[FFmpeg-cvslog] r11690 - trunk/libavcodec/h263.c

[Rich Felker]

On Thu, Jan 31, 2008 at 05:06:35PM +0100, Michael Niedermayer wrote:
> On Wed, Jan 30, 2008 at 09:32:35PM -0800, Mike Melanson wrote:
> > Rich Felker wrote:
> > >> Maybe thread-local storage? :)
> > > 
> > > Umm, the whole point was about valid C. There is no such thing in the
> > > C language and even if there were, its performance is MUCH worse than
> > > normal variables. Also it does not solve the reentrancy problem; for
> > > example, using a decoder from a signal handler would not be safe.
> > 
> > Yeah, I know. I just wanted to see your head explode. :) I hope there's 
> > some reasonable solution for making this warning go away so we can raise 
> > the S/N ratio on our stderr.
> 
> Theres a simple solution,
> have a list of harmless warnings and filter the output of gcc through that
> 
> That would also help alot with aes.c ...
> 
> And it shouldnt be hard to implement it, a mere
> gcc | egrep -v `cat harmless_warnings` instead of gcc should do
> 
> for aes.c
> aes.c:[0-9]*: warning: passing argument [1234] of '(addkey|crypt|init_multbl2|subshift|mix)' from incompatible pointer type
> 
> would remove all but 2 warnings

Indeed, this works except for your broken shell scripting that will
mangle the contents of the harmless_warnings regex file. Grep has a
nice option for what you want to do, egrep -f harmless_warnings...

Rich