From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 31267 invoked by alias); 25 Jan 2002 16:31:05 -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 31201 invoked from network); 25 Jan 2002 16:31:02 -0000 Received: from unknown (HELO dberlin.org) (64.246.6.106) by sources.redhat.com with SMTP; 25 Jan 2002 16:31:02 -0000 Received: from localhost (localhost.localdomain [127.0.0.1]) by dberlin.org (8.11.6/8.11.6) with ESMTP id g0PGUnn31854; Fri, 25 Jan 2002 11:30:49 -0500 Date: Fri, 25 Jan 2002 08:49:00 -0000 From: Daniel Berlin To: Robert Dewar cc: gcc@gcc.gnu.org, Subject: Re: g++ and aliasing bools In-Reply-To: <20020125160537.AA1A6F28AD@nile.gnat.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-SW-Source: 2002-01/txt/msg01641.txt.bz2 On Fri, 25 Jan 2002, Robert Dewar wrote: > < place all types in the right sets. If it doesn't, you are back to > undecidability. > >> > > Of course it is undecidable in the dynamic sense, but there we are not > interested in "the right sets", we are interested in sets that we can > prove are disjoint. It is quite "right" to put everything in the same > set, just not very efficient. We are NOT looking for an optimal solution, > here, that's obvious to anyone that that's unattainable. Of course. > > What we are looking for is improved, verifiable, principles for splitting > the sets more finely. > > Sometimes I really think they should not teach anyone about undecidability. > It always ends up with people looking at a perfectly simple problem like > this one (simple conceptually, not simple to get good solutions to), and > worrying about undecidability when it is a complete red herring. Not here, however. > > If you propose that two items are not aliased, and it is undecidable whether > they are aliased, that's fine, it just means they go in the same alias set, > no big deal! However, the important point is that the language semantics don't allow us to get it *right*, not just deciding. We may get it *wrong*, and think we have it *right*. --Dan