No More Free BitKeeper (KernelTrap)

[Posted April 6, 2005 by corbet]

KernelTrap is reporting that the free version of BitKeeper, used heavily in kernel development, is being withdrawn. "In a post to the Linux Kernel mailing list in February of 2005, Larry [McVoy] discussed a 16 bit limitation of the existing free product. With nearly 64,000 changesets in the mainline kernel tree, future development will quickly exceed this limitation. For that reason, it is likely that BitMover will provide one final release of its free BitKeeper product, allowing kernel developers a graceful transition. By the end of July, the goal is to have completed the migration, therby [sic] terminating the free product and focusing fully on the commercial product."

BitMover's Larry McVoy has confirmed the story (he pointed us to it, actually) and says that an announcement is forthcoming.

Bazaar

Posted Apr 6, 2005 14:14 UTC (Wed) by simon_kitching (guest, #4874) [Link]

Interestingly, I was reading just yesterday about Canonical´s efforts to create a new distributed version control system based more-or-less on Arch. Canonical is the company funded by Mark Shuttleworth to create Ubuntu Linux.

Bazaar is a cleaned-up gnu arch.

Bazaar-ng appears to be a more experimental project, inspired by a number of different version-control systems. It´s self-hosting already, though!

See:
http://wiki.gnuarch.org/Bazaar
http://bazaar.canonical.com/
http://www.bazaar-ng.org/

No More Free BitKeeper (KernelTrap)

Posted Apr 6, 2005 14:33 UTC (Wed) by gowen (guest, #23914) [Link] (2 responses)

Now, if only someone had warned that using a proprietary SCM system could you you at the whim of the vendor should they ever decide to unilaterally change the licensing terms.

Gosh darn it.

No More Free BitKeeper (KernelTrap)

Posted Apr 6, 2005 14:51 UTC (Wed) by allesfresser (guest, #216) [Link] (1 responses)

Hey, you don't suppose that limitation could have been designed in at the beginning... for this very purpose... do you? Nah. 16 bits should be enough for anyone, forever.

No More Free BitKeeper (KernelTrap)

Posted Apr 7, 2005 21:21 UTC (Thu) by cloose (guest, #5066) [Link]

LOL! Yeah a "world class" SCM with a 16-bit changeset limitation. *shakes
head*

No More Free BitKeeper (KernelTrap)

Posted Apr 6, 2005 16:07 UTC (Wed) by LinuxLobbyist (guest, #6541) [Link] (7 responses)

Ah, yes, the woes of using non-free software. I wonder how few people didn't predict something like this would happen.

What get's my goat is, once again, the bald faced arrogance of one Larry McVoy. To paraphrase, "out of 10, we're a four, but everyone else is a three, they just don't know it."

Reverse engineering, even post-DMCA, is protected and has been upheld by courts. And it is, dare I say, not an immoral act. Larry's years of gestapo-like enforcement of his "you can't develop a competing product," has made him many enemies that he didn't already have. And yet, he plays the victim here.

We only hear Larry's side of the story here regarding the OSDL employing the developer working on the reverse engineering, but even Larry's testimony about it indicts only him, not OSDL. For Larry to suggest that the OSDL should do something, given that the current reverse engineering is not being funded by OSDL is to promote the very worst of proprietary software company employee agreements: the IP agreements that purport to lay claim to everything the employee does on his own time, on his own equipment. That developer is solely responsible for what he is doing. Not the OSDL. To even hint otherwise, is despicable. You got a problem with him, go after him and leave OSDL out of it. It's none of their business, and it's none of your business where he works if he's doing the work on his own time.

But this is all moot, now. Even if it has not been the intention, this happening with that developer working on reverse engineering BK has had a happy ending. The end of using a non-free tool for the GPLed kernel. Larry is dead wrong about how this reflects on the FOSS community. There are many commercial companies that are well aware that their days are numbered if they don't change their revenue streams and they are doing just that. And they aren't disparaging the FOSS community as they do it.

