Hubert Feyrer wrote:

> On Wed, 21 Jul 1999, Simon Burge wrote:
> > I just had a look at the recent patches for gmp (both stand-along and the
> > ssh version).  Can we run into problems where a pentium or greater that
> > builds a binary package produces a package that can't run on an i386 or
> > i486?
> 
> We can, if gmp includes P5-only code. 
> I don't know if this is the case with our packages.

The following looks a bit suspect (from pkgsrc/devel/gmp/patches/patch-ab):

+       case "${target}" in
+         i[34]86*-*-*bsd*)
+               path="x86" ;;
+         i[56]86*-*-*bsd* | pentium-*-*bsd* | pentiumpro-*-*bsd*)
+               path="x86/pentium x86" ;;
+       esac ;;

so I'd guess the answer is "yes" unfortunately.  Maybe this sort of
behaviour could be overriden by a /etc/mk.conf variable perhaps.
Something like "IM_IN_BINARY_PACKAGE_MODE=yes", but shorter :-)

Simon.