From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-yb1-f176.google.com (mail-yb1-f176.google.com [209.85.219.176]) by mail.openembedded.org (Postfix) with ESMTP id 95EE37BF2E for ; Mon, 14 Jan 2019 12:56:39 +0000 (UTC) Received: by mail-yb1-f176.google.com with SMTP id x9so8560695ybj.5 for ; Mon, 14 Jan 2019 04:56:41 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:subject:date:message-id:mime-version :content-transfer-encoding; bh=AI0Y6tSJHGQ6oifepyxcMfOZaXHCmhgAy5teJ9vAYO4=; b=DrH9whfGnTV724eV1lSlfWan2zGcF+g5jqNOMiBuxefOZeNYNYfAYP2ibrn8ylGvCq jc/uvYDdnboMZr2DiBUa31zsZzQ70BNtvdhaYPjGKhFcvJj65R0nr1O371KZRKS2IS0Q OLdF+f+rO5evMIe2ZLeIksd2nIocCyo/xNk82ttLFrqYxgSo+uhWMgN6fnKGlHMqEPkK GicVrzBYUVukno1cJiwLGow54+NKBKV1YUe0/S8LBwGGeC2q38fb/zOI7/5ASINRfx/P BbOme3yeef3KK+okp0k3SMcgvSjNXARWzwXKQHoJy9oCNjebG0KoCjSVcG27v1me/Eei EjZA== X-Gm-Message-State: AJcUukfxOR+YdY2YFW+bKkVDxIIJS4+aJ++TgeKNhb7o50Hr7Oro28gN THBOy1gkRCd8y4xq0Ty85qAsI4ap X-Google-Smtp-Source: ALg8bN6SgJdMjqPe5wbAI8ilZgxBqktMWBz8yvlOr7jQxAOaHEnrHHl+u1LagT+KcnV2yuhBbbnnMA== X-Received: by 2002:a25:6652:: with SMTP id z18mr4219534ybm.242.1547470600466; Mon, 14 Jan 2019 04:56:40 -0800 (PST) Received: from tfsielt31850.garage.tyco.com ([77.107.218.170]) by smtp.gmail.com with ESMTPSA id h145sm115143ywc.72.2019.01.14.04.56.38 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Mon, 14 Jan 2019 04:56:39 -0800 (PST) From: =?UTF-8?q?Andr=C3=A9=20Draszik?= To: openembedded-core@lists.openembedded.org Date: Mon, 14 Jan 2019 12:56:20 +0000 Message-Id: <20190114125632.4780-1-git@andred.net> X-Mailer: git-send-email 2.20.1 MIME-Version: 1.0 Subject: [PATCH 00/12] RFC: util-linux: one package per binary 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, 14 Jan 2019 12:56:40 -0000 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This series of patches changes util-linux to splitting into one package per binary automatically, using do_split_packages() * the first 5 patches are a general clean-up * patch 2 could be classed as avoiding a potential crash * patch 5 is important, as it fixes up some existing packaging problems * the subsequent patches are the actual implementation The slight 1st problem was that update alternatives was running right after copying from D to PKGD, and at that point in time PACKAGES and ALTERNATVE_xxx were expected to be fully populated. So I changed this slightly, so that util-linux can generate PACKAGES now right after perform_packagecopy(), but before update-alternatives processing. The small 2nd problem was that FILES_${PN} must reference the renamed file name (after update-alternatives has run). Given this is quite hard to do automatically when populating packages using do_split_packages(), I updated update-alternatives to update the file name in FILES_${PN} if the non-renamed file was specified in there. This would also allow updating all recipes to drop knowledge about that renaming where this is done at the moment. Finally, I am splitting up the packages, one commit per directory. Things to note and questions: * it is unclear to me why util-linux RDEPENDS vs. RRECOMMENDS some of the manually splitted packages before this series, and some of those pre-existing packages weren't in RRECOMMENDS or RDEPENDS at all. For that reason, the RDEPENDS on packages from base_sbindir and base_bindir that existed before this series, weren't updated, only new packages from those two directories have been added to RRECOMMENDS of the meta-package. I would suggest to drop all RDEPENDS and simply have util-linux RRECOMMENDS *all* the packages, but I don't know the origial reasoning for that distinction. * /etc/default/mountall is the only file remaining in the util-linux package. I am not sure what to do with it as I coulnd't find any user anyway. Can it be dropped altogether, should it be moved into the util-linux-mount package instead? Note that the util-linux-mount and util-linux-umount existed before this series, but /etc/default/mountall was still packaged into util-linux. * The pre-existing util-linux-swaponoff is a meta-package now, that simply RDEPENDS on -swapon and -swapoff, simply for backwards compatibility reasons, but is it really necessary, given people can now add -swapon and -swapoff if they want to. No harm in keeping it either, though. * the -dev package RRECOMMENDS some (but not all) of the split -dev packages (which don't necessarily exist), e.g. util-linux-losetup-dev Should it RPROVIDE those itself instead? Should those be made ALLOW_EMPTY? Should the RRECOMMENDS be tweaked? What is the preferred approach here?