PR2076

[Geoff Keating]

> Date: Wed, 25 Apr 2001 14:17:19 -0700
> From: Richard Henderson <rth@redhat.com>
> Cc: gcc@gcc.gnu.org
> Mail-Followup-To: Richard Henderson <rth@redhat.com>,
> 	Geoff Keating <geoffk@geoffk.org>, gcc@gcc.gnu.org
> Content-Disposition: inline
> User-Agent: Mutt/1.2.5i
> 
> On Wed, Apr 25, 2001 at 02:00:11PM -0700, Geoff Keating wrote:
> > It's probably because of STRICT_ARGUMENT_NAMING.
> 
> Indeed it is.
> 
> > This is a compatibility hack that causes the last argument to a
> > varargs function to be treated as unnamed, because that's how GCC
> > used to do it and some backends rely on it (it could change the ABI).
> 
> Ug.  It does indeed change the abi.  What a crock.  Well, any
> idea how xlc treats this situation?  I suspect they don't load
> do this sort of thing, at which point we can make this change
> as a compatability bug fix.

I'm not sure.  If you can think of a testcase I can try it on bluey
and see what the native compiler there thinks.

-- 
- Geoffrey Keating <geoffk@geoffk.org>