From owner-freebsd-current Thu May 1 16:28:41 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id QAA24841 for current-outgoing; Thu, 1 May 1997 16:28:41 -0700 (PDT) Received: from time.cdrom.com (time.cdrom.com [204.216.27.226]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id QAA24836 for ; Thu, 1 May 1997 16:28:38 -0700 (PDT) Received: from time.cdrom.com (localhost.cdrom.com [127.0.0.1]) by time.cdrom.com (8.8.5/8.6.9) with ESMTP id QAA03418; Thu, 1 May 1997 16:28:34 -0700 (PDT) To: Richard Wackerbarth cc: Bruce Evans , current@FreeBSD.ORG Subject: Re: -current build is now broken.. In-reply-to: Your message of "Thu, 01 May 1997 16:58:13 CDT." Date: Thu, 01 May 1997 16:28:33 -0700 Message-ID: <3416.862529313@time.cdrom.com> From: "Jordan K. Hubbard" Sender: owner-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > Unfortunately, when I really had time to work on this, I was "shot down". Now Oh grrr, please don't start this again, Richard. You made certain demands as prerequisites to doing this work at all and we in the core team considered your demands to be bogus and said so. You then evidently weren't willing to do it unless your demands were met so took your ball and went home. End of story and entirely your decision. We have *always* considered the effort to fix the many known (from practically day one) build system bogosities to be worthwhile and we've only lacked the time to do so, not the desire. We'd also love to have someone dive in and start working seriously on it, just so long as they also fully understood that the implementation methods chosen *must* move the system forward in one single leap which leaves it functioning after the transition, it cannot simply be done incrementally to the existing -current tree as some have erroneously thought in the past. Why can't it be done incrementally? Because: 1. If someone from outside the core team, say Richard here, tries to do this then they will almost certainly be missing two very crucial elements which are necessary prerequisites for something like this to ever possibly succeed: o The complete trust of the FreeBSD development community (e.g. we trust you enough to play with the build system, essentially part of FreeBSD's CNS, and not break it). o A very detailed, public roadmap describing in painstaking detail what was planned and how it would be implemented. Just going through the nit-picking stage on such a document (and I think we're seriously talking about 300-400 pages if done to full "ISO 9000 spec" here) would probably take close to a year before everyone finally ratified it and construction could begin. 2. If someone from inside the core team tries to do this then even though they might have the trust of the development community, perhaps even so much as to waive the roadmap requirements (a long history of public successes can often be substituted for detailed plans, as good or bad as that might be :-), they'll only be nit-picked to death so badly by the other core team members during the implementation phase that they'll eventually throw up their hands and say "OK! OK!! Yuu guys want to drive? Fine! Drive! I'm outta here!!" [other core team members: "Aieeeee!! Somebody grab the wheel, the bastard's bailing out the window and we're doing 80MPH!"] So it has to be done all at once, with all the technology available up-front for inspection before adoption (and some reasonable plans for integration as I would hardly expect such a step to be painless). Sure, you're still going to have to deal with some of the same mistrust and nit-picking issues you would have gotten had you done it incrementally, but at least you've now got fait-accompli on your side ("It's *done*, damnit! You just need to take it now!") and you can demonstrate the system in action to any skeptical viewers, perhaps winning over enough of them that you start to get a chorus of voices shouting: "It's *done*, damnit! You just need to take it now!" That's definitely one of the ways that things get into the system. Finally, I see nothing wrong with multiple people working on this, perhaps trying different proof-of-concept implementations in the privacy of their own homes (and what goes on behind locked doors is none of my business). If someone comes up with something they truly deem presentable, they're also welcome to present it and the users are, in turn, welcome to either champion its introduction by popular acclaim (and trust me, a truly fine system which generated raves *would* be adopted) or hold their noses and go "Eueeggh! What were you thinking, you animal!" It's software darwinism at its finest. :) Jordan