Larry and BitMover are likely to be left behind in this disruptive force if they don't change. Oh, they'll make plenty of money in the meantime, but not indefinitely. This happening says more about Larry than it does about FOSS developers. He's playing the victim, through and through. The truth is that in an odd sort of way, we are all winners here. Larry no longer has to deal with those he insultingly refers to as whiners and can go on to focus on making money. The Linux kernel developers will now be free from the shackles of a non-free SCM tool.

-Paul Iadonisi
 Senior System Administrator
 Red Hat Certified Engineer / Local Linux Lobbyist
 Ever see a penguin fly?  --  Try Linux.
 GPL all the way: Sell services, don't lease secrets

No More Free BitKeeper (KernelTrap)

Posted Apr 8, 2005 0:39 UTC (Fri) by giraffedata (guest, #1954) [Link] (6 responses)

Reverse engineering, even post-DMCA, is protected and has been upheld by courts.

I think you mean making a copy for purposes of reverse engineering is not a violation of copyright. (It falls under the fair use exemption).

It is still possible for a person to enforceably trade a promise not to reverse engineer for valuable consideration such as a copy of a copyrighted program. I believe that's what happened here. There was a contract.

On the other hand, BitMover isn't pursuing the legalities, so this is all irrelevant. The morality question is more relevant, both the basic issue of whether it is moral to reverse engineer, the related question of whether it's moral to try to induce someone not to, and the higher level question of whether those moralities outweigh the morality of reneging on a deal.

No More Free BitKeeper (KernelTrap)

Posted Apr 13, 2005 13:59 UTC (Wed) by jschrod (subscriber, #1646) [Link] (5 responses)

According to Tridge, he did not use BitKeeper at all; thus he was not bound by BK's license.

Linus mentioned that Tridge tried to analyse the BK database format. So this might be the way that he approached the reverse engineering. As a co-founder of the Samba team, he has obviously quite some experience with reverse engineering...

According to a Slastdot post by Jeremy Allison from the Samba Team, Tridge is not allowed to speak about his view on the affair. So there might hide a legal followup in this whole mess.

Joachim

No More Free BitKeeper (KernelTrap)

Posted Apr 13, 2005 16:09 UTC (Wed) by giraffedata (guest, #1954) [Link] (4 responses)

According to Tridge, he did not use BitKeeper at all; thus he was not bound by BK's license.

Just to keep the legalities straight: a license doesn't bind the licensee in any case. It binds the licensor only. So the question is whether Tridge was bound by the BitKeeper EULA -- a contract. Or more precisely, whether the clause in there prohibiting reverse engineering applies to what he did.

But I wonder if Tridge is even a party to that contract. BitMover complained to OSDL, which suggests OSDL is the one that BitMover thought breached the contract, and OSDL responded only that Tridge was doing the reverse engineering as a personal project.

There are reports that BitMover extorted OSDL to fire Tridge or lose BitKeeper. That sounds incredible to me, but could easily be a distortion of this: BitMover claimed that as long as Tridge was in OSDL's employ, his actions were vicariously OSDL's and OSDL is in violation of the contract.

No More Free BitKeeper (KernelTrap)

Posted Apr 13, 2005 16:37 UTC (Wed) by jschrod (subscriber, #1646) [Link] (3 responses)

There are reports that BitMover extorted OSDL to fire Tridge or lose BitKeeper. That sounds incredible to me, but could easily be a distortion of this: BitMover claimed that as long as Tridge was in OSDL's employ, his actions were vicariously OSDL's and OSDL is in violation of the contract.

IMHO, it doesn't matter which of your both interpretations are true -- both are ridiculous. An employer is not responsible for the actions of their employees that they do on their own time. Not even the US legal system is as mad to claim that.

For me, as a CEO, this simply makes it clear that we will never ever do business with LmV. (I considered licensing BK, for some time.) It's obviously too dangerous to be in a contractual relationship with him. It seems that in his view, I might be responsible what my staff does in their free time -- but that is impossible by our law.

