[MPlayer-cvslog] r26095 - trunk/configure

[Diego Biurrun]

On Mon, Feb 25, 2008 at 05:14:22PM -0500, Rich Felker wrote:
> On Mon, Feb 25, 2008 at 11:02:48PM +0100, Diego Biurrun wrote:
> > On Mon, Feb 25, 2008 at 02:19:34PM -0500, Rich Felker wrote:
> > > On Sun, Feb 24, 2008 at 11:41:39PM +0100, diego wrote:
> > > > 
> > > > Log:
> > > > Disable internal VIDIX on OS/2, patch by Dave Yeo, dave.r.yeo gmail com.
> > > > 
> > > > --- trunk/configure	(original)
> > > > +++ trunk/configure	Sun Feb 24 23:41:39 2008
> > > > @@ -4257,7 +4257,7 @@ if test "$_vidix_internal" = auto ; then
> > > >    x86 && _vidix_internal=yes
> > > >    ppc && linux && _vidix_internal=yes
> > > >    alpha && linux && _vidix_internal=yes
> > > > -  qnx || beos || darwin && _vidix_internal=no
> > > > +  qnx || beos || darwin || os2 && _vidix_internal=no
> > > 
> > > IMO this approach is backwards. Vidix should be disabled by default
> > > and only enabled on exact cpu/os/toolchain combinations where it is
> > > KNOWN to work. x86 && _vidix_internal=yes is extremely broken..
> > 
> > Do you know where it works?  I don't ...
> 
> x86 && { linux || freebsd || netbsd || openbsd || darwin ; }

Umm, no, read what the code says about darwin :)

> Or something like that.
> The non-x86 cases already cover the OS as well it seems. The only
> bogus assumption seems to be that vidix works on ANY OS with x86 which
> could not possibly be true..

I don't mind changing the code, it just needs to be initialized with
something sane, i.e. those systems that are known to work.

Diego