[FFmpeg-cvslog] r11803 - in trunk: ffmpeg.c ffserver.c

[Reimar Döffinger]

Hello,
On Sat, Feb 02, 2008 at 11:41:56PM +0100, Michael Niedermayer wrote:
> On Sat, Feb 02, 2008 at 09:07:05AM +0100, Reimar D?ffinger wrote:
> > On Sat, Feb 02, 2008 at 01:54:37AM +0100, aurel wrote:
> > > Author: aurel
> > > Date: Sat Feb  2 01:54:37 2008
> > > New Revision: 11803
> > > 
> > > Log:
> > > non-const 2nd parameter for strtol/strtod
> > > 
> > > Modified:
> > >    trunk/ffmpeg.c
> > >    trunk/ffserver.c
> > > 
> > > Modified: trunk/ffmpeg.c
> > > ==============================================================================
> > > --- trunk/ffmpeg.c	(original)
> > > +++ trunk/ffmpeg.c	Sat Feb  2 01:54:37 2008
> > > @@ -2396,16 +2396,17 @@ static void opt_frame_aspect_ratio(const
> > >      int x = 0, y = 0;
> > >      double ar = 0;
> > >      const char *p;
> > > +    char *end;
> > >  
> > >      p = strchr(arg, ':');
> > >      if (p) {
> > > -        x = strtol(arg, (char **)&arg, 10);
> > > -        if (arg == p)
> > > -            y = strtol(arg+1, (char **)&arg, 10);
> > > +        x = strtol(arg, &end, 10);
> > > +        if (end == p)
> > > +            y = strtol(end+1, &end, 10);
> > 
> > While this maybe avoids a warning, I do not really like it, since the
> > thing that "end" points to is supposed to be constant.
> 
> How do you suggest to fix it?

Hmm.. Well, ignoring the warning is a possibility.
Then also possible is
> x = strtol(arg, &end, 10);
> arg = end;
> if (arg == p)

etc. at least reducing the risk.
Then we could also add a av_const_strtol or some such that uses the
right arguments.
Given the number of places where this is an issue this might actually
make sense...

Greetings,
Reimar D?ffinger