From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 11562 invoked by alias); 16 Oct 2004 21:29:19 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 11553 invoked from network); 16 Oct 2004 21:29:18 -0000 Received: from unknown (HELO vlsi1.ultra.nyu.edu) (128.122.140.213) by sourceware.org with SMTP; 16 Oct 2004 21:29:18 -0000 Received: by vlsi1.ultra.nyu.edu (4.1/1.34) id AA27728; Sat, 16 Oct 04 17:33:03 EDT Date: Sun, 17 Oct 2004 08:54:00 -0000 From: kenner@vlsi1.ultra.nyu.edu (Richard Kenner) Message-Id: <10410162133.AA27728@vlsi1.ultra.nyu.edu> To: pinskia@physics.uc.edu Subject: Re: [RFC] Gimple and language independent Cc: gcc@gcc.gnu.org X-SW-Source: 2004-10/txt/msg00665.txt.bz2 Really gimple should be language independent in the sense the types are defined by gimple and not the other way around (the only exception is structs and unions). What do people think about this? I think that structs and unions are among the most interesting cases, since the question of which are compatible is so language-dependent, so I think we can't exclude that case. I know there has been talk about this recently and how gimple is dependent on the type system of the source language which I think is wrong and we loose some optimizations because of it. I don't see that.