Joachim

No More Free BitKeeper (KernelTrap)

Posted Apr 14, 2005 1:57 UTC (Thu) by giraffedata (guest, #1954) [Link] (2 responses)

IMHO, it doesn't matter which of your both interpretations are true -- both are ridiculous. An employer is not responsible for the actions of their employees that they do on their own time.

Except that we don't know this was on Tridge's own time. We just know (and only through the grapevine) that OSDL thinks it was. There could easily be a legitimate disputed fact here.

The law does make certain allowances to ensure that a corporation can't escape liability for its actions by always claiming they were independent actions of employees. So it might not be cut and dried.

No More Free BitKeeper (KernelTrap)

Posted Apr 14, 2005 2:06 UTC (Thu) by jschrod (subscriber, #1646) [Link] (1 responses)

Your hypothetical cases doesn't fit here.
Tridge wrote it was on his own time, and OSDL wrote it was on his own time.

Do you have any reason not to believe both of them? Any message by Linus or other OSDL staff that says otherwise? Or do you just want to cast doubt on a respected open source developer?

Joachim

No More Free BitKeeper (KernelTrap)

Posted Apr 14, 2005 2:53 UTC (Thu) by giraffedata (guest, #1954) [Link]

First of all, I haven't seen anything written by either. The statement you just made that these writings have happened is the most direct evidence that has appeared so far in this article or its comments of whether the work was done on Tridge's own time.

Second, I'm assuming they said it and believe it.

It doesn't matter. There are disputed facts where no one is lying. History is full of hot disputes where people couldn't agree on whether an employer and employee had fully separated their activities. Furthermore, we don't even know if it's a relevant fact. OSDL might have promised not to employ any person who is known to be involved in reverse engineering period. This thread contains almost no information on the substance of the dispute. Hence my speculation.

Here's more info about OSS/FS SCM systems

Posted Apr 6, 2005 16:24 UTC (Wed) by dwheeler (guest, #1216) [Link] (1 responses)

See my paper Comments on OSS/FS Software Configuration Management (SCM) Systems for more information on OSS/FS SCM systems. I review several, list ones I know of, and reference lots of other places for more information.

Here's a list of many OSS/FS SCM systems: CVS, Subversion (SVN), GNU arch, Monotone, Aegis, CVSNT, Darcs, FastCST, OpenCM, Vesta, Codeville, Bazaar, and Bazaar-NG.

...and ArX

Posted Apr 7, 2005 23:10 UTC (Thu) by kevinbsmith (guest, #4778) [Link]

I know ArX is mentioned on your page, but somehow got left out of your list here.

http://www.nongnu.org/arx/

No More Free BitKeeper (KernelTrap)

Posted Apr 8, 2005 3:24 UTC (Fri) by brianomahoney (guest, #6206) [Link]

First of all, once again, I am profoundly disappointed by the reaction, in the community to this issue, for two reasons

(a) this was entirely predictable [ see Stallman rants -- he isn't always wrong, and on strategic
issues, like this, is very often right], and

(b) BitKeeper is now a very good product, much of the corner cases honed on the kernel.

Now, what lessons do we learn from this?

1. Follow the old saw ' if you dine with the Devil, use a long spoon'

2. McVoy is deeply, personally and fixedly focused on his vision, which has
proved excellent and generated a fine product, far superior to the main
commercial products, e.g. ClearCase and PCVS which cost a lot of money
and are both buggy and bad,

3. However the problem, for McVoy, is that he has created an excellent
proof of concept, the essentials of which will now easily be
re-engineered, but that could have been covered by patent, _AND_ he has
now demonstrated to the Open Source community that he is not to be
trusted. The famous McVoy temper.

I really do hope we learn from this, because as we really begin to win,
these issues become more and more central, imagine if 'gcc' was encumbered.

So, while I wish LmV every success, I personally will now NEVER promote his
product in the commercial marketplace, why?
... because with him it is always temper, temper ... not just rational business.