LWN.net Logo

Larry McVoy on BitKeeper, kernel development, Linus Torvalds and Bruce Perens (LinuxWorld)

Joe Barr talks with Larry McVoy in this LinuxWorld article. "McVoy's biggest contribution to free software may be BitKeeper, his proprietary source management system. The story of how BitKeeper has come to be Torvalds' (and many other kernel hackers) tool of choice in maintaining the Linux development tree is worthy of a book. It's not just an unlikely outcome, given the animosity that often flares up when proprietary and open source types gather in the same space, it has been a frustratingly painful one. McVoy tells me that it was his desire to help Linus that has resulted, to use his own words, in "a miserable last five years.""
(Log in to post comments)

Larry McVoy on BitKeeper, kernel development, Linus Torvalds and Bruce Perens (LinuxWorld)

Posted Jan 28, 2003 12:52 UTC (Tue) by leandro (guest, #1460) [Link]

What this guy does is disgusting. He wants to be in the position of sole proprietary vendor in a free software world. Even more appalling is that people fall from him.

Suffice it so say that, if too many free software developers thought like him, we would have had no Linux. Linus could say, See, I am supporting the GNU system with my free kernel. Either pay for it or you cannot use it to develop the Hurd, and (put additional restrictions here).

Larry McVoy on BitKeeper, kernel development, Linus Torvalds and Bruce Perens (LinuxWorld)

Posted Jan 28, 2003 15:57 UTC (Tue) by dang (guest, #310) [Link]

Bull. He created a really useful tool that has really streamlined the kernel patch process and he has done it in such a way that allows him to make it available freely for GNU project while still being able to charge for proprietary project. Think of it as a service version of the LGPL and you can even see that it is GNU at heart. I fail to see the problem. Moreover I see zero evidence to support your claim ( which to me looks borderline libelous ) that he seeks a monopoly on anything. In the case of kernel development, the post-bitkeeper world is one in which people who want to use BK can and in which people who don't want to use BK can use what they will.

And as for "disgust," get over yourself already.

One thing I don't understand about what McVoy's doing Perens (LinuxWorld)

Posted Jan 28, 2003 12:57 UTC (Tue) by rknop (guest, #66) [Link]

I haven't followed BitKeeper that closely, but an obvious question comes up in my mind.

McVoy claims that he wants to support everybody: free software, also vendors and closed-source software. He also says that he's identified something which the closed-source people want but the free software people don't need: code privacy.

As such, why can't he just release two versions? Release the free software version *without* code privacy under a truly free licence (even the GPL), and sell the version with code privacy under a proprietary licence to the people who want it? Some who misunderstand the GPL would say this is not legal, but it is: if his company owns the copyright, they can release the code under as many different licences as they want. How would his doing this not accomplish what he claims in the article to want to accomplish? Plus, it would silence the flames once and for all.

-Rob

One thing I don't understand about what McVoy's doing - Perens (LinuxWorld)

Posted Jan 28, 2003 16:00 UTC (Tue) by lm (guest, #6402) [Link]

We tried pretty close to what you are suggesting, we used to release source
under a license which said "you can do pretty much anything you want except
remove the openlogging features". The default target in the Makefile displayed the license.

The first thing people did was to remove the openlogging stuff in direct
violation of the license.

The thing you are missing is that if we release under any source license at
all we make it possible for people to remove the very thing which provides
us with a revenue stream.

One thing I don't understand about what McVoy's doing - Perens (LinuxWorld)

Posted Jan 28, 2003 16:35 UTC (Tue) by rknop (guest, #66) [Link]

OK, my misunderstanding was that I thought it was something *added* to the commercial version, rather than something present in the FreeBeer version the removal of which made it work for the commercial version.

-Rob

One thing I don't understand about what McVoy's doing Perens (LinuxWorld)

Posted Jan 28, 2003 16:08 UTC (Tue) by jamesh (guest, #1159) [Link]

Probably because a company wanting privacy could take the GPL version and remove the change set advertising code, which would remove the incentive to buy the proprietary version. The code managed by an SCM is not a derived work of the SCM, so the GPL'd version wouldn't place any restrictions on the managed code.

Larry McVoy on BitKeeper, kernel development, Linus Torvalds and BrucePerens (LinuxWorld)

Posted Jan 28, 2003 14:57 UTC (Tue) by MarkSwanson (guest, #9328) [Link]

I've written my first Linux device driver in 1993 so I'm not new to Linux
or open source. I've also read a fair bit about Larry and Bitkeeper over
the years.

