Linux founder Linus Torvalds may soon be using a proprietary, closed source code management tool for Linux kernel development, to the dismay of many in the open source community. Torvalds today diplomatically postponed his final decision, explaining that he's taking a week offline, and he says actively exploring alternatives to the Bitkeeper software that's served as his primary code management tool in recent years. Torvalds' announcement follows a decision by Bitkeeper's Larry McVoy yesterday that he would focus on the proprietary version of the software, orphaning the free client.
It's a crunch for Torvalds. However, if he continues to use Bitkeeper, according to senior figures who spoke to the The Register on condition of anonymity, it could severely undermine his authority in the Linux community. Torvalds' decision to use Bitkeeper effectively obliges hundreds of kernel developers to follow suit.
Torvalds adopted Larry McVoy's BitKeeper tool for managing submissions to the Linux kernel three years ago, and this week McVoy said he was phasing out the free client, hoping to complete the transition by July.
In a press release issued yesterday, McVoy explained that "our commercial users exceed those using BitKeeper on Open Source projects. Focusing our energy on our cutting edge commercial products will ensure we maintain our competitive advantage..." McVoy hardly mollified his critics recently by branding them "politically correct".
The Linux founder hasn't made a decision, but software libre supporters were drawing some solace from the fact that Torvalds was today referring to Bitkeeper in the past tense -
"I've decided to not use BK [Bitkeeper] mainly because I need to figure out the alternative," he wrote on the kernel mailing list this morning ... and rather than continuing 'things as normal', I decided to bite the bullet and just see what life without BK looks like. So far it's a gray and bleak world."
"I'm just asking that people who continue to maintain BK trees at least try to also make sure that they can send me the result as (individual) patches, since I'll eventually have to merge some other way."
Torvalds' decision has a symbolic importance. In 1985, Richard Stallman began the GNU project to create a free operating system by building a free toolchain - the GNU compiler and GDB debugger - for developers to use. Open source critics will point out that if Torvalds can't eat his own dog food, why should anyone else?
"There are consequences," former Debian lead Bruce Perens pointed out on a Slashdot discussion this morning. "1000 people are going to have to learn a new facility, that facility is going to have to be deployed and files are going to have to be moved into it in a laborious version by version process to convert them, etc. There is also all of the surplus heat produced by the multi-year argument that Bitkeeper brought and some loss of productivity because of that, including some untold number of people who would otherwise have worked on the kernel but bugged out because of the Bitkeeper decision."
McVoy claims his decision to walk away from the open source community was prompted by a consultant contracted to ODSL - which is also Torvalds' employer - who reverse engineered the Bitkeeper protocols. McVoy said this project undermined the business model beneath his company Bitmover, which develops Bitkeeper. However others dispute McVoy's version of events.
"Larry changed the deal, repeatedly," wrote Perens on Slashdot.
McVoy had made it impossible to for an ethical open source developer to use Bitkeeper.
"It started out that we just had to use [McVoy's] 'notification server', and then other odd terms came up at intervals like termination of the license for those who attempt to make other software compatible with Bitkeeper through reverse-engineering. OSDL refused to terminate an employee or consultant who was also reverse-engineering Bitmover as a hobby Open Source project outside of OSDL. Had they terminated that person, the hue and cry would have been greater.
"There was never a chance that this relationship could work, because of the lack of an Open Source license and the mercurialism Larry regularly displayed."
Tensions have been simmering for years. McVoy first showed his designs for a distributed source code management system to Torvalds in 1998, and BitMover began hosting the kernel tree in 2001. But McVoy has never made a secret of his intention to sustain a business of paid professional developers, in which non-free licenses play a crucial role.
So whether you take the view that Bitkeeper isn't compatible with the principles of the Linux project, or vice versa, is moot. It's simply a wonder it took so long for things to come to a head. ®