From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 8921 invoked by alias); 15 Oct 2002 07:22:32 -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 8863 invoked from network); 15 Oct 2002 07:22:31 -0000 Received: from unknown (HELO egil.codesourcery.com) (66.92.14.122) by sources.redhat.com with SMTP; 15 Oct 2002 07:22:31 -0000 Received: from zack by egil.codesourcery.com with local (Exim 3.36 #1 (Debian)) id 181Lz0-0004CX-00; Tue, 15 Oct 2002 00:19:14 -0700 Date: Tue, 15 Oct 2002 02:18:00 -0000 To: Mark Mitchell Cc: "Kaveh R. Ghazi" , "doko@cs.tu-berlin.de" , "debian-gcc@lists.debian.org" , "gcc-bugs@gcc.gnu.org" , "gcc@gcc.gnu.org" Subject: Re: gcc-3_2-branch bootstrap failure when using bison-1.50 Message-ID: <20021015071914.GB15067@codesourcery.com> References: <200210130308.XAA13661@caip.rutgers.edu> <34920000.1034663177@warlock.codesourcery.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <34920000.1034663177@warlock.codesourcery.com> User-Agent: Mutt/1.4i From: Zack Weinberg X-SW-Source: 2002-10/txt/msg00789.txt.bz2 On Mon, Oct 14, 2002 at 11:26:17PM -0700, Mark Mitchell wrote: > > >I'll leave it to our release manager to decide if this issue warrants > >backporting the relevant patches or not. > > I think that your point that we will include the generated files on the > branch is a good one; let's not backport the patches given that. Counter-argument - Since I (like Matthias) track Debian unstable, I no longer have bison 1.35 to use with the 3.2 branch. I can, I suppose, grab an old tarball, build it, install it as bison1.35, and pass BISON= every time I do a build on the 3.2 branch, but this is a fair amount of work. Worse, any OS distribution which (a) includes gcc 3.2, (b) does so by tracking the 3.2 CVS branch instead of working from offical release tarballs, and (c) does not maintain side-by-side packages of bison 1.35 and 1.50, will have the same problem. And it may be necessary for them to start maintaining side-by-side packages of 1.35 and 1.50 in order to make the problem go away. Debian unstable is in precisely this boat. Thus, provided that we can nail the noncompile/920923-1.c regression, and provided that bison 1.35 continues to work, I think we probably should backport the fixes for bison 1.50. zw