A couple of questions come to mind:

1. It's Larry's company. Why should anyone begrudge him the right to make
a profit? It's Larry's choice to give away the source for free (or not) as
much as it is your choice to choose your religion. Why dislike him for
that?

2. Larry has spent a lot of money developing a product that free software
developers can use for free. Shouldn't we be thankful?

3. Charging for goods is a proven business model. On planet Earth - given
all of its incredibly vast and experienced free software developers - the
only way a product like BitKeeper could be born, grow, and survive was by
using its current business model. CVS (arch, et. all) despite being useful
are not bitkeeper. If the history of CVS development is any indication
they never will be. Shouldn't we learn from this?

4. The only way free software has made money is through the consulting
model. Can every software product morph to fit into the consulting model?
Can every developer/company fit into the consulting model?

Free software is great and I devote a lot of my personal and professional
time to it. However, through years of experience I have witnessed serious
morality problems with the consulting business model. I believe it has
fundamental sales and marketing issues that can not be avoided and only
grow worse the larger a company gets. I believe that in the real world not
every developer/product/company can or should give away their IP and
become a consulting company.

I do not begrudge Larry for making a living selling proprietary software.
I hope no one will begrudge me for it either.

Backstabbing. Perens (LinuxWorld)

Posted Jan 28, 2003 20:27 UTC (Tue) by ncm (subscriber, #165) [Link]

I was with Larry all the way, until recently.

He certainly has every right to run his business his way. In one sense, if he offers something to help out Free Software, we can take his help or leave it, on the terms he spells out, and be as grateful or indifferent as we like. That's not the whole story, though. Like the rest of us, Larry has benefitted overwhelmingly by the generosity of others in the Free Software movement. Like the rest of us, he has been given so much that there is no possibility of his ever giving back anything close to what he has received. So, in a strict legal sense, while BK is his property to take or leave, it takes disgraceful nerve to demand the special treatment he does for his contribution.

What changed my mind about him personally was when he stiff-armed the very people that he thanks so effusively in this interview. I'm referring to his changing the BK license, after the kernel team had become dependent on the software, to exclude anyone who (or whose employer!) might contribute to any potentially competing product. That's a bait-and-switch tactic, and even if no fraud statute applies, it's deeply unethical.

If BK is better than the alternatives, it will command a price. If it's not better, he doesn't deserve the price. Playing nasty license tricks is Microsoft's game. It fouls the well. Anybody who has contributed any substantial amount of code to the Linux kernel, or to most other Free Software products (FSF/GNU products excepted), is free to change the license to that code to exclude Larry and his customers from the right to use that code. I would hate to see that start to happen, but that is the kind of food-fight that Larry has started.

It's not too late for him to apologize and acknowledge his mistake.

Backstabbing (NOT) Perens (LinuxWorld)

Posted Jan 28, 2003 21:58 UTC (Tue) by lm (guest, #6402) [Link]

Hey Ncm, I'm sorry you feel that way. The reason that clause is in there has nothing to do with the kernel guys, it has to do with sourceforge. The sourceforge team had integrated BK into sourceforge without working out a deal with us and given that they were headed down a path to compete with us we did not feel that it was prudent to allow them free access to our technology.

In general, I question your outrage. If you asked IBM for an eval license for any product, they asked you what you wanted it for, you tell them that you want to use it to learn how to copy it and release a free version, then do you believe that IBM should grant you that license? The BitKeeper free use license is a lot like an eval license where we don't have the chance to filter out the people who are trying to do material harm to our business. That means that the license itself has to spell it out.

Almost all products have licenses which say you may not reverse engineer, etc. No company stays in business if they license their products in a way which allows other entities to copy those products.

As a business, we are committed to doing whatever it takes to stay healthy. As the CEO of the business, I am committed to helping the kernel folks and the open source community. You may disagree with how we are going about it, in which case you are free to demonstrate a better path. Until you do, however, we'll take the path which we believe works for as many people as possible. The fact that the path doesn't include people who want to steal our technology is a problem for those people and those people only.

Backstabbing (NOT) Perens (LinuxWorld)

Posted Jan 30, 2003 12:36 UTC (Thu) by rknop (guest, #66) [Link]

No company stays in business if they license their products in a way which allows other entities to copy those products.

