From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 5056 invoked by alias); 20 Oct 2009 16:58:39 -0000 Received: (qmail 5042 invoked by uid 22791); 20 Oct 2009 16:58:39 -0000 X-SWARE-Spam-Status: No, hits=-2.5 required=5.0 tests=AWL,BAYES_00,SPF_PASS X-Spam-Check-By: sourceware.org Received: from fg-out-1718.google.com (HELO fg-out-1718.google.com) (72.14.220.154) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Tue, 20 Oct 2009 16:58:32 +0000 Received: by fg-out-1718.google.com with SMTP id d23so1845712fga.8 for ; Tue, 20 Oct 2009 09:58:30 -0700 (PDT) Received: by 10.86.228.16 with SMTP id a16mr3889766fgh.49.1256057909833; Tue, 20 Oct 2009 09:58:29 -0700 (PDT) Received: from ?192.168.2.99? (cpc2-cmbg8-0-0-cust61.cmbg.cable.ntl.com [82.6.108.62]) by mx.google.com with ESMTPS id e20sm24551fga.10.2009.10.20.09.58.28 (version=SSLv3 cipher=RC4-MD5); Tue, 20 Oct 2009 09:58:29 -0700 (PDT) Message-ID: <4ADDEFBA.7060302@gmail.com> Date: Tue, 20 Oct 2009 17:11:00 -0000 From: Dave Korn User-Agent: Thunderbird 2.0.0.17 (Windows/20080914) MIME-Version: 1.0 To: Richard Guenther CC: gcc-patches@gcc.gnu.org, spop@gcc.gnu.org, Loren James Rittle Subject: Re: [PATCH] Fix GRAPHITE configure References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit X-IsSubscribed: yes Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org X-SW-Source: 2009-10/txt/msg01303.txt.bz2 [ Cc added: Loren, who emailed me off-list about the top-level configure tests reporting 'yes' even when ppl/cloog aren't going to be configured in. ] Richard Guenther wrote: > On Tue, 20 Oct 2009, Richard Guenther wrote: > >> It seems that even though the documentation states that --with-cloog >> and --with-ppl need only be specified if cloog/ppl do not reside in >> the default search paths graphite is disabled if I do not specify >> them (with a correct path even). >> it seems clear that without specifying --with-cloog with_cloog will >> be "no" and thus while the header checks will report success (and >> thus also config.log is silent about not using graphite) CLOOGLIBS >> will be empty and thus gcc/configure will set HAVE_cloog to 0. >> >> So, what is wrong here, the documentation or the configure? I think someone just forgot to throw the switch. The current state is a hangover from when they were optional: see http://gcc.gnu.org/ml/gcc/2009-05/threads.html#00121 http://gcc.gnu.org/ml/gcc-patches/2009-05/threads.html#00263 http://gcc.gnu.org/ml/gcc-patches/2009-05/threads.html#00259 > Even >> just specifying --with-cloog --with-ppl will not work as then I get >> >> HOST_CLOOGLIBS = -Lyes/lib -lcloog >> HOST_CLOOGINC = -Iyes/include -DCLOOG_PPL_BACKEND >> >> but still >> >> checking for version 0.10 of PPL... yes >> checking for correct version of CLooG... yes >> >> so I believe all these checks are somehow broken. Very likely. I didn't try to fix the underlying problems at the time because I expected the whole thing to get a proper rewrite by the Graphite maintainers when Graphite was made mandatory, but that seems to have been overlooked until you picked it up now. > 2009-10-20 Richard Guenther > > * configure.ac: Adjust the ppl and cloog configure to work as > documented. Disable cloog if ppl was disabled. Omit the version > checks if they were disabled. > * configure: Re-generate. Loren, perhaps you'd like to test Richard's patch(*) and see if it does what you wanted? cheers, DaveK -- (*) - http://gcc.gnu.org/ml/gcc-patches/2009-10/msg01274.html