From owner-freebsd-current@FreeBSD.ORG Sun Dec 20 18:24:45 2009 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id C05AD1065672 for ; Sun, 20 Dec 2009 18:24:45 +0000 (UTC) (envelope-from sgk@troutmask.apl.washington.edu) Received: from troutmask.apl.washington.edu (troutmask.apl.washington.edu [128.208.78.105]) by mx1.freebsd.org (Postfix) with ESMTP id 845958FC0A for ; Sun, 20 Dec 2009 18:24:45 +0000 (UTC) Received: from troutmask.apl.washington.edu (localhost.apl.washington.edu [127.0.0.1]) by troutmask.apl.washington.edu (8.14.3/8.14.3) with ESMTP id nBKIObk6031852; Sun, 20 Dec 2009 10:24:37 -0800 (PST) (envelope-from sgk@troutmask.apl.washington.edu) Received: (from sgk@localhost) by troutmask.apl.washington.edu (8.14.3/8.14.3/Submit) id nBKIObLs031851; Sun, 20 Dec 2009 10:24:37 -0800 (PST) (envelope-from sgk) Date: Sun, 20 Dec 2009 10:24:37 -0800 From: Steve Kargl To: Anton Shterenlikht Message-ID: <20091220182437.GA31691@troutmask.apl.washington.edu> References: <20091220114619.GA94146@mech-cluster241.men.bris.ac.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20091220114619.GA94146@mech-cluster241.men.bris.ac.uk> User-Agent: Mutt/1.4.2.3i Cc: freebsd-current@freebsd.org Subject: Re: g95 as a system fortran compiler? X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sun, 20 Dec 2009 18:24:45 -0000 On Sun, Dec 20, 2009 at 11:46:19AM +0000, Anton Shterenlikht wrote: > > I understand that gfortran is not an ideal choice for > many reasons, not least that it doesn't build on ia64. Do you have any details to support this claim (other than the fact that you can't get gcc to build on ia64)? Benchmarks seem to offer a different viewpoint. http://www.polyhedron.com/compare0html > I think g95 could be a better choice for the following > reasons: > > 1. very few dependencies, easy to maintain: > > Port: g95-0.92.20090624 > Path: /usr/ports/lang/g95 > Info: Fortran 95 compiler from g95.org > Maint: gahr@FreeBSD.org > B-deps: gettext-0.17_1 gmake-3.81_3 libiconv-1.13.1 > R-deps: > WWW: http://www.g95.org/ You left out the dependency that it uses gcc-4.0.3 as it's base gcc. > 2. builds fine on amd64, i386, ia64, sparc64 (and alpha, but this > is no longer relevant). Also builds fine on linux ppc, mips, arm > (haven't got fbsd stats for these archs). Install the g95 port and be done with it. > 3. actively developed, including co-arrays, a fortran 2008 feature g95 has a single developer who has not been heard from since June as far as patching g95 is concerned. Your statement that g95 is 'actively developed' suggests that you believe that gfortran isn't actively developed. 235 patches have been committed to gfortran and its runtime library since July 1, 2009. Note, these aren't all bug fixes suggesting gfortran is low quality. Many are new features such as OOP or changes to the gcc infrastructure (ie., use of MPC for complex types, autoparallelization, and vectorization). The co-arrays feature is a closed source shareware product. http://www.g95.biz/ As for F2003 and F2008, gfortran has implemented many more of the newer features including some of the OOP features in Fortran. http://gcc.gnu.org/wiki/Fortran2003Status http://gcc.gnu.org/wiki/Fortran2008Status > 4. proven track record of successful integration with > numerical libraries, scientific software, etc. ? If it's Fortran, gfortran will compile it. Not sure that this last point supports inclusion of g95 over any other Fortran compiler. > I think inclusion/exclusion of system fortran is best > controlled via /etc/src.conf, for those who have no use > for it. > > Any comments? Install the g95 port and move on in life. Fortran was removed from the base system when GCC deprecated g77 in the gcc-4.x series and FreeBSD moved to the newer compiler. Fortran should not return to base system. PS: Guess who is an active gfortran developer? -- Steve