From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mga02.intel.com ([134.134.136.20]) by linuxtogo.org with esmtp (Exim 4.72) (envelope-from ) id 1TdiRZ-0001pk-7C for openembedded-core@lists.openembedded.org; Wed, 28 Nov 2012 15:16:53 +0100 Received: from orsmga002.jf.intel.com ([10.7.209.21]) by orsmga101.jf.intel.com with ESMTP; 28 Nov 2012 06:02:37 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.84,177,1355126400"; d="scan'208";a="248519807" Received: from unknown (HELO helios.localnet) ([10.252.122.248]) by orsmga002.jf.intel.com with ESMTP; 28 Nov 2012 06:02:36 -0800 From: Paul Eggleton To: "Robert P. J. Day" Date: Wed, 28 Nov 2012 14:02:34 +0000 Message-ID: <2455772.UuKjZbS977@helios> Organization: Intel Corporation User-Agent: KMail/4.9.3 (Linux/3.2.0-33-generic-pae; KDE/4.9.3; i686; ; ) In-Reply-To: References: <5890106.Vb1bZqkAH2@helios> MIME-Version: 1.0 Cc: openembedded-core@lists.openembedded.org Subject: Re: what is "packagegroup-core-nfs-server"? X-BeenThere: openembedded-core@lists.openembedded.org X-Mailman-Version: 2.1.11 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: Wed, 28 Nov 2012 14:16:53 -0000 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="us-ascii" On Wednesday 28 November 2012 08:45:15 Robert P. J. Day wrote: > On Wed, 28 Nov 2012, Paul Eggleton wrote: > > ... snip for brevity... > > > So the bit that might be missing here is that the PACKAGE_GROUP > > structure and the packagegroup recipes/class are actually not really > > related, except that the latter may be used to populate the former. > > "packagegroup" is the new name for what used to be known as a "task" > > in versions before danny and much better reflects what these recipes > > do. PACKAGE_GROUP_* actually came first as a way to define items for > > IMAGE_FEATURES that map to one or more packages, and I'm not sure > > the name choice was a particularly good one - I wonder if it would > > be worth considering renaming it to something like > > IMAGE_FEATURE_PACKAGES in order to avoid confusion, although > > obviously every rename like this has a cost associated. > > > > In any case, I would definitely recommend using the term "package > > group" to refer to packagegroup recipes only, otherwise you're only > > likely to increase people's level of confusion. > > one more observation just to mess things up a bit more -- the fact > that, despite what one reads in core-image.bbclass, the IMAGE_FEATURES > variable can be used to customize your image in ways that have nothing > to do with packages or package groups. > > witness this from meta/classes/rootfs_ipk.bbclass: > > if ${@base_contains("IMAGE_FEATURES", "read-only-rootfs", "true", > "false" ,d)}; then if grep Status:.install.ok.unpacked ${STATUS}; then echo > "Some packages could not be configured offline and rootfs is read-only." > exit 1 > fi > fi > > so there's a potential "image feature" called "read-only-rootfs" which > is clearly not a package group or a package, and of which there is no > mention in core-image.bbclass. Yep, that's intended - there are a number of these. IMAGE_FEATURES controls features in the image, whether they are sets of packages to be installed or other aspects to enable during image construction. This particular IMAGE_FEATURES item (read-only-rootfs) is not listed probably because it's limited to the ipk backend only and not widely used. There is a bug open to implement it for rpm and deb, at which time it should probably be added to the list there and to the one in the manual. Cheers, Paul -- Paul Eggleton Intel Open Source Technology Centre