From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-64716-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 28290 invoked by alias); 16 Dec 2002 17:09:37 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 28264 invoked from network); 16 Dec 2002 17:09:35 -0000
Received: from unknown (HELO momotombo.TechFak.Uni-Bielefeld.DE) (129.70.136.107)
  by sources.redhat.com with SMTP; 16 Dec 2002 17:09:35 -0000
Received: from xayide.TechFak.Uni-Bielefeld.DE.TechFak.Uni-Bielefeld.DE (xayide.TechFak.Uni-Bielefeld.DE [129.70.137.35])
	by momotombo.TechFak.Uni-Bielefeld.DE (8.11.6+Sun/8.11.6/TechFak/pk+ro20010720) with ESMTP id gBGH9WU25156;
	Mon, 16 Dec 2002 18:09:32 +0100 (MET)
From: Rainer Orth <ro@TechFak.Uni-Bielefeld.DE>
Message-ID: <15870.2247.822105.412724@xayide.TechFak.Uni-Bielefeld.DE>
Date: Mon, 16 Dec 2002 09:24:00 -0000
To: "Kaveh R. Ghazi" <ghazi@caip.rutgers.edu>
Cc: gcc-bugs@gcc.gnu.org, gcc@gcc.gnu.org, libstdc++@gcc.gnu.org,
   oldham@codesourcery.com
Subject: Re: Irix6 long doubles implemented wrong?  (27_io/ostream_inserter_arith)
In-Reply-To: <200212142145.QAA25890@caip.rutgers.edu>
References: <200212142145.QAA25890@caip.rutgers.edu>
Mime-Version: 1.0 (generated by tm-edit 7.106)
Content-Type: text/plain; charset=US-ASCII
X-SW-Source: 2002-12/txt/msg00890.txt.bz2

Kaveh R. Ghazi writes:

> As you can see, gcc's sizeof (long double) doesn't agree with that
> from the system compiler and I would assume therefore that gcc also
> disagrees with libc's notion of long double.  This means that the %Lf
> specifier is hosed and cannot be used from gcc on irix6.
> 
> I believe this is the cause of the 27_io/ostream_inserter_arith.cc
> failures we've been getting on irix6.
> 
> See also http://gcc.gnu.org/ml/gcc-bugs/2001-08/msg00088.html which
> apears to be the same issue.

It is.  The following comment from config/mips/abi64.h (lost when the
definitions from that file were merged into mips.h) still applies:

/* ??? Unimplemented stuff follows.  */

/* ??? Add support for 16 byte/128 bit long doubles here when
   mips_abi != ABI32.  */

> Does anyone have any thoughts on this?  (Or better yet a fix?)

I didn't have much time to work on gcc lately, but David Edelsohn's support
or 128-bit long doubles in AIX (implemented as a pair of 64-bit doubles)
applies to IRIX 6 long doubles as well, since both platforms use the same
format as documented in math(3M).

	Rainer