer just becomes a perfect implementation of the ffmpeg libraries.
Working on a G2-like approach will only lead to one more media player
attempt in the wild and multimedia in Linux (as good as it may be) as
imho already suffered too much of things not being factorized and
reinvented (MPlayer/xine/gst/vlc each have already reinvented their
own lavf equivalent for various reasons over the time).

Ben

-- 
"My life, and by extension everyone else's is meaningless."
Bender, Futurama