Date: Thu, 5 Sep 1996 13:04:31 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: msmith@atrad.adelaide.edu.au (Michael Smith) Cc: rkw@dataplex.net, msmith@atrad.adelaide.edu.au, nate@mt.sri.com, current@freebsd.org Subject: Re: Latest Current build failure Message-ID: <199609052004.NAA09730@phaeton.artisoft.com> In-Reply-To: <199609050522.OAA09648@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Sep 5, 96 02:52:13 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> with them. The FreeBSD project is (read Terry again) not > policy-driven but technology-driven. Given human nature, reality and > the current limits of the technology, policy simply gives named form > to reality. Correction: policy gives constraints on future configurations of reality. The purpose of policy is not to say "here is how things are", but instead, "here is how things should be". If policy simply codifies existing practice, then you have accomplished nothing more than en entrenchment of the status quo. This would be an error. > You are saying "I cannot do what I want", and then you are saying "the only > things that I actually want to do are the second phase". You make no sense. Actually, this makes a lot of sense to me, probably because I've had a similar experience. 1) I want to work on advanced file systems design 2) To do that, I have to fix the framework 3) Fixing the framework without first doing the advanced file systems work yields no net benefit to people who don't also want to work on advanced file systems 4) Fixing the framework is opposed as "having no immediate benefit" It is a catch-22 that prevents me from working on advanced file systems design. > "We" don't "have to" do anything. You, on the other hand, could do with > trying to make your statements mean something. So far, I read you as > saying : > > - You cannot use sup/ctm/whatever as a top-level feed to simulate the > distribution process. > (false) Doing so artificially constrains the implementation. > - Something works "by inspection". > (What? What does "bu inspection" mean anyway?) "By inspection" is a term used in the hard sciences. The Physicists (my self included) refer to it as "IOTTMCO" -- "Intuitively Obvious To The Most Casual Observer". > - It is necessary to bring all of the FreeBSD distribution process under > your heel. > (Possibly; but not a move likely to endear you to anyone) I don't see where this is a requirement. > Born to Lose, as they will be able to screw themselves just as well > attempting to build a tree that wasn't validated on their particular system > under their exact circumstances.) Yet today, air bags are required as standard equipment to enable those "born to lose" to survive to wreak even greater havoc on our roadways... it *is* possible to build safety systems. > > The real thing that is different is that "current" would no longer exist in > > quite the same way as a bunch of unsynchronized snapshots. > > If you are suggesting that I should not be able to track, as close as I > care to chose, the bleeding edge, then you can take your neat ideas, fold > them until they are all sharp corners, and attempt to pack your appendix > with them from the wrong end. You should not be permitted to do so at the expense of everyone else (by your pursuit destableizing the tree for everyone else). The point is to build this requirement into the process whereby you interact with the tree. It is a safety system. It also protects you from the ill-executed changes of others. > The current distribution system, as can be demonstrably shown, works > quite well. And here is the crux of the disagreement. The statement is true when it is modified as: "the current distribution system works quite well for those people who are members of the core team, and passably well for those people who have direct commit priviledges". CVSup fixed a number of problems for people without direct commit priveledfes, but it certainly did not fix them all. Specifically, it did not fix the problems endemic to the lack of consistency enforcement (which would cause the tree to always be buildable), nor did it fix the problem of what to do with local branches to terminate them gracefully once the differences between them and the main line source have been integrated into the main line source. > You can produce, in pilot, a _complete_duplicate_ of the _entire_ > distribution system, using exactly the same input as the real thing > has. This is available to you _right_now_. But the name OpenBSD has already been taken by a similar pilot project in reaction to similar pressures within the NetBSD "core team" framework... what would he call such a duplicate distribution? I am only half-joking here. > The "organisation" is more than willing to help. The "organization", as an entity, will not work to reduce its own power any more than government will vote to reduce its own power. The problem isn't the organization, but it's intrenchment. It is a derivative (in the mathematical sense) in the same replationship momentum bears to kinetic energy. Richard wants, what is in effect, a decentralization of authority. The authority is not to be given to Richard, it is to be dispersed. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199609052004.NAA09730>