The BitKeeper Break
In the previous series we saw the Linux kernel grow from a short post into a project with thousands of contributors. But the bigger it got, the harder one simple question became: how do you manage all these changes?
The years of patches and email
According to the official Git book, from 1991 to 2002 changes to the kernel were passed around as patches (files showing the difference between old and new code) and archived files. Programmers emailed patches, and Torvalds read and applied them one by one.
That works for a small project, not for one this size. LWN later wrote that in those days people said Torvalds "did not scale": patches were dropped, developers complained and he was getting tired.
A closed tool for an open project
A programmer named Larry McVoy, head of the company BitMover, offered a solution: his company's version control software, BitKeeper. According to LWN, he had worried since 1999 that Torvalds was heading for burnout and believed the right tool would fix it. In 2002 Torvalds decided to give it a try.
The result was striking. BitKeeper established a model with no central repository: each developer could keep a full, independent copy of the code's history and "pull" changes from one copy to another. LWN wrote that the rate of merged patches skyrocketed and developers were happy.
But there was a big problem: BitKeeper was proprietary. Its free version came with strict conditions, including that users could not work on any other version control system or reverse engineer it. Many free software advocates objected from the start: why should the world's largest open project rely on a closed tool?
The break
In April 2005 the fragile balance collapsed. Andrew Tridgell, an Australian programmer known for creating Samba, who, like Torvalds, was a fellow at OSDL, tried to build a free tool that could interoperate with BitKeeper repositories by studying how it behaved. BitMover saw this as a breach of its terms. As LWN reported, the company ended up withdrawing the free version of BitKeeper. Days later, as ZDNet reported, Tridgell rejected claims that he had violated BitKeeper's code, demonstrating on stage that the information he used was available from BitKeeper's own online help.
Torvalds announced he had stopped using BitKeeper to figure out the alternatives and wrote wryly: "So far it's a gray and bleak world." LWN noted it was not clear that any existing free tool could handle a project as large as the kernel, and going back to patches and email was not an option either.
The turning point
Overnight, the world's largest open project lost its main working tool. It was exactly the danger critics had warned about for years. LWN wrote at the time that proprietary software can be good stuff, but it always carries the threat that "you never really know if it will be there for you tomorrow or not." At the same time, LWN fairly pointed out that BitKeeper had served the project for three years and the kernel would not have reached its level of capability without it.
Lesson for businesses
Critical dependence on a tool you don't control is a hidden risk. It stays invisible while things go well; but the day a supplier changes its terms, your whole business is caught. Always ask: "What do we do if this is taken away tomorrow?"
Second, take contract terms as seriously as product quality. Technically, BitKeeper was excellent; the problem was the restrictions in its licence.
Third, don't forget gratitude. Even when a relationship ends, recognising the value it created leads to wiser decisions.
Next: Torvalds decided not to wait. He disappeared for a few days and came back with a new tool he jokingly described as "the stupid content tracker".
منابع / Sources
"A Short History of Git", Pro Git book, git-scm.com, n.d. — https://git-scm.com/book/en/v2/Getting-Started-A-Short-History-of-Git
"The kernel and BitKeeper part ways", LWN.net, Jonathan Corbet, 2005-04-06 — https://lwn.net/Articles/130746/
"Tridgell speaks out in BitKeeper war", ZDNet, Angus Kidman, 2005-04-20 — https://www.zdnet.com/article/tridgell-speaks-out-in-bitkeeper-war/
