From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 4779 invoked by alias); 26 Mar 2004 19:50:13 -0000 Mailing-List: contact libc-hacker-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-hacker-owner@sources.redhat.com Received: (qmail 4755 invoked from network); 26 Mar 2004 19:50:12 -0000 Received: from unknown (HELO myware.akkadia.org) (24.221.190.179) by sources.redhat.com with SMTP; 26 Mar 2004 19:50:12 -0000 Received: from redhat.com (drepper@myware.akkadia.org [192.168.7.70]) (authenticated bits=0) by myware.akkadia.org (8.12.11/8.12.10) with ESMTP id i2QJngY7015867; Fri, 26 Mar 2004 11:49:42 -0800 Message-ID: <40648956.20700@redhat.com> Date: Sat, 27 Mar 2004 08:18:00 -0000 From: Ulrich Drepper Organization: Red Hat, Inc. User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7b) Gecko/20040314 MIME-Version: 1.0 To: Jakub Jelinek CC: Richard Henderson , Glibc hackers Subject: Re: [PATCH] long double IEEE double -> quad switch References: <20040324160427.GM15946@sunsite.ms.mff.cuni.cz> In-Reply-To: <20040324160427.GM15946@sunsite.ms.mff.cuni.cz> X-Enigmail-Version: 0.83.3.0 X-Enigmail-Supports: pgp-inline, pgp-mime Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-SW-Source: 2004-03/txt/msg00136.txt.bz2 I don't like any of this very much, people should just bite the bullet. But if this is not possible we'll have to make the best out of it. The most important part are the public headers. There is too much added so far. Especially for loser compilers. Hide that stuff in some extra header which only exists for the affected platforms and which is only included in case such a compiler is used. I'm mainly talking about the long list of defines in and . The defined __LONG_DOUBLE_MATH_OPTIONAL || !defined __NO_LONG_DOUBLE_MATH expression probably deserves it own #define to make tests shorter. All the redirect macros should be defined in a central place: . No need to duplicate this. And: out of sight. is not that often looked at. The DBL_IS_LDBL_DECL and DBL_IS_LDBL macros probably should be defined in include/libc-symbols.h and not in individual files. This will also eliminate some #ifdef in function definition parameter lists. If you do something like +#include +#ifdef __LONG_DOUBLE_MATH_OPTIONAL +# define vasprintf(s, f, a) ___vasprintf (s, f, a, 0) +#else +# define vasprintf(s, f, a) _IO_vasprintf (s, f, a) +#endif it probably is better to just use ___vasprintf everywhere. On the lighter side, some preprocesor directives are not correctly indented. -- ➧ Ulrich Drepper ➧ Red Hat, Inc. ➧ 444 Castro St ➧ Mountain View, CA ❖