FreeTDS Development Group

["James K. Lowden" <jklowden AT schemamania.org>]

On Sat, 19 Jul 2003 16:18:00 -0500, GUTIERREZ Pablo TESIS-TAMSA
<Tesgup AT tamsa.com.mx> wrote:
> 
> 1) I'll send you a patch for to include <select.h> if appropriated.

Got it, thanks! 

> 2) The problem is that even when I already made 
> 
> #if defined (__QNX__) && defined (__WATCOMC__)
> # define vsnprintf _vbprintf
> # define  snprintf  _bprintf
> #endif /* QNX */

You don't need "&& defined (__WATCOMC__)" if QNX is providing the
functions.  

> ./configure says that the function vsnprintf() is not present, so
> ./include/config.h doesn't define HAVE_VSNPRINTF => All the code that
> use vsnprintf is not compiled.
> What can I do?

Ah, that makes sense, and makes the test more complex.  First, we need
./configure to look for _bprintf and _vbprintf.  I added that to
configure.in.  Can you generate your own configure script?  You need:

$ (autoconf --version; automake --version; libtool --version) |grep GNU
autoconf (GNU Autoconf) 2.57
automake (GNU automake) 1.7.3
ltmain.sh (GNU libtool) 1.4a (1.641.2.255 2001/05/22 10:39:30)

Our test becomes something like:

#if defined (__QNX__) 
# if !HAVE_VSNPRINTF && HAVE__VBPRINTF
#  define vsnprintf _vbprintf
# endif /* vsnprintf */
# if !HAVE_SNPRINTF && HAVE__BPRINTF
#  define  snprintf  _bprintf
# endif /* snprintf */
#endif /* QNX */

That's a little pedantic, but it is what we're testing.  If someone
else is using QNX with another libc, the above won't interfere.  

--jkl