From owner-cvs-ports@FreeBSD.ORG Sun Sep 26 14:58:58 2010 Return-Path: Delivered-To: cvs-ports@FreeBSD.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id B4CCC1065670; Sun, 26 Sep 2010 14:58:58 +0000 (UTC) (envelope-from gerald@pfeifer.com) Received: from vexpert.dbai.tuwien.ac.at (vexpert.dbai.tuwien.ac.at [128.131.111.2]) by mx1.freebsd.org (Postfix) with ESMTP id 7356B8FC0C; Sun, 26 Sep 2010 14:58:58 +0000 (UTC) Received: from acrux.dbai.tuwien.ac.at (acrux.dbai.tuwien.ac.at [128.131.111.60]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by vexpert.dbai.tuwien.ac.at (Postfix) with ESMTPS id 92E2A1E06F; Sun, 26 Sep 2010 16:58:55 +0200 (CEST) Date: Sun, 26 Sep 2010 16:58:58 +0200 (CEST) From: Gerald Pfeifer To: Alexey Dokuchaev In-Reply-To: <20100822112542.GB58812@FreeBSD.org> Message-ID: References: <201008201617.o7KGHVAG037310@repoman.freebsd.org> <20100822112542.GB58812@FreeBSD.org> User-Agent: Alpine 2.00 (LNX 1167 2008-08-23) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: cvs-ports@FreeBSD.org, cvs-all@FreeBSD.org, ports-committers@FreeBSD.org Subject: Re: cvs commit: ports/lang/gcc45 Makefile distinfo X-BeenThere: cvs-ports@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: CVS commit messages for the ports tree List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 26 Sep 2010 14:58:58 -0000 Hi Alexey, On Sun, 22 Aug 2010, Alexey Dokuchaev wrote: >> Replace NOT_FOR_ARCHS by ONLY_FOR_ARCHS. [1] >> >> Suggested by: linimon > I don't really see the point. Ideally, any port (including gcc) should > work across all architectures. If the port inherently cannot work on > some arch (meaning it is not simply BROKEN there, but cannot be possibly > fixed without e.g. total rewrite, or if it is binary-only, etc.), it can > be marked IGNORE (old way) or (ONLY|NOT)_FOR_ARCHS (new way). what you describe matches my understanding for general ports. When it comes to a compiler which needs explicit support for different processor architectures, operating systems, and the combinations thereof (which is not automagic even if given support for each of the two in general), I do feel things are indeed a bit different. For example, GCC has been broken on FreeBSD/ia64 for three release series and apparently only a single user cared about it without really managing to get things addressed, though. And when some general infrastructure around was created in GCC that needs specific configuration for every platform nobody stepped upfor months. I did address both (the former still leaving some work in progress) and one or two others now, even though I am not a compiler hacker, but it shows that things are not straightforward for software like a compiler. > Now, since GCC is open source software, I believe it could/should be > fixed to work on every Tier 1/2 arch; which means that ONLY_FOR_ARCHS > will grow. In this light, the change is controversial at best. Any help to fix this is very welcome! Some of our fellow committers certainly can attest that I've managed to help getting things addressed upstream and patches included there and I will continue to do my best to help with that. In fact, I am in the process of testing and pinging one where Kan has helped a user to address an issue he found. > I'd like to hear Mark's rationale for [1]. I'm not Mark, and do not feel too strongly about the use of ONLY_FOR_ARCHS here, but usually if portmgr makes a request I try to honor it. :-) Gerald