Re: trivial modification to gtk/configure.in

[Owen Taylor <otaylor redhat com>]

[ Following up to gtk-devel-list, instead of gtk-bugs since that's
  where this sort of discussion should be directed ]

Thanks for asking.

This was discussed recently: 

 http://mail.gnome.org/archives/gtk-devel-list/2000-October/msg00015.html

My opinion is that it would be better to change the check to check for
siglongjump, which isn't supposed to be a macro, than to add the more
complicated configure test.

Regards,
                                        Owen

Peter Hawkins <peter hawkins emu id au> writes:

> Hi there...
> 
> I would just commit this but the README.cvs-commits says "ASK" in big letters, so I'm asking.
> 
> sigsetjmp on my copy of glibc is a preprocessor define, not an actual function. Therefore the AC_CHECK_FUNCS test fails, even though sigsetjmp really does exist in the library. Fix is below.
> 
> =)
> Peter
> 
> --- configure.in	2000/10/18 15:50:13	1.171
> +++ configure.in	2000/10/22 09:13:38
> @@ -194,7 +194,9 @@
>  fi
>  AC_SUBST(REBUILD)
>  
> -AC_CHECK_FUNCS(lstat mkstemp sigsetjmp)
> +AC_CHECK_FUNCS(lstat mkstemp)
> +AC_TRY_COMPILE([#include <setjmp.h>],[sigjmp_buf env; int r=sigsetjmp(env,0);],+	AC_DEFINE(HAVE_SIGSETJMP))
>  
>  AC_MSG_CHECKING(whether make is GNU Make)
>  STRIP_BEGIN=