[MPlayer-dev-eng] Rework of vidix: now a static lib.

[Nico Sabbi]

Benjamin Zores wrote:
> Hi,
> 
> I rework and rewrote the vidix library.
> As requested by people from here and there, it is now a completely
> static library that got all drivers built-in. The rewrite was
> necessary as the vidix1 lib was using shared libraries for drivers and
> uses dlopen/dlsym everytime to get all symbols.
> 
> The vidix2 lib is now way easier and it registers all drivers directly.
> I've planned many other changes to the library (like porting all
> changes from upstream vidix) but first step is to commit the current
> work.
> I've tried (and managed to) not to modify the public API. Only drivers
> loading and entry point of changed (so only internal changes). The
> libvo interface suffers from no changes at all and drivers code has no
> functionnal change either (so everything should be still working).
> 
> I'd like you to test it by:
> - applying the attached patch
> - using the vidix directory available at
> http://geexbox.org/~ben/vidix-20070329.tar.bz2 in place of the current
> one (just rename the old one). The changes are huge so a patch is
> useless.
> 

as root here it works, as normal user and with dhahelper modprobed it 
doesn't work because it can't scan the pci bus;
using -vo xvidix:radeon_vid.(a|so) doesn't work either (what subdevice
should I use? )

Anyway I still consider nonsense having all drivers compiled 
statically because I'll NEVER need any of them except one
(what do you do when you compile your kernel: enable
all available drivers statically? or do you compile all possible
ao and vo in mplayer? I doubt it) so I'm for either:
- giving the user the possibility to compile them dynamically like now
- or giving the user the possibility to disable all drivers but X

-- 
"Without a frontend, mplayer is useless" - someone in mplayer-users