rj...@apache.org wrote: > > Modified: httpd/mod_ftp/trunk/modules/ftp/ftp_internal.h > URL: > http://svn.apache.org/viewvc/httpd/mod_ftp/trunk/modules/ftp/ftp_internal.h?rev=816074&r1=816073&r2=816074&view=diff > ============================================================================== > --- httpd/mod_ftp/trunk/modules/ftp/ftp_internal.h (original) > +++ httpd/mod_ftp/trunk/modules/ftp/ftp_internal.h Thu Sep 17 07:00:24 2009 > @@ -84,6 +84,8 @@ > > #if APR_HAVE_SYS_STAT_H > #include <sys/stat.h> > +#elif HAVE_SYS_STAT_H > +#include <sys/stat.h> > #endif
NAK. The fragment above makes zero sense, please revert. #if APR_HAVE_SYS_STAT_H should be sufficent. If apr does not provide it consistently, and httpd has, then the test becomes #ifdef HAVE_SYS_STAT_H which is an altogether different beast. First clue that the code above was wrong is that you included the same code for both cases. So provided that you had no clue if APR consistently provided this and you wanted to rely upon config.h, then it becomes #if (defined(APR_HAVE_SYS_STAT_H) && APR_HAVE_SYS_STAT_H) \ || defined(HAVE_SYS_STAT_H) #include... but we know we don't need to go that far. I'll answer the question you raised on list tomorrow, although it certainly seemed like an apr question and not an httpd question.