From owner-freebsd-ports@FreeBSD.ORG Wed Sep 3 15:15:39 2014 Return-Path: Delivered-To: ports@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id CDC9B765; Wed, 3 Sep 2014 15:15:39 +0000 (UTC) Received: from hades.sorbs.net (hades.sorbs.net [67.231.146.201]) by mx1.freebsd.org (Postfix) with ESMTP id AE4B818AD; Wed, 3 Sep 2014 15:15:39 +0000 (UTC) MIME-version: 1.0 Content-transfer-encoding: 7BIT Content-type: text/plain; CHARSET=US-ASCII Received: from isux.com (firewall.isux.com [213.165.190.213]) by hades.sorbs.net (Oracle Communications Messaging Server 7.0.5.29.0 64bit (built Jul 9 2013)) with ESMTPSA id <0NBB00429ZWFEV00@hades.sorbs.net>; Wed, 03 Sep 2014 08:19:29 -0700 (PDT) Message-id: <54073097.6000006@sorbs.net> Date: Wed, 03 Sep 2014 17:15:35 +0200 From: Michelle Sullivan User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.1.24) Gecko/20100301 SeaMonkey/1.1.19 To: Tom Evans Subject: Re: [HEADSUP] pkg(8) is now the only package management tool References: <20140901195520.GB77917@ivaldir.etoilebsd.net> <540522A3.9050506@sorbs.net> <54052891.5000104@my.hennepintech.edu> <54052DFA.4030808@freebsd.org> <54053372.6020009@my.hennepintech.edu> <5405890F.8080804@freebsd.org> <20140902125256.Horde.uv31ztwymThxUZ-OYPQoBw1@webmail.df.eu> <5405AE54.60809@sorbs.net> <1D2B4A91-E76C-43A0-BE75-D926357EF1AF@gmail.com> <5405E4F5.4090902@sorbs.net> <5406BD65.705@digsys.bg> <5406ED34.7090301@sorbs.net> <5406F00C.6090504@digsys.bg> <358B9E99-5E02-47BA-9E30-045986150966@gromit.dlib.vt.edu> <540711FF.3050409@sorbs.net> <47F4AAAA-2D88-4F03-8602-880C4B129305@gromit.dlib.vt.edu> <54072011.7030800@sorbs.net> <540723EC.5000908@sorbs.net> In-reply-to: Cc: "ports@freebsd.org" , freebsd-stable X-BeenThere: freebsd-ports@freebsd.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: Porting software to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 03 Sep 2014 15:15:40 -0000 Tom Evans wrote: > On Wed, Sep 3, 2014 at 3:21 PM, Michelle Sullivan wrote: > >> Tom Evans wrote: >> >>> On Wed, Sep 3, 2014 at 3:05 PM, Michelle Sullivan wrote: >>> >>> >>>> I think portsnap should provide 'stable' - tested, known >>>> working, security patched... >>>> >>>> >>> 100%, and as soon as someone comes along who is prepared to do and pay >>> for that, I think we would all enjoy it. >>> >>> Unfortunately, someone like that doesn't yet exist, so it is >>> unrealistic to just expect that infrastructure to be there. >>> >>> >> Well as I was one of the people trying to raise funds for FreeBSD (for >> general stuff, not specifically this) and as $employer will *not* be >> adopting FreeBSD now the chances of having such just reduced. >> >> > > That's a fallacious argument; "if *someone* doesn't put the > infrastructure in to place then *we* can't contribute more". > Didn't say that, though I can see how it looks like that (because you're not taking into account other emails.) If the system had not been broken over night I would still be well on to my way of to getting corporate support, but with echos of Mandrake (co-incidentally produced in the same country as the latest breakage) they're going to stay with enterprise OSs. > This is what Linux distributions spend their money on; employing > people to do infrastructure engineering. When a new release of httpd > happens, people at Red Hat manually back-merge fixes to the version of > httpd that is in their package repository. > > FreeBSD has volunteers who maintain the ports tree, they have no time > to manually merge and test fixes, so when a new release of httpd > happens in FreeBSD, the version changes and you get all the new > features and bug fixes. > Nope, as an ex-maintainer I can vouch for this. > So if you use FreeBSD, that infrastructure is not there; you need to > do it in house. Like I was doing. (and am not anymore because the entire build system is now screwed and I'm not going to build it again.) > How tricky that is depends on the size of your house - > Netflix have no problems, Yahoo have no problems, SMEs like the one I > work for - problems. > > I'm not denying the problem; just that specifying what should or > shouldn't happen with the ports tree is not productive if you aren't > proposing to actually do it yourself. > > Here's the problem, I saw the EOL last October, I read it, I understood that after Sept 1, 2014 the old packaging system will no long be supported as it's "EOL" I continued the production database upgrade. I continued my building of a completely new Puppet Environment for the production servers. I learned and built my own build system using jenkings, virtualbox and poudriere. I even built it so it would build both pkgng and pkg_* versions of the repos... and with the pkgng without docs (just reverse engineering the public FreeBSD pkg system.) I continued freebsd-update'ing production servers to 9.2 then 9.3 where possible (testing, and doing it all by hand) then integrating it into puppet - in many cases writing my own puppet modules and patches to make it work with FreeBSD (like the facter patch that gives interface aliases.) I learned that the ports tree was being updated so much that things would break every day in just 580 packages I have. I changed the system so it would only start building when triggered and would continually cycle until it got a complete and stable repo (with regression testing)... sometimes this took over a month to get stable (mostly just a few days.) I took over maintainership of some ports to get staging done and to help others (including virtuoso - which was no small task and something which I don't use at all - amongst others I don't use.) At this point (July/August 2014) I saw a convo between bapt and someone else that led me to question , "So Sept 1, 2014 the entire ports builds will change and pkg_* will be completely broken" .. - not just EOL, but updated so they no longer work at all Then around mid August after some patches had *finally* been applied I triggered a new build which continued to cycle due to a bad TCL update until August 30 when it 'fixed itself' ... but the build continued to slowly make its way through got to 9.2-i386 on Sept 1... and guess what... something caused it to restart because of a bad update and so Sept 2 came and bapt deliberately and knowingly broke pkg_* in the ports tree and my entire repo for non-pkgng started building itself for pkg ... which means my environment cannot be tested, so it'll never complete and I can not upgrade without going to every server manually... not to mention I have to rebuild the build environment completely to make it work now, then I have to rebuild the testing environment to cope with the switch from pkg_* to pkg, and then finally I have to switch all servers to pkg and tell puppet to use pkg instead of pkg_* ... I've had about 2 months - even with keeping up the notices - and all I got was, "No, you should plan better"... Well that's a real good way of getting people who can help to keep helping... FUCK THAT! Michelle -- Michelle Sullivan http://www.mhix.org/