From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 2820 invoked by alias); 5 Jul 2005 20:24:14 -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 2775 invoked by uid 22791); 5 Jul 2005 20:24:08 -0000 Received: from smtp-102-tuesday.noc.nerim.net (HELO mallaury.nerim.net) (62.4.17.102) by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Tue, 05 Jul 2005 20:24:08 +0000 Received: from uniton.integrable-solutions.net (gdr.net1.nerim.net [62.212.99.186]) by mallaury.nerim.net (Postfix) with ESMTP id 26A0E4F3E3 for ; Tue, 5 Jul 2005 22:23:54 +0200 (CEST) Received: from uniton.integrable-solutions.net (localhost [127.0.0.1]) by uniton.integrable-solutions.net (8.12.10/8.12.10/SuSE Linux 0.7) with ESMTP id j65KMwKY007793 for ; Tue, 5 Jul 2005 22:22:58 +0200 Received: (from gdr@localhost) by uniton.integrable-solutions.net (8.12.10/8.12.10/Submit) id j65KMw0O007792; Tue, 5 Jul 2005 22:22:58 +0200 To: gcc@gcc.gnu.org Subject: Re: tr1::unordered_set bizarre rounding behavior (x86) References: <42CABDCC.3070508@suse.de> <20050705181025.GB2315@synopsys.com> <42CAD4AC.6040008@suse.de> <42CAE536.7010303@suse.de> From: Gabriel Dos Reis In-Reply-To: <42CAE536.7010303@suse.de> Date: Tue, 05 Jul 2005 20:24:00 -0000 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-SW-Source: 2005-07/txt/msg00195.txt.bz2 Paolo Carlini writes: | Gabriel Dos Reis wrote: | | >If you regard the object representation as an array of bytes, does it | >take long realize it is not much different from hashing a character | >string? | > | It takes less if your proposal comes together with a specific one for | character string hashing: You do have one already. Start from there. Measure, then we can argue later. Now that you realize it is same issue, we've made significant progress. [...] | involved. For example, I'm not at all sure that the hash function | suggested by submitter of 21193 would also work statistically well for | floats (note that most of the real world succesful applications | mentioned in the linked web page are about *real* text). I don't think Yes, but note also that those depend on some assumptions about the distribution of some words. Not every specific context would equally benefit from that. The point being that once you have reduced the problem to one previously known solved problem, it is only a matter of arguing about little details. I think that is far better than current situation. -- Gaby