From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mga03.intel.com ([143.182.124.21]) by linuxtogo.org with esmtp (Exim 4.72) (envelope-from ) id 1T206P-00063W-No for openembedded-core@lists.openembedded.org; Thu, 16 Aug 2012 15:27:10 +0200 Received: from azsmga002.ch.intel.com ([10.2.17.35]) by azsmga101.ch.intel.com with ESMTP; 16 Aug 2012 06:15:11 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.77,778,1336374000"; d="scan'208";a="134975987" Received: from unknown (HELO helios.localnet) ([10.252.121.76]) by AZSMGA002.ch.intel.com with ESMTP; 16 Aug 2012 06:15:10 -0700 From: Paul Eggleton To: Koen Kooi Date: Thu, 16 Aug 2012 14:15:09 +0100 Message-ID: <4232599.R290qD4K8p@helios> Organization: Intel Corporation User-Agent: KMail/4.9 (Linux/3.2.0-29-generic-pae; KDE/4.9.0; i686; ; ) In-Reply-To: References: MIME-Version: 1.0 Cc: openembedded-core@lists.openembedded.org Subject: Re: How to override bitbake.conf defaults in layers properly? X-BeenThere: openembedded-core@lists.openembedded.org X-Mailman-Version: 2.1.11 Precedence: list Reply-To: Patches and discussions about the oe-core layer List-Id: Patches and discussions about the oe-core layer List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 16 Aug 2012 13:27:10 -0000 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="us-ascii" On Thursday 16 August 2012 13:57:44 Koen Kooi wrote: > I spent the morning backporting kernel.bbclass changes to the meta-oe denzil > branch and I've run into a problem: STAGING_KERNEL_DIR needs to change to > match the updated class. In this specific case I can work with Scott and > Eric to see if those kernel.bbclass changes can go into oe-core and meta-oe > at the same time. That's something for next week :) In the meantime I'm > wondering what the best way is to fix this in cases where you can't > influence oe-core. So let's be clear, having a copy of kernel.bbclass in meta-oe is half the reason you have this problem. The sooner we get out of this situation the better, IMHO. > I see a number of options: > > 1) set STAGING_KERNEL_DIR in layer.conf. I don't know if that's parsed > before or after bitbake.conf, but it feels like the cleanest solution layer.conf for each layer is parsed very early - before bitbake.conf, so if you're looking to alter things that are set in bitbake.conf there it's not really going to work unless you use override hacks. > 2) set it in $DISTRO.conf. Easy enough, breaks every other distro out there To answer for everyone else's benefit, this is normally the correct place to globally override defaults set in bitbake.conf, and for the avoidance of confusion, settings in the distro config will absolutely take precedence. It's not workable in this situation, but this situation is a special one. > 3) copy over bitbake.conf. Might work with some BBPATH magic, but in the end > doesn't scale when >1 layers do it With all of the other settings in bitbake.conf and the maintenance pain this would lead to this is not a good idea. > 4) provide a custom denzil branch of oe-core, another easy way, but very > antisocial Perhaps, but if the changes were not accepted in OE-Core denzil branch it would be more than a little antisocial to expect to impose them via meta-oe. Cheers, Paul -- Paul Eggleton Intel Open Source Technology Centre