Hi Paul,
> That may be the best we can do on short notice, so I did that by installing
> the
> attached. Thanks for the diagnosis.
the patch fixes the problem for me, too. Thank you very much!
Best regards,
Peter
Ángel González wrote:
For this specific issue, another solution would be to add a #ifndef
_GCC_MAX_ALIGN_T
guard to the replacement unistd.h.
Although the latter doesn't fix the underlying problem.
That may be the best we can do on short notice, so I did that by installing the
attached. Thanks
On 01/04/16 11:30, Peter Simons wrote:
Apparently, the second (failing) build decides that max_align_t is not
available, but the gnulib-generate test build thinks it is. Not sure how
to explain that.
For what it's worth, the full source code of my project is available at
http://git.savannah.gnu.
Hi Paul,
> Can you reproduce the problem this way?
>
> ./gnulib-tool --dir foo --create-testdir unistd
> cd foo
> ./configure
> make
> make check
No, that sequence of commands gives no error. The check phase reports 12
out of 12 successful tests. The relevant bit from config.log is
availab
I'm not observing the problem on Fedora 23 with gcc (GCC) 5.3.1 20151207 (Red
Hat 5.3.1-2).
Can you reproduce the problem this way?
./gnulib-tool --dir foo --create-testdir unistd
cd foo
./configure
make
make check
As I understand it, GCC 5.3.1 should not need a substitute stddef.h. So if your
Hi,
I'm using the current "master" of gnulib at v0.1-744-gf0be2ae to build a
C project on Linux/x86_64. This works fine with older compilers, but gcc
5.3.1 cannot build my project because of the following error:
gcc -DHAVE_CONFIG_H -I. -I.. -DDEBUG -g -O2 -MT unistd.o -MD -MP -MF
.deps/un
