Re: [gentoo-dev] RFC: emboss.eclass as replacement for embassy.eclass

[justin]

On 27/03/11 16:50, Mike Frysinger wrote:
> On Sun, Mar 27, 2011 at 10:34 AM, justin wrote:
>> So I need one last hint, how to correct following correctly?
>>
>>
>> #if defined (HAVE64) && !defined(AJ_MACOSXLF) && !defined(AJ_HPUXLF) &&
>> !defined(AJ_FreeBSDLF) && !defined(AJ_AIXLF)
>>    struct dirent64 *dp;
>> #else
>>    struct dirent *dp;
>> #endif
>>
>> #if defined (HAVE64) && !defined(AJ_MACOSXLF) && !defined(AJ_HPUXLF) &&
>> !defined(AJ_FreeBSDLF) && !defined(AJ_AIXLF)
>>    struct stat64 sbuf;
>> #else
>>    struct stat sbuf;
>> #endif
>
> neither should be necessary with LFS.  if you call
> AC_USE_SYSTEM_EXTENSIONS or AC_SYS_LARGEFILE, the system will take
> care of translating stat into stat64 as needed.
>
> but in practice, i guess what they'll want to do is:
>  - call AC_USE_SYSTEM_EXTENSIONS at top of configure script
>  - add some AC_TRY_COMPILE's:
> AC_CACHE_CHECK([for stat64], ac_cv_struct_stat64,
>   [AC_TRY_COMPILE([#include <sys/stat.h>],
>   [struct stat64 st],
>   ac_cv_struct_stat64=yes, ac_cv_struct_stat64=no)])
>   if test "x$ac_cv_struct_stat64" = xyes; then
>     AC_DEFINE(HAVE_STRUCT_STAT64)
>   fi
>  - change the code to look at HAVE_STRUCT_STAT64 instead of random
> system defines
>
> (largely untested :P)
> -mike
>

Thanks Mike,

compiletime and runtime tests are fine. I really owe you one!