Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msunews!agate!violet.berkeley.edu!jkh From: jkh@violet.berkeley.edu (Jordan K. Hubbard) Newsgroups: comp.unix.bsd Subject: Re: Linux vs. BSD?! Date: 20 Feb 1995 15:55:21 GMT Organization: University of California, Berkeley Lines: 174 Message-ID: <3iae19$8do@agate.berkeley.edu> References: <3i7ar8$ahv@marton.hsr.no> <3i83js$avl@ivory.lm.com> <3i9aa3$sbp@fido.asd.sgi.com> NNTP-Posting-Host: violet.berkeley.edu In article <3i9aa3$sbp@fido.asd.sgi.com>, Larry McVoy <lm@slovax.engr.sgi.com> wrote: >reality. Some points to consider (and while you are reading this please >note that I was a died in the wool BSD bigot; I was one of the kernel >hackers at Sun during the SunOS 4.x (the last BSD based SunOS) days. >I loved that OS.) > [Linux points deleted] Well, I gotta say - if it were anybody but Larry making these comments then I probably wouldn't even reply, but I feel compelled to correct what I feel are some overly simplistic generalizations contained herein. Oh yes, let me also first add the disclaimer that although I'm deeply involved with FreeBSD, I also highly respect what Linux has accomplished and consider Linus a friend. He and his band of merry men have pulled off a real coup, and they deserve some serious kudos for it. Nonetheless, let's not give credit (or take it away from BSD) where it's not really due. > . Given the volumes, there will be more commercial applications > for Linux than for BSD. I'm not so sure of that. Now? Maybe. However, in the long run a number of companies will be a lot more eager to play footsie with the BSD folks than I think were initially prepared to do so with Linux. For one thing, BSD is generally an easier port. Sure, Linux has a lot of great POSIX features, but BSD in all of its incarnations formed the mainstay of commercial Unix platforms for a *long time*. There's a lot of SunOS code out there.. There's also our friend the GPL, which despite further clarifications with the GLPL and what to do about linking with GPL'd libraries or tools, does scare a lot of companies. The legal pitfalls lying in wait for your average software developer are enough of a challenge without adding things like the GPL to the pot. Don't get me wrong, for many years I was a great champion of the GPL and even released a fair amount of my own work under it (xinfo, awl, xtrlib, etc..) but I've come to see in recent years that it's created new restrictions of its own. More on that later. > . Linux is covered by the GNU copyleft. Users that have been > bitten by braindead management decisions (such as Sun and DEC > users) may appreciate an OS that is truly open. *BSD most > certainly are not. Even the original BSD crowd, now at BSDI, > are not shipping all their source. That is, quite frankly, an utterly ridiculous argument. To paraphrase Forest Gump, "Open is as Open does." *BSD most certainly ARE open, thank you very much, and they have every intention of staying this way. Do you see any source code missing from FreeBSD or NetBSD? Do you see any way of denying source code to anyone who wants it now that Novell has stood down with their lawsuit in the post-4.4 era? So let's say some company in Moosefart, Wisconsin decides to form "GreedBSD, Inc." and turns totally commercial with it. So what? The other *BSD groups go on churning out their product and nobody needs to know or even care that GreedBSD even exists. "Oh!" you say - "Wait a minute! But what if GreedBSD develops the Wonder Hack and then refuses to give it to us!" Well, again, I say "so what?" Do you think they'd have developed it at all if they didn't have a vested interest? What's the difference between GreedBSD forming and creating the hack and GreedBSD being scared away by the GPL and never forming to do the hack in the first place? Either way, there's no hack available and at least in the former case you've got an EXAMPLE for others to follow if they want to do their own free implementations of it. In fact, I would argue that I'd much rather have a couple of GreedBSD companies around than none at all. They all know which side their bread is buttered on, and if they're getting all their base technology from us then it's only in their best interest to work with us on folding their bug fixes in. This gives me the equivalent of a couple of paid FreeBSD programmers working in those companies and I'm certainly not going to kick 'em out of bed for eating salt crackers! Finally, if it's any consolation, The FreeBSD Project is now incorporating as a non-profit organization strictly so that the overall direction of the project remains in nice, altruistic and unimpeachable hands. It continues its mission of providing an Open System, as you liked to call it (though I have grown to hate that phrase and would just like to remind everyone that Larry used it first! :-), and with the framework necessary to ensure that there's some actual future for it. > . DEC is paying DEC engineers to provide Linux for the DEC Alpha. Um, actually, that's a skunk-works project and you know it, Larry! I've