[Ffmpeg-devel] Coverity defect scan

[Steve Lhomme]

Reimar D?ffinger wrote:
> Hi,
> On Sat, Mar 11, 2006 at 09:10:18AM +0100, Diego Biurrun wrote:
>> On Fri, Mar 10, 2006 at 11:43:54PM +0100, Michael Niedermayer wrote:
>>> calling these bugs or errors stretches the definition by quite a large 
>>> amount id call them warnings which is what they are, and like the gcc 
>>> warnings some of them point to actual bugs but most are irrelevant
>>> in practice IMHO
>> Well, judging from the commits I'm seeing on mplayer-cvslog, quite a few
>> of them seem to be real bugs.
> 
> Well, most of them aren't really relevant, though good to fix in the
> spirit of getting cleaner code (though MPlayer is in more need in that
> respect).
> E.g. the memleaks are practically all at error conditions and thus
> unlikely to matter to 99% of all users.

Well, only if you consider that's "good enough" for users and that we 
live in a perfect world where all files are clean (and FFMPEG compliant).

IMO, that's a problem with many OSS projects that consider 99% is enough 
and fail on the last mile. As I often say "when you offer something to 
someone, make sure it's not rotten first".

I'm not saying this is the case of FFMPEG which mostly works and is very 
useful. It's just a general comment :)