From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dan.rpsys.net (5751f4a1.skybroadband.com [87.81.244.161]) by mail.openembedded.org (Postfix) with ESMTP id 14FA37A0E2 for ; Mon, 1 Oct 2018 14:30:34 +0000 (UTC) Received: from hex ([192.168.3.34]) (authenticated bits=0) by dan.rpsys.net (8.15.2/8.15.2/Debian-10) with ESMTPSA id w91EUXMg029224 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Mon, 1 Oct 2018 15:30:35 +0100 Message-ID: From: Richard Purdie To: Kevin Hao , openembedded-core@lists.openembedded.org Date: Mon, 01 Oct 2018 15:30:33 +0100 In-Reply-To: <20180929070147.88203-4-kexin.hao@windriver.com> References: <20180929070147.88203-1-kexin.hao@windriver.com> <20180929070147.88203-4-kexin.hao@windriver.com> X-Mailer: Evolution 3.28.1-2 Mime-Version: 1.0 X-Virus-Scanned: clamav-milter 0.100.1 at dan X-Virus-Status: Clean Subject: Re: [PATCH 3/4] bitbake.conf: Introduce variables used to access the other recipe specific sysroot directory in multilib case X-BeenThere: openembedded-core@lists.openembedded.org X-Mailman-Version: 2.1.12 Precedence: list List-Id: Patches and discussions about the oe-core layer List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 01 Oct 2018 14:30:35 -0000 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit On Sat, 2018-09-29 at 15:01 +0800, Kevin Hao wrote: > When multilib is enabled, there would be two different recipe > specific > sysroot directories: > libxx-recipe-sysroot > recipe-sysroot > > We do need to access the files under the "recipe-sysroot" directory > in some cases. So introduces these variables for this purpose. > > Signed-off-by: Kevin Hao > --- > meta/conf/bitbake.conf | 4 ++++ > meta/conf/multilib.conf | 1 + > 2 files changed, 5 insertions(+) This doesn't look like a good idea to me. With multilib its very easy just to add in special case options and make it do the right thing until something else needs 'fixing'. What is harder but works out better in the long run is to figure out the real underlying problem and the solution. This patch adds datadir but paves the way to add all the other *dir variants. Also, the name "RECIPE_ALT_SYSROOT" is horrible, its near impossible to know what "ALT" means. Alternative to what? So whilst I understand what this is trying to do, we need to come up with a better way, sorry. I've tried thinking how to do it differently but don't have a proposal right now, sorry. Cheers, Richard