got people working on FreeBSD inside of Sun, for that matter, but I don't presume to say that Sun is now officially lining up behind the project. The same goes for many other large companies that suffer the presence of a few maverick groups internally going off in their own directions on something. It doesn't mean that Palmer is going to be announcing the death of OSF/1 and the birth of Linux as DEC's new official OS, now does it. > . Novell's former CEO, Ray Noorda, just spun off a company to do > Linux. Yeah yeah.. Let's just see where it goes, OK? And if what comes out of the project ends up looking anything like Linux, for that matter. >On the other hand, as these things become known (and, more importantly, >become a real problem - the run queue thing isn't an issue for your >average workstation but is for a internet server provider), they get >fixed. I'm constantly amazed at the rate at which things get addressed >in Linux. As am I. But allow me to indulge in a little bit of doomsaying here at the possible risk of pissing off Linus if he's reading this. I think he sort of knows this already.. As Linux gets larger, its guerrilla development strategy will begin to work in subtle (and probably not-so-subtle) ways against it, hindering its progress and causing a credibility gap to widen in certain circles. A lack of central organization is no big deal during the forming and storming stages (to quote Crosby), and you get lots of great ideas as a result of the pot boiling merrily away with lots of different competing groups (and believe you me, they DO compete!) all rolling out their own flavors of Linux-du-jour. But as the initial glow fades and people start putting down their tools for a moment to rest and reflect on just what they're doing, it becomes less cute of a situation to deal with. Where do you contribute infrastructure changes? Whom do you even argue them with? Patrick Volkerding and his merry Slackware crew? Adam Richter at Yggdrasil? Wait, he's commercial, right? What about those Debian folks - are they still around? Who are these "bogus" people? Can I talk Matt Welsh into championing my cause? I can't just talk it over with Linus, he only deals with the kernel and besides he's way too busy anyway! Sure, you probably have fair answers to all those questions since you've been in the thick of it for so long, but how does Joe Average figure all this out? Or the company that really wants to commit to Linux but just isn't having much luck figuring out exactly what that MEANS? Just WHO do they support? How long can those people be counted on to be around? Are they tax deductible? What happens if they die in a fatal juggling accident - does the company need to then start over with an entirely different group? These are the kinds of problems that will become increasingly irksome as Linux becomes more successful, to say nothing about bug tracking and source control. The *BSD people take the use of tools like CVS and GNATS almost completely for granted when it comes to creating the kind of infrastructure necessary to REALLY support a customer base, but the Linux people take an almost perverse pride in NOT using such things and I think that this can only come back to bite them hard on a sensitive part of the anatomy. When I attended the Linux symposium in Amsterdam last November, I remember sitting with folks like Patrick and Matt and talking about how to solve certain problems like generating slipstream releases for important strategic partners. I mentioned that we in the FreeBSD group would simply do something like check out a source at the last major release date, bring the changes we wanted from some later dated back into it, and then do a `make release' from the top to roll a complete distribution with just those fixes in it. They looked at me owl-eyed and sort of shook their heads. When I pressed them, it came out that they accomplished the same thing by just pulling down a complete dist off the net someplace and then scouting around for the various patch files and then going through and trying to identify those bits of the patch files that were relevant. Talk about labor intensive! When I laughed and said that they were a bunch of crazy cowboys, I could tell that they were actually somewhat pleased by the description. I think that pretty much sums it all up right there! P.S. Larry, there's one crucial difference between us and Sun dropping SunOS. Sun never released the source code to SunOS to each and every customer. If they had, I can assure you that things would have turned out quite a bit differently! The users would have enhanced it 8 ways to Sunday and Solaris would have been greeted by general laughter, SunOS no doubt already ported and running on Sun's newest hardware by its users before Sun even _began_ to work the bugs out of their own product. So run into the arms of the GPL if you must, but not for that reason. With *BSD, the genie is already out of the bottle and no number of short-sighted or small-minded marketdroids are ever going to manage to stuff it back in again. Jordan