From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [128.178.224.219] (helo=smtp0.epfl.ch) by linuxtogo.org with smtp (Exim 4.69) (envelope-from ) id 1MgCZo-0001tz-2s for openembedded-devel@openembedded.org; Wed, 26 Aug 2009 09:05:52 +0200 Received: (qmail 26067 invoked by uid 107); 26 Aug 2009 06:47:50 -0000 X-Virus-Scanned: ClamAV Received: from unknown (85.4.79.108) (authenticated) by smtp0.epfl.ch (AngelmatoPhylax SMTP proxy); Wed, 26 Aug 2009 08:47:50 +0200 Message-ID: <4A94DA90.8090306@epfl.ch> Date: Wed, 26 Aug 2009 08:47:44 +0200 From: Valentin Longchamp User-Agent: Thunderbird 2.0.0.23 (X11/20090817) MIME-Version: 1.0 To: openembedded-devel@lists.openembedded.org References: <4A937831.2000304@4d-electronics.co.nz> <4A93A9E3.7090605@epfl.ch> <4A9461C3.2000406@4d-electronics.co.nz> In-Reply-To: <4A9461C3.2000406@4d-electronics.co.nz> Subject: Re: [PATCH] Add toolchain file generation to cmake bbclass X-BeenThere: openembedded-devel@lists.openembedded.org X-Mailman-Version: 2.1.11 Precedence: list Reply-To: openembedded-devel@lists.openembedded.org List-Id: Using the OpenEmbedded metadata to build Distributions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 26 Aug 2009 07:06:31 -0000 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Matthew Dombroski wrote: >> I have had no problem with boost and with Qt4, I have been able to >> achieve similar results without the toolchain file, only by using a >> qt.conf file as Zecke had suggested to me. > I'm using boost-1.39 and found i needed to set Boost_COMPILER or it > wouldn't find libraries. (Oh how I hate bjam, but thats another story...) Ok, I think I had 1.36. > > Setting Qt environment data in cmake.bbclass may seem a bit out of > place. When you consider that cmake is distributed with scripts for > configuring/detecting Qt it doesn't seem too bad. > It is definitly tidier than generating a qt.conf in every recipe. Adding > all the Qt info in EXTRA_OECMAKE is doing the same work twice. Well that's a point of view. From my point of view, not all cmake based programs need that qt info, that's why I find it's cleaner to enable it in the recipes that need these libraries. About adding the EXTRA_OECMAKE, I have to check, but I think I needed to define some so that the right qmake is found. I will try without or by adding only the needed definitions. > >> Although the toolchain file is the preferred value according to the >> cmake documentation, it has not been necessary to me. > > The effect of a toolchain file is the same as that of defining lots of > options in EXTRA_OECMAKE. > personally, I think it is cleaner. The ability to use it outside of > openembedded is really useful to me also. I agree that it is the same. It's just that it was done until now like that in OE, so I chose that way. However one advantage that I see of defining things directly in the recipes and not in a toolchain file, is that you can easily add definitions using EXTRA_OECMAKE in every recipe. Is cmake able to take two toolchain files (one from cmake.bbclass and one from the recipes with the needed libraries) without having to rewrite the do_configure in the recipe ? This comes back to the point above: do we need to define all the libraries that could come with cmake-based programs in cmake.bbclass ? From my point of view, this is definitely something related to the recipes. Val