[Ffmpeg-devel] More signed fun

[Ivan Kalvachev]

2006/3/26, M?ns Rullg?rd <mru at inprovide.com>:
> Diego Biurrun <diego at biurrun.de> writes:
>
> > On Thu, Mar 23, 2006 at 10:32:53AM +0000, M?ns Rullg?rd wrote:
> >>
> >> Steve Lhomme said:
> >> >
> >> > Depending on how your preprocessor handle data, not using () in macros
> >> > can lead to problems. Like interpreting (- - variable) (invalid C
> >> > syntax)instead of -(-variable). Here is a fix.
> >>
> >> - -x is perfectly valid C syntax.  If your compiler chokes on it,
> >> it's broken.
> >>
> >> That said, the fix is appropriate for other reasons.
> >
> > So will you commit it?
>
> Actually, looking closer those parens are not needed at all.  The
> macro is only used a few times just below the definition, and always
> with a single number for each argument.  Those parens only add useless
> clutter in this case.

Bulshit.

Giving that the code that use is:
 CHECK_BIDIR2( 0,-1, 0, 1)
I was wondering how it compiles at all, well gcc -E reveield 
something like this:
...
(- -1<=0 || motion_fx+- -1<=xmax) && (-0<=0 || motion_fy+-0<=ymax) &&
 (-1<=0 || motion_bx+-1<=xmax) && (-1<=0 || motion_by+-1<=ymax) &&(-
-1>=0 || motion_fx+- -1>=xmin) && (-0>=0 ||
...
For these who don't use monospace font, there is space between both - -.
I'm not sure where this space have came from, but i doubt it is part
of the standard. It is better to don't risk.
I'm going to commit the following code really soon:

#define CHECK_BIDIR2(a,b,c,d)\
CHECK_BIDIR((a),(b),(c),(d))\
CHECK_BIDIR((-(a)),(-(b)),(-(c)),(-(d)))