[MPlayer-cvslog] r25521 - trunk/libmpdemux/demux_ogg.c

Rich Felker dalias at aerifal.cx
Thu Jan 3 04:54:41 CET 2008


On Wed, Jan 02, 2008 at 11:53:45PM +0100, Balatoni Denes wrote:
> Hi!
> 
> Wednesday 02 January 2008 21:01-kor Diego Biurrun ezt írta:
> > On Thu, Dec 27, 2007 at 03:10:07PM +0100, Balatoni Denes wrote:
> > > Thursday 27 December 2007 12:44-kor Diego Biurrun ezt írta:
> >
> > There is no doubt that contributing to MPlayer can be a slow and tedious
> > process.  However, I disagree that this is a question of attitude.  It
> > is a matter of manpower.  We receive around an order of magnitude more
> > patches than we have time to review.
> 
> You are perfectly right that manpower is an issue. But I am also critical of 
> the reviewing process, where each patch is picked apart to it's last letter.. 
> I think it would be suffcient to point out blatant errors, and violations of 
> (preferably written) policy - but not force the contributor into compliance 
> in every minor (sometimes subjective) issue. It would be a form of generosity 
> towards the contributor if you wish, and without mentioning the obvious 
> advantages imho it also wouldn't lead to any visible or practical degradation 
> to the quality of the codebase (we are talking about the mplayer codebase 
> here afterall). I am sure that this point is controversial to many of you - 
> but this is the compromise and attitude I have mentioned before.

I don't think any of the show-stoppers have been minor issues like
indention preference or choice of function names or whatnot. After
all, those things are all trivial for the contributor to fix. The
show-stoppers have almost always been MAJOR design considerations with
long-term impact on the maintainability and longevity of the code.

No one is willing to accept hacks into the codebase that will lead to
a forced choice between feature regressions and massive rewrites a
year down the line, possibly after the contributor who understood all
the hacks has long-since vanished. I've dealt with too much code like
that left over by Arpi, which while I'm sure he understood the
motivations and usefulness perfectly, is pretty much indeciperable to
us now... Thus the dead-end of MEncoder... After I gave up on it, Oded
made some progress as did a few other people, but it's essentially
still stuck.

Rich



More information about the MPlayer-cvslog mailing list