Unable to buy a copy of BK [LWN.net]

Unable to buy a copy of BK

Posted Sep 26, 2004 4:11 UTC (Sun) by kevinbsmith (subscriber, #4778)
In reply to: An Interview with Tom Lord of Arch (O'ReillyNet) by lm
Parent article: An Interview with Tom Lord of Arch (O'ReillyNet)

This is really interesting!

In the past, on various discussion lists, several people have said that it's reasonable for BK to rescind the gratis license for anyone they consider to be a competitor, because that person could always choose to buy a copy. While I strongly dislike that policy, it is defensible.

Now, we discover that competitors (and apparently employers of individuals who may be competitors) cannot even *buy* a copy. While this policy is (probably) still within BK's legal rights, it certainly changes the debate.

This seems to be yet another piece of strong evidence that relying on proprietary software deep in your toolchain is risky. Especially one with a history of the license becoming more and more restrictive over time.

(Log in to post comments)

Unable to buy a copy of BK

Posted Sep 26, 2004 5:04 UTC (Sun) by lm (guest, #6402) [Link]

Sigh. You guys are missing the point. Every commercial software license in the world has words to the effect "no reverse engineering". Go look, they all have it.

That clause is far more restrictive than "you can't compete with us". It means you aren't allowed to poke at the software for *any* reason as opposed to only if you are chosing to compete with us. There is easily a 1000:1 ratio of random engineers to source management engineers, we were trying to avoid sweeping everyone into the same category.

"You can't compete" is a *subset* of no reverse engineering. The intent of that clause was to only target those people who wanted to destroy our livelihood. It was explicitly designed to leave the door open for people who had to poke at the software to do something we hadn't anticipated. In other words, the clause you hate so much was designed to help you unless you had an active goal of hurting us. Oh, my, how unreasonable of us.

Unable to buy a copy of BK

Posted Sep 26, 2004 6:08 UTC (Sun) by HoserHead (subscriber, #828) [Link]

"You can't compete" is a *subset* of no reverse engineering.

It most certainly is not. I can compete with Bitkeeper without ever looking at its output, code or feature set.

I won't comment as to the enforceability of your anti-reverse-engineering clause, because those in general have not been tested in American courts, and I'm not an American.

Unable to buy a copy of BK

Posted Sep 26, 2004 6:09 UTC (Sun) by dvdeug (subscriber, #10998) [Link]

"Reverse engineering" is not poking at the software. It's disassembling the software. So, no, "you can't compete" is not a subset of "reverse engineering". In any case, "reverse engineering" is specifically permitted, irrespective of license, in many jurisdictions, and I find it very questionable ethically to tell people they can't poke around and see how something works.

Instead of just building the best mousetrap, you want to try and hide how your mousetrap works. Legal or illegal, it doesn't help get the best mousetrap on the market, and you can't really expect us to like it.

Unable to buy a copy of BK

Posted Sep 26, 2004 7:09 UTC (Sun) by Ross (subscriber, #4065) [Link]

On the meaning of reverse engineering, it depends on who you ask. In the
academic and most traditional meaning of the word it means any process to
turn a working product back into specifications, normally with the
intention to use those to produce another product with some or all
aspects of the original.

But yes, most people just use it to mean "disassemble".

Unable to buy a copy of BK

Posted Sep 26, 2004 19:25 UTC (Sun) by dvdeug (subscriber, #10998) [Link]

I can see that more general definition. However, "poking at" doesn't imply a process to me; keeping your eyes open during the normal course of using the program is not reverse engineering. Even playing around with a program to see how it works and how you could improve your program isn't really reverse engineering to me.

I seriously doubt that most of these people who were refused licenses were going to reverse engineer the program, instead just looking at it and possibly poking at it.

Unable to buy a copy of BK

Posted Sep 26, 2004 7:05 UTC (Sun) by Ross (subscriber, #4065) [Link]

But few have a "we won't sell to anyone who employs or has employed
someone who has or has potentially discussed reverse engineering (or
our product(s) or the protocols they use" (not an actual quote from a BM
license or sales person, it's just a characterization of what LM said).

No-reverse-engineering clauses really rub me the wrong way even without
that extra twist.

I'm changing my mind again: no BitKeeper for me, ever.

Unable to buy a copy of BK

Posted Sep 26, 2004 11:34 UTC (Sun) by nix (subscriber, #2304) [Link]

Quite. I was going to recommend it at work (~1500 seats): I'm certainly not now. It's arch with lots of wrapper scripts for us.

It'll be more annoying for me, but less disruptive than having to have everyone who's ever worked for us watch what they say on every net-accessible source in case we say something that annoys BitMover.

Note: I've said things that have annoyed representatives of Sun, and Oracle, and even Microsoft; I've made it plain that I'm in favour of competitors to Oracle springing up, as well, and the Sun people knew I used Linux. None of them have decided to stop selling me (or the company I work for) things because of it.

(And, yes, my non-lawyer's memory agrees that the no-reverse-engineering clause is not enforceable across the EU, although doubtless the copyright lunatics are bringing in something to reverse that regrettably just state of affairs.)

