From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 3115 invoked by alias); 17 Oct 2004 18:11:58 -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 3107 invoked from network); 17 Oct 2004 18:11:57 -0000 Received: from unknown (HELO mail-out3.apple.com) (17.254.13.22) by sourceware.org with SMTP; 17 Oct 2004 18:11:57 -0000 Received: from mailgate2.apple.com (a17-128-100-204.apple.com [17.128.100.204]) by mail-out3.apple.com (8.12.11/8.12.11) with ESMTP id i9HIGXML011043 for ; Sun, 17 Oct 2004 11:16:33 -0700 (PDT) Received: from relay2.apple.com (relay2.apple.com) by mailgate2.apple.com (Content Technologies SMTPRS 4.3.14) with ESMTP id ; Sun, 17 Oct 2004 11:11:56 -0700 Received: from [17.219.198.228] ([17.219.198.228]) by relay2.apple.com (8.12.11/8.12.11) with ESMTP id i9HIBrxr025713; Sun, 17 Oct 2004 11:11:54 -0700 (PDT) In-Reply-To: <200410171158.29730.stevenb@suse.de> References: <200410161217.43614.stevenb@suse.de> <4172086B.4080106@codesourcery.com> <200410171158.29730.stevenb@suse.de> Mime-Version: 1.0 (Apple Message framework v619) Content-Type: text/plain; charset=US-ASCII; format=flowed Message-Id: <05B05D49-2068-11D9-A3A0-000393B2ABA2@apple.com> Content-Transfer-Encoding: 7bit Cc: Mark Mitchell , Jakub Jelinek , GCC Patches From: Matt Austern Subject: Re: [Committed] Use special-purpose hash table to speed up walk_tree Date: Sun, 17 Oct 2004 18:45:00 -0000 To: Steven Bosscher X-SW-Source: 2004-10/txt/msg01394.txt.bz2 On Oct 17, 2004, at 2:58 AM, Steven Bosscher wrote: > On Sunday 17 October 2004 07:51, Mark Mitchell wrote: >> I don't think there is any such policy one way or the other. >> Certainly, >> there is precedent for patches being approved offline. > > I know there is, and I think it's wrong. More eyes see more > things. > > >> Matt sent me the >> patch, and it looked good to me. I didn't see any reason to test it >> on >> other architectures. > > Well, we've have some pretty serious breakage a few times in the > past few weeks. Already three times the cause of this breakage > was that some patch wasn't tested on some architecture. So I > see lots of reasons to require better testing for patches when > the mainlne is in stage3. > > Can we make it a requirement that larger patches like this should > be tested on three platforms when the mainline is in stage3? That may be a good idea, but bear in mind that it probably would not have caught this problem. I don't regularly use a platform where pointers are 64-bit, so if I were testing this on three platforms it probably would have been three 32-bit platforms. If we're looking for diversity, it had probably better be something like this: at least one non-Linux platform, at least one non-ELF platform, at least one non-x86 platform, at least two different word sizes. --Matt