Eric Blake writes:
> We've started using the _GL_ prefix, rather than GL_, because the
> package OpenGL uses the GL_ prefix
> s/one one/one/ ...
> Do you also want to check that it actually increments? And wouldn't
> __LINE__ be a slightly better fallback than 0?...
All good points, thanks;
On 05/05/2010 01:55 PM, Paul Eggert wrote:
> To fix the problem you mentioned with verify.h, I installed the
> following patch into gnulib. It uses the new __COUNTER__ macro of GCC
> (4.3 and later), which is designed for this problem. So this should fix
> the warnings for GCC 4.3 and later. Old
Eric Blake writes:
>* This implementation exploits the fact that GCC does not warn about
> the last declaration mentioned above. If a future version of GCC
> introduces a warning for this, the problem could be worked around
> by using code specialized to GCC, e.g.,:
>
>
Eric Blake wrote:
> why do they exist if no test module mentions them?
> tests/test-stpncpy.c
> tests/test-stpncpy.out.aix433
> tests/test-stpncpy.out.glibc
This is a manual test of how stpncpy is working. It could be removed;
I had it committed in case this function is buggy in other platforms a
I noticed that on libvirt, two uses of verify() caused compilation
failure when using libvirt's preferred set of gcc warnings and -Werror:
util/util.c:84: error: redundant redeclaration of 'verify_function__'
[-Wredundant-decls]
util/util.c:83: error: previous declaration of 'verify_function__' wa
