From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 497 invoked by alias); 10 Jun 2004 20:04:00 -0000 Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org Received: (qmail 489 invoked from network); 10 Jun 2004 20:03:59 -0000 Received: from unknown (HELO mail.codesourcery.com) (65.74.133.5) by sourceware.org with SMTP; 10 Jun 2004 20:03:59 -0000 Received: (qmail 23647 invoked from network); 10 Jun 2004 20:03:58 -0000 Received: from taltos.codesourcery.com (zack@66.92.218.83) by mail.codesourcery.com with DES-CBC3-SHA encrypted SMTP; 10 Jun 2004 20:03:58 -0000 Received: by taltos.codesourcery.com (sSMTP sendmail emulation); Thu, 10 Jun 2004 13:03:57 -0700 To: Andreas Jaeger Cc: Phil Edwards , law@redhat.com, Jerry Quinn , gcc-patches@gcc.gnu.org Subject: Re: [patch] Minor improvement to typeclass.h References: <877julhhlb.fsf@codesourcery.com> <1086817343.18174.320.camel@speedy> <878yevcm11.fsf@taltos.codesourcery.com> <20040610183105.GA22168@disaster.jaj.com> From: Zack Weinberg Date: Thu, 10 Jun 2004 21:26:00 -0000 In-Reply-To: (Andreas Jaeger's message of "Thu, 10 Jun 2004 21:14:16 +0200") Message-ID: <87vfhz9o42.fsf@taltos.codesourcery.com> User-Agent: Gnus/5.110003 (No Gnus v0.3) Emacs/21.3 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-SW-Source: 2004-06/txt/msg00677.txt.bz2 Andreas Jaeger writes: >> I tried getting rid of this header when I was doing compile-time type >> reflection/introspection on Gaby's now-dead branch. It turned out that >> glibc used it for some twisted purpose, even though gcc doesn't. > > I just did a grep through the current glibc sources and see no > references to typeclass.h at all. Especially tgmath.h does not > include it. It might be that tgmath.h could be written in a better > way with typeclass... I was unclear. glibc tgmath.h uses __builtin_classify_type. It doesn't use typeclass.h (nor could it - gcc doesn't install it); it has hardwired numeric constants for the builtin's return value. Y'all might find __builtin_types_compatible_p a useable replacement, but I can't be sure. (Take a look at gcc/config/rs6000/altivec.h for example usage.) zw