[FFmpeg-devel] [RFC] Applying trivial things vs. sending patches.

[Reimar Döffinger]

On Sat, Aug 30, 2008 at 03:12:53AM +0200, Aurelien Jacobs wrote:
> Michael Niedermayer wrote:
> > On Sat, Aug 30, 2008 at 02:29:33AM +0200, Aurelien Jacobs wrote:
> > > Michael Niedermayer wrote:
> > > > As the question of applying trivial things (typos, trivial optimizations,
> > > > simplifications, ...) occasionally causes fights and flames i propose here
> > > > the heal all ill solution ;)
> > > > 
> > > > What about,
> > > > Everyone who is ok with or is especially not ok with some things being
> > > > directly commited to his code, should say so clearly in the MAINTAINERS
> > > > file.
> > > 
> > > Before resorting to this, is there anyone against trivial commit without
> > > patch submission in code he maintains ?
> > 
> > > 
> > > (note that I'm not sure whether your latest dca.c commit can really be
> > > considered trivial or not, I would say it is borderline)
> > 
> > hmm, i considered it trivial, or rather i considered it
> > uncontroversal, i mean less code, faster code, less memory ...; who would
> > object to that?
> 
> I agree that it should be uncontroversial, but that's not the same as
> trivial...

I agree, optimizations are hardly ever trivial, and I would prefer it to
be restricted to trivial (not that there is a clear definition for it).
Optimizations IMO are very easy to do wrong (e.g. miss a special case),
and I would not really want them to get less review by getting applied
directly.