Date: Wed, 30 Aug 2000 22:52:21 +0200 From: Poul-Henning Kamp <phk@critter.freebsd.dk> To: Jordan Hubbard <jkh@winston.osd.bsdi.com> Cc: Bill Fumerola <billf@chimesnet.com>, arch@FreeBSD.ORG Subject: Re: build tools as separate distribution Message-ID: <22213.967668741@critter> In-Reply-To: Your message of "Wed, 30 Aug 2000 13:25:43 PDT." <87939.967667143@winston.osd.bsdi.com>
next in thread | previous in thread | raw e-mail | index | archive | help
>What I've said, and more times than I can count now, is that something >new and better is NEEDED. More recently, I've also pointed out the >existence of the libh project as one potential candidate for filling >that need. As one of the original perpetrators and in fact the one who laid the foundations of sysinstall and "make release" where they lie to this day and then abandonned it to Jordan to maintain the monster, and as a card-carrying old asshole I would like to make a number of observations on this topic: 1. Yes sysinstall is not a pretty sight. (On the other hand, sysinstall was written to run in a rather "DIY" environment and therefore many things were initially not even possible which has latter become possible due to more RAM in the minimum configuration. Remember how long you could install FreeBSD with a single floppy ?) 2. Yes it would be nice if we had a well structured, 5th generation, object-oriented, modular, configurable, scripted, extensible, and provably correct install program. 3. Yes there are things sysinstall doesn't do very well. BUT: 4. Sysinstall does a hell of a good job for a lot of people and has done so for many years by now. They are used to it and know how to get the job done with it. 5. Any rewrite of sysinstall is bound to suffer "second systems syndrome" in fact Jordans evangelism for the project over the last 5 years to some extent proves this already: Anything rutinely called "the all singing and dancing foo" is bound to suffer from 2SS. (If you don't know what 2SS is, you need to read "The Mythical Man-Month" by Fred. P. Brooks). 6. Unless significant perceived improvement over the current sysinstall is realized, a transistion to a new tool faces a significant probability of a "New Coke" fate: "Why on earth did they fuck with this stuff, it worked fine ?!!!" Until now, I have always assumed that observation number 5 would prevent the sysinstallNG from ever being completed (seems I was right so far) and as a backup I assumed that observation number 6 would effectively kill it in early infancy if it did in fact carry to term with number 4 nailing the lid in the coffin firmly shut. I may be totally wrong. I just happen to remember how long time it took before the last people stopped telling us how much better the FreeBSD 1.X shellscripts were at installing FreeBSD than the "monster" we had created. No, I have not even looked at libh. I don't know what it can, what it does or how it does it. And I don't intend to find out unless I absolutely bloody have to. I helped write sysinstall, somebody else do the next one. With that off my chest: I want to wish the people who decide to tackle this task a well meant "Good luck - enjoy it!" I personally found it a very interesting and challenging task to write sysinstall, and I think you will find it a very very interesting task, only too bad the bar is so much higher to clear for you than it was for me, but so is life. You have my source and shoulders to stand on, so it should not be impossible, only just very very hard. Poul-Henning, Release-engineer FreeBSD 2.0 etc etc -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD coreteam member | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?22213.967668741>