From owner-freebsd-python@FreeBSD.ORG Wed Dec 4 13:13:22 2013 Return-Path: Delivered-To: python@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id A109BF18 for ; Wed, 4 Dec 2013 13:13:22 +0000 (UTC) Received: from ainaz.pair.com (ainaz.pair.com [209.68.2.66]) by mx1.freebsd.org (Postfix) with ESMTP id 7BEF21FCF for ; Wed, 4 Dec 2013 13:13:22 +0000 (UTC) Received: from tuna (ip-2-206-0-140.web.vodafone.de [2.206.0.140]) by ainaz.pair.com (Postfix) with ESMTPSA id 55CE73F445; Wed, 4 Dec 2013 08:13:10 -0500 (EST) Date: Wed, 4 Dec 2013 14:13:00 +0100 (CET) From: Gerald Pfeifer To: William Grzybowski Subject: Re: python packages struggling with GCC 4.7 In-Reply-To: Message-ID: References: <201311060508.rA658XmR012155@svn.freebsd.org> <20131106063906.GA1385@medusa.sysfault.org> <20131106070255.GE60770@FreeBSD.org> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: python@freebsd.org X-BeenThere: freebsd-python@freebsd.org X-Mailman-Version: 2.1.17 Precedence: list List-Id: FreeBSD-specific Python issues List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 04 Dec 2013 13:13:22 -0000 On Wed, 4 Dec 2013, William Grzybowski wrote: > The main reason foe the python failures, if not all, is math/p-numpy. > It is compiling just fine, but the package is not working: > > ImportError: /lib/libgcc_s.so.1: version GCC_4.6.0 required by > /usr/local/lib/gcc47/libgfortran.so.3 not found > > Looks like it is linking to /lib/libgcc_s.so.1 and should use > /usr/local/lib/gcc47/libgcc_s.so.1, am I right? #1, you rock. #2, you are right. Is it possible that math/p-numpy fails to use LDFLAGS which would have added -Wl,rpath=/usr/local/lib/gcc47/libgcc_s.so.1 when linking? Or is it a question of getting the order of the dynamic library search paths right? Unfortunately I am far from an expert in this area (and having two libraries with the same version, one of which is a superset, is tricky). (Why is this not happening when building with lang/gcc being GCC 4.6 or lang/gcc46, though? That puzzles me. Also, if this were a general issue, wouldn't all ports using Fortran be affected?) Can you somehow fix this in/for math/p-numpy? Gerald