[MPlayer-dev-eng] [RFC] rc2 at the beginning of October

[Aurelien Jacobs]

Uoti Urpala wrote:

> On Tue, 2007-09-18 at 13:21 -0400, Rich Felker wrote:
> > On Tue, Sep 18, 2007 at 12:52:42PM +0200, Diego Biurrun wrote:
> > > On Sun, Sep 16, 2007 at 04:16:00AM +0200, Attila Kinali wrote:
> > > > On Sat, 15 Sep 2007 12:18:10 +0200
> > > > Diego Biurrun <diego at biurrun.de> wrote:
> > > > I agree with Ivan that only a few people have Matrox graphic
> > > > cards, so it does not make sense to enable them by default.
> > > > They should rather be autodetected. Either by using lspci
> > > > output, trying to find /dev/mga_vid* or any combination thereof.
> > > 
> > > The disadvantage of this autodetection is that you have to recompile
> > > MPlayer after you have set up mga_vid...
> > 
> > I agree, this is a major disadvantage. These drivers are tiny and will
> > not cause any ill effects if left to fallback at runtime. There's no
> 
> IMO whatever setup is chosen for the rare drivers it must not result in
> any additional error (or other) messages in the default output on
> ordinary systems. Enabling mga_vid etc by default as they currently are
> would fail that; they all print errors during autodetection.
> 
> There are a few ways to fix that; I don't particularly care which one is
> picked.
> 
> - Make them not print errors when they fail. Has the disadvantage that
> on systems where they should work or when -vo xyz is specified
> explicitly the failure is not obvious.
> 
> - Use autodetection during compile time. Has the disadvantage mentioned
> above, but OTOH makes it reasonable and possibly useful to print errors
> when they fail (there was a reasonable expectation it SHOULD have
> worked, and maybe the user should investigate).
> 
> - Move them to the end of the autodetection order and require the user
> to specify them in the default config. IMO reasonable at least for
> devices which require extra steps like creating special devices to work.
> 
> - Leave them disabled by default.

There is one more way which would probably be acceptable:

- Make them print errors only when they were enabled explicitly. That way
faster drivers are still auto-detected, but if auto-detection fails, it
don't print errors. Still when you use -vo xyz, it print errors.

Does it sounds like a good compromise ?

Aurel