Re: gint32 / guint32

["Gary V. Vaughan" <gvaughan oranda demon co uk>]

Raja R Harinath wrote:
> 
> "Gary V. Vaughan" <gvaughan@oranda.demon.co.uk> writes:
> > Andreas Cittlau wrote:
> > > [[snip]]
> > >
> > > The error I get (a little piece of it included below) indicates
> > > that I dont have gint32 and guint32 defined(?).
> >
> > This indicates that the new glibconfig.h generation code (generated
> > in config.status from configure.in) is not working... either the
> > AC_CHECK_SIZEOF macros are not working, or the case statements
> > generate the typedefs for glibconfig.h are not matching.
> >
> > In my case I hacked configure.in slightly so that it hardcodes the
> > correct values for my machine.
> 
> Please tell us why you needed to hardcode the values.  AFAIK, the
> configure.in checks are very similar to what was originally in
> `glib.h' in coverage -- if the old code in `glib.h' checked for
> SIZEOF_INT and SIZEOF_LONG, the new code in configure.in checks for
> $ac_cv_sizeof_int and $ac_cv_sizeof_long.  If the new checks are
> failing, the old ones must have, too, unless there's a problem with
> the new checks.

I am compiling with bleeding edge libtool/automake/autoconf under cygwin
b20 with a patched cygwin1.dll on NT4SP3.  `Nuff said =)O|

The old code worked fine with the particular NT/cygwin setup I had at
the time, but the new code triggers a bug somewhere which causes ^M's to
be inserted in the case statements that set the ac_cv_sizeof_* vars.
OTOH both the old and new code work fine on my Linux box.

Cheers,
	Gary V. Vaughan.