From owner-freebsd-bugs@FreeBSD.ORG Fri Feb 4 19:10:19 2005 Return-Path: Delivered-To: freebsd-bugs@hub.freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id B555116A4FF for ; Fri, 4 Feb 2005 19:10:19 +0000 (GMT) Received: from freefall.freebsd.org (freefall.freebsd.org [216.136.204.21]) by mx1.FreeBSD.org (Postfix) with ESMTP id 76ED643D48 for ; Fri, 4 Feb 2005 19:10:19 +0000 (GMT) (envelope-from gnats@FreeBSD.org) Received: from freefall.freebsd.org (gnats@localhost [127.0.0.1]) by freefall.freebsd.org (8.13.1/8.13.1) with ESMTP id j14JAGPS014112 for ; Fri, 4 Feb 2005 19:10:16 GMT (envelope-from gnats@freefall.freebsd.org) Received: (from gnats@localhost) by freefall.freebsd.org (8.13.1/8.13.1/Submit) id j14JAGlr014111; Fri, 4 Feb 2005 19:10:16 GMT (envelope-from gnats) Date: Fri, 4 Feb 2005 19:10:16 GMT Message-Id: <200502041910.j14JAGlr014111@freefall.freebsd.org> To: freebsd-bugs@FreeBSD.org From: Brooks Davis Subject: Re: bin/77082: src/usr.sbin/pkg_install - Add 3 new macros to clean pkg-plist X-BeenThere: freebsd-bugs@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: Brooks Davis List-Id: Bug reports List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 04 Feb 2005 19:10:19 -0000 The following reply was made to PR bin/77082; it has been noted by GNATS. From: Brooks Davis To: Florent Thoumie Cc: Brooks Davis , FreeBSD-gnats-submit@freebsd.org, freebsd-bugs@freebsd.org, hq@freebsd.org Subject: Re: bin/77082: src/usr.sbin/pkg_install - Add 3 new macros to clean pkg-plist Date: Fri, 4 Feb 2005 11:04:57 -0800 On Fri, Feb 04, 2005 at 07:41:29PM +0100, Florent Thoumie wrote: > Brooks Davis wrote: > > >It seems like dirrmtry should take an optional message to emit if the > >event that the directory can not be delete. That way the user can be > >informed that the directory should be removed if they are really done > >using the port. > > I asked myself if I should put this feature in my patch and > I finally haven't because it required some extra-stuff > (handling optional arguments for @ commands is painful), and a > simple '@unexec [ -d ${PREFIX}/etc ] && echo ...' is easier I > guess. But that's no problem for me to include that if everybody > thinks it worth it. Given this workaround, it's probably not a high priority to add this. Hmm, what about a seperate @echoifexists or similar command? > >Have you thought about how to solve the boot strapping problems with > >pkg_install/pkg_delete? > > I have absolutely no idea what you're talking about, I started > looking at pkg_install source yesterday at night. Could you > give me some pointers about that ? The issue is that you need to find a way to keep users from installing packages they can't uninstall. If you add new commands and they are used in ports, users with older systems won't have the necessicary pkg_delete commands to make them work. The current system doens't even give a graceful way of detecting this condition both in the port and when the users installs a pkg from the -stable collections online. Longer term, we need some versioning in the plist and ports, but first we've got to solve the problem we're stuck with now. > >Our nominal pkg_install maintainer is MIA at the moment. > > Ok, actually I knew eik has been working on it, but I didn't > know who was the active maintainer now. Last I heard, eik was the one working it, but no one has heard from him in a while. He's been gone long enough that someone else could certaintly commit to pkg_install given public review. > I have thought of a new purge command, that would act like > dpkg --purge on Debian but AFAIK that would be impossible > since it would need persistent package records (that still > exists after a package has been removed as long as we have > some configuration files for this port in the tree). This would be a really cool feature. Off hand, you'd probably want to create another dirctory under /var/db to store these records. That would certaintly be allowed to support such a feature. -- Brooks