FreeTDS Development Group

[Brian Bruns <camber AT ais.org>]

On Fri, 10 Nov 2000, Mark Schaal wrote:

> > 
> > 
> > There are some autoconf rules for getting the size of particular data
> > types. AC_CHECK_SIZEOF(x) will set $ac_cv_sizeof_x which can be check in
> > configure.in to set the right sizes.  This will probably have wider
> > portability than <inttypes.h>
> > 
> > BTW, this is what glib does to define its types.
> > 
> > What do you think?
> > 
> > Brian
> 
> That's a possibility.  My main preference for looking for the int32_t
> family header file is that once you have one you have them all, plus it
> seems simple to implement, plus I had the hope that we would find them
> defined in the same place on most platforms.  If you can steal already
> tested logic straight from glib, that sounds great to me.
> 
The glib stuff seems pretty straight forward, except it uses the
information to generate a platform specific glibconfig.h file that gets
included by glib.h.  So we would need to have configure generate a file
like include/tdstypes.h that would be included from tds.h etc...
Definitely work for the 0.52 series ;-)

And I'm not sure how this would affect the rpm/deb builds.

BTW, I'm fixing up the last few things for 0.51 (gonna release pre2
tonight hopefully), I've decided to leave 64bit and TDS7 on big endian out
with the hopes of releasing a 0.52 in the next two months or less.  I'd
like to get ideas about what should be on the short list for that
release. So far I've got:

1. TDS7 on big endians
2. 64bit platforms fix
3. ctlib needs to return rows affected back correctly (see sqsh)
4. col names and col info in server code
5. intermittant php/apache crashes (hopefully)
6. new freetds.conf config file format (although we will still support
   sybase interfaces format as well).  This is mostly to give extra
   configurability beyond what the interfaces can reasonably support, and 
   make freetds a better fit on linux distros...)

Brian