On Thu, 25 Oct 2001, Alex Kanavin wrote:
> Or do you want me to comment on the "mmx code skipping bug" - the
> symptoms, the cause, the fix, and who's guilty?
Yes. I knew about the original FUD, I'm interested in everything else,
the cause and the fix in particular. (Because those will help other people
who wrote similar code).
> Mplayer still can't be included in distributions since it relies heavily
> on compile-time optimisation for the particular processor you have.
That would actually be a simple patch. Since the mplayer code is quite
modular, we could simply replace some of the static links with shared
libraries and dlopen(), add checks at runtime, and load different plugins
based on the results.
Imagine this:
/usr/lib/mplayer/plugins/i386/ffmpeg-out.so
/usr/lib/mplayer/plugins/i386-mmx/ffmpeg-out.so
/usr/lib/mplayer/plugins/i686/ffmpeg-out.so
if(ssetest())
foo=dlopen(LIBDIR "i686/ffmpeg-out.so")
else if(mmxtest())
foo=dlopen(LIBDIR "i386-mmx/ffmpeg-out.so")
else
foo=dlopen(LIBDIR "i386/ffmpeg-out.so")
There's an entirely different thing that keeps mplayer and similar things
out of distributions: Evil governments that have been bought out by even
more evil corporations.
While ffmpeg has free implementations of all important codecs, some of the
important ones are still covered by software patents in the US and other
countries, so you can't legally use or distribute ffmpeg in those
locations.
And since "they can only be used to view illegally ripped movies anyway"
(just like the one and only purpose of decss is supposed to be pirating
DVDs), politicians are unlikely to step in on our side.
> > So far the only drawback I've seen when comparing mplayer to xine is that
> > xine can't play Video-CD rips (".bin files"), but that should be fixable.
>
> Well I tried xine long ago and it crashed like hell.
Haven't seen the latest versions (0.9.1 and 0.9.2) crashing - but I
haven't used them all that much.
LLaP
bero
_______________________________________________
Redhat-devel-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/redhat-devel-list
