From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: sparse (cgcc) vs lintool Date: Fri, 6 Jun 2008 15:25:57 +0200 Message-ID: <20080606132557.GA27189@lst.de> References: <20080606123434.GA23323@lst.de> <20080606124955.GE600@ins.uni-bonn.de> <20080606130756.GA24879@lst.de> <20080606132123.GF600@ins.uni-bonn.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from verein.lst.de ([213.95.11.210]:38304 "EHLO verein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753973AbYFFN0G (ORCPT ); Fri, 6 Jun 2008 09:26:06 -0400 Content-Disposition: inline In-Reply-To: <20080606132123.GF600@ins.uni-bonn.de> Sender: linux-sparse-owner@vger.kernel.org List-Id: linux-sparse@vger.kernel.org To: Christoph Hellwig , linux-sparse@vger.kernel.org On Fri, Jun 06, 2008 at 03:21:23PM +0200, Ralf Wildenhues wrote: > Well, thing is, the typical way things work is that a libtool script is > generated as part of the configure process, and it sets things for the > $CC, $CXX, etc. used for that configure. With this package, however, > /usr/bin/libtool is used, and that typically has been configured only > for /usr/bin/gcc etc. > > I suppose a distro could add a section for cgcc, yes, or make it assume > --tag=CC for 'cgcc'. But really passing --tag=CC is the sane thing to > do: the package knows that your code is C code, whereas libtool > inferring it from the command line is error-prone (which is why the tag > check was made stricter in the first place). Thanks, that explanation makes a lot of sense. I'll see if I can fix up the xfsprogs buildsystems, it's a little odd anyway.