From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 872 invoked by alias); 1 Jul 2005 03:10:19 -0000 Mailing-List: contact binutils-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: binutils-owner@sources.redhat.com Received: (qmail 794 invoked by uid 22791); 1 Jul 2005 03:10:09 -0000 Received: from w098.z064220152.sjc-ca.dsl.cnc.net (HELO bluesmobile.specifixinc.com) (64.220.152.98) by sourceware.org (qpsmtpd/0.30-dev) with ESMTP; Fri, 01 Jul 2005 03:10:09 +0000 Received: from [127.0.0.1] (bluesmobile.corp.specifix.com [192.168.1.2]) by bluesmobile.specifixinc.com (Postfix) with ESMTP id 730C716A66; Thu, 30 Jun 2005 20:10:07 -0700 (PDT) Subject: Re: [PATCH] ia64: unwind directive handling From: James E Wilson To: Jan Beulich Cc: binutils@sources.redhat.com In-Reply-To: <42BFF1C3020000780001D842@emea1-mh.id2.novell.com> References: <42BFF1C3020000780001D842@emea1-mh.id2.novell.com> Content-Type: text/plain Message-Id: <1120187407.32003.100.camel@aretha.corp.specifixinc.com> Mime-Version: 1.0 Date: Fri, 01 Jul 2005 03:10:00 -0000 Content-Transfer-Encoding: 7bit X-SW-Source: 2005-07/txt/msg00000.txt.bz2 On Mon, 2005-06-27 at 03:32, Jan Beulich wrote: > Here's the updated patch, with (I believe) all issues addressed (except for where I gave reasons for why things were done the way they are in an earlier reply). Additionally to that there are four more new warnings (noting inconsistencies between .prologue and later .save-s). Sorry about the delays. I'm hoping to be more responsive now that the GCC Summit is over. It messed up my schedule the last few weeks. I have no problem with any of the answers you've given, or the subsequent changes you made. About the default in the middle of the switch statement issue, I missed the fact that there was a missing break. There is a relatively common convention to add a /* FALLTHOUGH */ marker in this case. This makes it obvious that the missing break is intentional, and not a bug. It also makes it obvious that the case labels can't be reordered without breaking the code. Anyways, in this case, I think the rewrite is better. Your original code was too clever for its own good. This invariably leads to problems, as at some point in the future someone you don't know will fail to understand how clever the code is, assume it is broken, and then proceed to fix it. Clear code is better than clever code, even if it is a few lines longer. And yes, gotos should be avoided. I said I was going to do some testing with the first version of the patch. glibc built OK. I found two minor typos in the linux kernel, spurious characters at the end of an unwind directive that were now caught whereas they used to be ignored. Both have been reported, and I think both have already been fixed. In any case, there is nothing to worry about here. I may have forgetten to do the gcc bootstrap. I don't recall. The patch is OK. -- Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com