Unable to buy a copy of BK

Posted Sep 27, 2004 3:42 UTC (Mon) by bronson (subscriber, #4806) [Link]

1500 corporate Arch users? There just ain't no way, no matter how many wrapper scripts you write. Arch barely scales to 1/10 that many users right now. Watch the Arch mailing list and IRC and notice the number of merge issues that need to be resolved by hand. It seems a small issue, but with 1500 users, this sort of manual labor will bury you.

You'd best give them a year or two to get the speed up and simplify the merging. Otherwise, you're going to have 1500 people using Arch as nothing more than a poor-man's RCS.

If you do manage to use Arch's features with even 200 corporate developers, please write an article about it! That would certainly be a potent milestone.

Unable to buy a copy of BK

Posted Sep 27, 2004 5:12 UTC (Mon) by atai (subscriber, #10977) [Link]

can BK or even Clearcase handle 1500 users working on the same code base? I assume that's what you mean in your message.

Unable to buy a copy of BK

Posted Oct 1, 2004 1:04 UTC (Fri) by lm (guest, #6402) [Link]

Yes, BK can handle 1500 users working on the same source base, or pretty close to it. There are about that many people checking stuff into the linux kernel trees. You can check it out on www.bkbits.net but truth in advertising, not all of those users listed are actually using BK, some of them are proxied into the system via patches imported by someone else.

That said, I'm pretty sure there are >1000 actual BK users working on the Linux kernel tree directly. BK uses a lease based model for its licensing which means that those users connect to openlogging.org once a month to get a lease (this all happens in the background, nobody realizes it is there, which is - in my opinion - how a license server should work, I'm not a flexlm fan). We can correlate those with the stuff on bkbits.net and openlogging.org and a 1000 looks to be a lower bound. I think it's quite a bit more than but I haven't worked through all the data to be sure (we have many many GB of logs). So 1500 all on one source base? I don't think we have any commercial users with that many users all working on the same source base, that's pretty atypical anyway, source is usually broken up into chunks with different groups working on different chunks.

Unable to buy a copy of BK

Posted Sep 27, 2004 10:47 UTC (Mon) by nix (subscriber, #2304) [Link]

They don't commit that often :) many of them are using it as a distribution mechanism.

But I guess it'll be SVN for now, anyway: it seems that `like CVS' is more important than I thought it was.

Unable to buy a copy of BK

Posted Sep 28, 2004 3:50 UTC (Tue) by Talli (guest, #25044) [Link]

Nix,

I am starting a company with Tom Lord to develop a company around GNU arch. If you would like to chat about how GNU arch can be used at your dev shop, we would be delighted to discuss that with you.

You can reach me at talli-at-seyza.com

Thanks and looking forward to chatting.

Talli

Unable to buy a copy of BK

Posted Sep 26, 2004 13:15 UTC (Sun) by mbp (guest, #2737) [Link]

There is a difference between "reverse engineering" and "poke at". (OK, I am not an IP lawyer...) No proprietary licence allows you to disassemble the binaries; that's fine with me. But no reasonable licence forbids the customer from looking at the files containing their own data, or thinking about the design, or "poking" the program.

I completely defend your free-market right to refuse to sell to whoever you want, and certainly your right to choose a proprietary licence. But I have to say all these shifts in the licence worry me. Just the other day I recommended bk to a business user. Had I known about this I would have thought again. I couldn't recommend storing code in a system whose licence can be denied because you dislike something that one employee said.

My employer competes with IBM products, but are they going to suddenly revoke our Clearcase licence, or refuse to sell a new one? I certainly hope not. I suppose the possibility is a strong argument for using a free system as soon as there is one that's good enough.

Unable to buy a copy of BK

Posted Sep 26, 2004 13:47 UTC (Sun) by hppnq (subscriber, #14462) [Link]

The thing with licenses is, that they have to be extremely accurate in what is permitted and what not. So if your clause is intended for a particular reason, but it does not cover it, it is wrong -- you cannot blame people for not trusting you will adhere to the original "spirit" of the license.

So you'll have to explain it a bit better, if you care. Right now, I would never recommend using Bitkeeper *exactly* because of this clause. Maybe I don't fully get it, but as a licensee I would be worried about letting you know that I'm thinking about switching to another version control system.

Unable to buy a copy of BK

Posted Nov 25, 2004 18:45 UTC (Thu) by djao (subscriber, #4263) [Link]

Larry,

I freely admit that I am not at all involved with or affected by anything related to bitkeeper, but even as a detached party, your license shifts do worry me.

Two years ago you said, or at least strongly implied, that BKCL had no non-compete clause. In fact you even specifically mentioned that BKCL did not, in order to defend yourself against the (at the time) controversial notion of BKL having a non-compete clause.

Now we find that not only the free users but also even the commercial users of BK are not permitted to develop competing software. Can you imagine (say) Microsoft refusing to license Windows to any person developing competing OSs, or to any company employing any person developing competing OSs?

That is EVIL.

You can't claim on the one hand that the no-reverse-engineering clause already prohibits competing against you, and on the other hand refuse to sell BK to competitors. This very position is a contradiction in terms -- if the clause is sufficient protection, then why do you feel the need to further protect yourself with discriminatory sales tactics?

I can understand your logic in not making BK free software, but I cannot understand your logic in refusing to sell BK to competitors. The no-reverse-engineering clause already provides you enough protection. Any incremental gain to BK from sales discrimination is not worth committing such an evil act, if only for public relations reasons.