From owner-freebsd-questions Wed Jan 24 16: 4:13 2001 Delivered-To: freebsd-questions@freebsd.org Received: from harumscarum.mr.itd.umich.edu (harumscarum.mr.itd.umich.edu [141.211.125.17]) by hub.freebsd.org (Postfix) with ESMTP id BD57337B404 for ; Wed, 24 Jan 2001 16:03:55 -0800 (PST) Received: from tim.elnsng1.mi.home.com (c1129767-a.elnsng1.mi.home.com [24.183.248.20]) by harumscarum.mr.itd.umich.edu (8.9.3/3.3s) with SMTP id TAA05272; Wed, 24 Jan 2001 19:03:45 -0500 (EST) From: Tim McMillen To: Roelof Osinga , Mike Meyer Subject: Re: OT: non-Unix history (Was: FreeBSD vs linux) Date: Wed, 24 Jan 2001 19:08:02 -0500 X-Mailer: KMail [version 1.1.99] Content-Type: text/plain; charset="us-ascii" Cc: "Albert D. Cahalan" , questions@FreeBSD.ORG References: <14957.31196.939559.889627@guru.mired.org> <14959.23870.728403.859934@guru.mired.org> <3A6F61DC.39E9CF0D@nisser.com> In-Reply-To: <3A6F61DC.39E9CF0D@nisser.com> MIME-Version: 1.0 Message-Id: <01012419080209.24525@tim.elnsng1.mi.home.com> Content-Transfer-Encoding: 8bit Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Ok, sorry this is way off topic to -questions, and I like to avoid that, but since the discussion is already here... I suppose beyond this we should cc -chat and drop -questions. On Wednesday January 24, 2001 18:14, Roelof Osinga wrote: > Mike Meyer wrote: > > ... > > Actually, Ultrix is a BSD derivative, and didn't require a SysV > > license. But they changed it later. Ditto for Sun and Mips - both > > started on BSD with a SysIII license, then went to SysV. But The > > Unix market splintering that way was what "the kernel APIs being > > different" was referring to. You couldn't port an application to > > Unix - you had to port it to each variant. Since VMS - and later NT > > - were usually a larger market than any single Unix vendor, even if > > it wasn't as big as all of them put together, it got preference. > > That does not mean I agree on the point of API difference. Apps > got ported allright, back then. Many did. But empirically many vendors either gave up support for Unix altogether or only supported a few Unix types because it was simply too expensive to provide support for them all. Do you think anyone really develops for Windows because of it's technical superiority? Of course not. They develop for windows because of its ubiquity. They can write the program once, learn on platform to support it on and have acces to a large market. Sure porting the software is not that expensive (it IS an expense though), but the support gets very expensive. You have to hire people that know the ins and outs of each platform that your software runs on (if you want to do it right that is.) I believe that is the biggest reason Unix did not succeed more than it did. I also believe it is the biggest thing holding it back now. Unix was obviously (and still probably is) far technically superior to Windows, but it was the universality of the Windows API that allowed them to capture market share. Add that to extinguishing competition and that allowed MS to gain a monopoly with Windows. Right now it's far from trivial to even write an app that works on all of the Linux distributions. Then look at the fact that you have to have another one for each of FreeBSD, OpenBSD, IRIX, HP-UX, SCO, etc to even have support for most of the Unix platforms. That, unfortunately makes it economically sensible to develop for Windows, becuase it give you access to a bigger market with less expense, even though the platform is technically inferior to pretty much any Unix. That is what has created the lock in that we have into the technically inferior OS. Free competition has failed to result in the best solution in this case. (and it does in many cases of infrastructure--like the CA power situation. I'll save my rant on that for another forum, and I may even write up an econ paper on it.) The best case would be if there was a unified API (or ABI, I don't understand the diff) that all applications could be written for. This would of course result in compromise since no one API can be best for all things, but having one API that all applications were written for would force Windows and Word to compete an an even field with other OS's and Apps. It wouldn't take long for them either fail, or shape up in order to compete. Unfortunately I don't think this is possible, but who knows. If even major opensource OS's that try to do The Right Thing (tm) can't agree on a single API due to technical issues and differences of opinion, then how could commercial vendors do it also? Tim To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message