This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: mips/mips.h bug with WCHAR_TYPE?


FYI: a similar inconsistency is/was in gcc 2.95.2 on AIX, where the patched
<stddef.h> defines wint_t to unsigned int while the system <ctype.h> (which
isn't replaced by gcc) defines wint_t to int. I've been meaning to submit a bug
report on this but haven't so far. Don't know if this is still a problem in the
latest sources.

Regards
Martin

$ grep -n wint_t /usr/include/ctype.h 
101:    typedef int     wint_t;
127:    extern int      is_wctype(wint_t, wctype_t);

$ grep -n wint_t
/package/1/compilers/g++/2.95.2/lib/gcc-lib/rs6000-ibm-aix4.3.3.0/2.95.2/include/stddef.h 
5:    || defined(__need_wint_t)
12:     && !defined(__need_wint_t))
273:#if defined (_STDDEF_H) || defined (__need_wint_t)
280:typedef __WINT_TYPE__ wint_t;
282:#undef __need_wint_t

$ grep -n __WINT_TYPE__
/package/1/compilers/g++/2.95.2/lib/gcc-lib/rs6000-ibm-aix4.3.3.0/2.95.2/include/stddef.h 
277:#ifndef __WINT_TYPE__
278:#define __WINT_TYPE__ unsigned int
280:typedef __WINT_TYPE__ wint_t;

Neil Booth wrote:
> 
> I think I've found what causes David's testsuite failures for wide
> characters.  In mips/iris.h we have
> 
> #define WCHAR_TYPE      "long int"
> #define WCHAR_TYPE_SIZE LONG_TYPE_SIZE
> 
> For -mabi=64, this is an 8-byte wide char!  Surely you jest.
> 
> I supect these should be
> 
> #define WCHAR_TYPE      "int"
> #define WCHAR_TYPE_SIZE INT_TYPE_SIZE
> 
> Would someone with more knowledge of MIPS please confirm or deny?
> Does this fix the bug for you David?
> 
> Thanks,
> 
> Neil.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]