Date: Tue, 20 Feb 2001 00:32:09 -0800 From: "Crist J. Clark" <cjclark@reflexnet.net> To: Alfred Perlstein <bright@wintelcom.net> Cc: Leif Neland <leif@neland.dk>, freebsd-current@FreeBSD.ORG Subject: Re: Do we need a 3. level between stable and cuurent? Message-ID: <20010220003209.V62368@rfx-216-196-73-168.users.reflex> In-Reply-To: <20010219103017.M6641@fw.wintelcom.net>; from bright@wintelcom.net on Mon, Feb 19, 2001 at 10:30:17AM -0800 References: <20010216025147.A25002@dragon.nuxi.com> <xzp3ddjmij6.fsf@flood.ping.uio.no> <20010216104409.A99749@enigma.whacky.net> <20010216025147.A25002@dragon.nuxi.com> <200102161703.f1GH37E29782@billy-club.village.org> <20010217101500.A10023@enigma.whacky.net> <00d001c09a94$b280d9c0$0e00a8c0@neland.dk> <20010219103017.M6641@fw.wintelcom.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Feb 19, 2001 at 10:30:17AM -0800, Alfred Perlstein wrote: > * Leif Neland <leifn@neland.dk> [010219 08:54] wrote: > > We all know: -current is bleeding edge, expect it to break at random. Don't run it if you don't know how to fix it. > > -stable is for production, it works all the time. > > > > Do we need a level in between for people who just run current for the fun of it and for testing. > > So after the hardcore has tested it in -current, they commit it to all the monkeys trying to break it, and we then try it on n^m' combinations of hardware/software. > > > > I might not be able to fix a problem, but I can report what happens, and if my -current breaks for a few days, it is no big deal. > > > > While -current is not for everybody, I believe people like me helps in quality testing before the stuff hits -stable. > > > > Perhaps not a level, just a separate file, which contained the date of the last known version without known major problems. (or "." if no known problems) > > This is a good idea, however it would take someone dedicated to > maintaining this as well as doing regression testing. Those > regression tests could easily be ported to -stable making for > happier -stable as well as -current users. > > Are you volunteering? :) IMHO, people who are not actively developing -CURRENT, those running it for fun or only occasionally hacking non-essential portions just need to realize that blindly tracking -CURRENT with a daily, weekly, or perhaps any regular make-world schedule is just not going to work. Find a reasonably stable version of -CURRENT and stick with it until you know that the current -CURRENT is also in decent shape. For example, I did a make world early in February and started getting kernel panics at boot. I tried to get some help and provide some data on freebsd-current to fix it, but I just did not have the time and tools. I ended up just CVSuping a back to a version I knew was in good shape and the box has been fine. I can use it for what I want, and I'm not upping it until I hear things have settled a bit... And I hope that's soon, I don't want it to diverge too much or else merging my changes in might get painful. What would help me and others like me is some way to actually figure out if the current -CURRENT is stable (I like that sentence). I am not saying it would be a substitute for freebsd-current and the commit list, but those can be very, very hard to follow. Rather than two versions of -CURRENT provide the data for people to fairly quickly determine if today's -CURRENT is solid enough for them. In some sense, each commit makes a new version, people need to decide up to what recent earth-shaking commit they can go to and be safe. Possibly some webpage summarizing what is broken, which essential pieces are in active development, etc. Obviously, this would take effort, but it may be worth it to those who want to track a stable -CURRENT. It is definately less work than maintaining two versions. How does it differ than reading the mailing lists? You don't need to dig through a dozen me-toos for each problem. Someone more knowledgeble on things can filter the claims that -CURRENT is broken from people who can't read UPDATING. And so on. And no, I am not volunteering. -- Crist J. Clark cjclark@alum.mit.edu To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010220003209.V62368>