Hi Stepan, On 11/28/06, Stepan Kasal <kasal@xxxxxx> wrote:
On Mon, Nov 27, 2006 at 08:26:20AM -0800, Bruce Korb wrote: > In this case, truth is, I don't think "wchar.h" requires "runetype.h", > but rather wint_t type *normally* requires "wchar.h" but, for OS/X, > it instead requires "runetype.h". well, I do not see the difference here: wint_t shall be defined in wchar.h.
Except it is (on OS/X) defined in runetype.h and runetype.h is not #included by wchar.h. "self sufficient" means that this module:
#include <wchar.h> int foo = 0;
will compile without error. "I think" (without knowing for certain) that this will compile on OS/X. However, this module:
#include <wchar.h> wint_t foo = 0;
will not. But this latter is not a self-sufficiency test. It is a standards test. It fails that. Therefore, a rigorous CHECK_WINT_T configury test would need to automatically check for this runetype.h header and use it if found to test for a system definition of wint_t.
The message which started this thread indicates that if you include wchar.h alone, the compilation breaks.
Only in conjunction with using the wint_t type.
It seems that including runetype.h before wchar.h fixed that. So it seems that wchar.h requires runetype.h.
wint_t requires runetype.h. I do not think wchar.h needs it. There really is a difference, even if a bit subtle. Regards, Bruce _______________________________________________ Autoconf mailing list Autoconf@xxxxxxx http://lists.gnu.org/mailman/listinfo/autoconf