I hope you're wrong; otherwise, Free Software will return to the fringe of the hacker (meaning that you have to be very careful to find what hardware is supported, and forget data formats), as there will be no companies out there that can really produce it.

There are currently counter-examples (such as RedHat), and I hope that counter examples continue to exist indefinitely. Otherwise I will be sad.

-Rob

Backstabbing (NOT)Perens (LinuxWorld)

Posted Jan 31, 2003 18:51 UTC (Fri) by xoddam (subscriber, #2322) [Link]

You're saying that a re-implementation contravenes the use
provisions of the license. Isn't there something in the
Fair Use doctrine which allows reverse engineering for
the purposes of interoperability?

You're free to release your software under whatever licence
you like and you are entitled to the same legal enforcement
of that license as any other copyright holder -- but no more.
So long as the law allows reverse engineering (in some
jurisdiction), your license clause which says Thou Shalt Not
is unenforceable (in that jurisdiction).

I'm firmly in favour of Fair Use, as I think most people who
use or develop Free Software are. As for 'stealing technology',
it's an *idea*, man, kudos for you for coming up with it and
shame on anyone who uses it without acknowledging you. But
you don't *own* it!

Use-restricted licenses

Posted Feb 1, 2003 12:17 UTC (Sat) by Peter (guest, #1127) [Link]

You're saying that a re-implementation contravenes the use provisions of the license. Isn't there something in the Fair Use doctrine which allows reverse engineering for the purposes of interoperability?

You might think fair use would apply in a number of situations where it apparently doesn't. For example, if you buy a copy of Microsoft Windows NT Workstation, you might consider it fair use to install your copy on your computer and use it to run a high-volume web server. You paid for your copy, so you might think it would be fair use to install and use it on your computer - that's about all you can do with it, after all - and beyond that, it's none of Microsoft's business what you do with it (so long as you don't redistribute it, or make a derived work, either of which would go against copyright law).

Turns out the MS end-user license agreement says you can't do that. You were supposed to buy NT Server instead.

Or download the free personal version of JBuilder. Page through the license and read all about how Borland or whatever they are calling themselves today don't think you should have the right to use copyrighted material you legally downloaded for the purpose of building commercial projects.

The point is, arbitrary use restrictions in end-user licenses are nothing new. Larry didn't invent them. I personally think they are severely stretching enforceability under copyright law, but apparently Silicon Valley either disagrees or is willing to bluff it until a court weighs in. And so, this practice probably won't be challenged in court, or if it is, many courts would choose to avoid the perceived mayhem that would be caused by invalidating such use restrictions. Undeniably, this would be a large boat to rock.

(NOTE: even if such restrictions are held invalid under copyright law, there is no reason software couldn't be distributed on the basis of contracts ... except that getting customers to individually sign pieces of paper before allowing them to purchase or download software would be a major PITA for most vendors. Knowing this, vendors today try to claim that click-through license dialogs actually are contracts, but that is just part of the bluff. Vendors have no right to speak for me in saying that my "clicking" is intended as a legally binding signature or even a handshake.)

Religions and examples

Posted Jan 31, 2003 23:29 UTC (Fri) by Max.Hyre (subscriber, #1054) [Link]

ncm writes:
Like the rest of us, Larry has benefitted overwhelmingly by the generosity of others in the Free Software movement. Like the rest of us, he has been given so much that there is no possibility of his ever giving back anything close to what he has received.

Dear me---does this mean everyone who uses Free Software code is obligated to GPL all her own code? Sounds awfully close to virality to me.

RMS & al. wrote the GPL to, among a number of other things, avoid precisely this claim---you can do any damn' thing you want with the code, down to and including modelling fission in Iraq.

To GPL something is to recognize that there will be those (like me)-: who take without giving, but to give anyway, perhaps to lead those others by example. Any other stance is the first step toward the extremely slippery slope of thousand-line licenses naming the people and purposes who are restricted from using the software.

Let Mr. McVoy do his thing, and you do yours, without mud-slinging. As Mr. Swanson hints at above, this is a religious matter, and your taste should have nothing to do with anyone else's actions. Instead, let everyone be gladdened by the spirit of friendship embodied in the GPL, and inspired to add her bit to the amazing stew of Free Software.

Now back to the piece I hope to toss in....

Best wishes,
Max Hyre

Copyright © 2003, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds