OneAndOneIs2

If you've not come across it already, there's quite a good talk by Linux Torvalds on YouTube that he gave at Google on git, the software that manages the Linux kernel.

One of the big things it highlights is that the Free Software zealots were wrong about the whole bitkeeper thing.

For those not familiar with it, basically it boils down to this:

The kernel was managed by software called bitkeeper, which was (shock horror) proprietary software that the kernel devs were allowed to use for free. Right from the start, RMS & co. claimed this was a Bad Thing (TM).

Some years later, for various political reasons, BitMover (the owners) withdrew the free-as-in-beer status. Cue a smug round of "We told you so"'s from the FS brigade, who hammered home repeatedly the point that if you use proprietary software you run the risk that support for it can be taken away at any time.

This, they claimed, was yet another reason why you should never use any kind of proprietary, non-free software. It's still not uncommon to see people citing BitKeeper as an "RMS was right all along" argument.

Thing is, they're completely wrong.

Why?

Unless you happened to see the story reported on the various news sites, you would never have known anything had happened: Linux kernels continued to be released and updated without any significant pauses or problems. I guarantee you never saw any headlines about distros shipping the same kernel two releases in a row because the sudden loss of BitKeeper caused huge kernel issues.

The replacement for BitKeeper, git, was written by Linus from scratch in a matter of days. From Wikipedia: The development of Git began on April 3, 2005. The project was announced on April 6, 2005, and became self-hosting as of April 7, 2005. The first merge of multiple branches was done on April 18, 2005.

Bitkeeper was replaced in no time at all. Its loss was a bit of a nuisance, but nothing more.

Essentially, Linus was faced with a choice all those years ago: He could use BitKeeper, at the time the best available tool for the job, and have the risk of having to write git hanging over him the whole time; or he could write git right from the start.

FS advocates would have you believe that the former choice was a bad one. They ignore the fact that the git Linus could have written back then would have been written with far less experience of managing the kernel development and with no experience of a tool well-suited to doing the job.

By using BitKeeper whilst it was available and then writing git, the result was a far better git right from the start. The cost? An effectively unnoticeable hiccup in the kernel development as everybody migrated.

Either choice would have required git to be written. By delaying its creation by several years, he ensured that when it was written, it was written well. It does clever things and performs well and is uniquely suited to managing the kernel development. Using BitKeeper was a beneficial move, not the disaster that FS advocates would have you believe.

The important distinction, I think, is in the replaceability: Specific versus generic. Bitkeeper was just software that did a job: Any skilled coder could come up with an alternative system. It's a generic "manage this" task.

Proprietary formats, however, are different, because you can't just sit down and write a WMV player or .doc reader. These are specific tasks that can only be solved one way.

It's the difference between "You shouldn't use MS Word to write letters" and "You shouldn't use MS Word to save documents" - the first argument is not valid, because any word processor can be used to write letters so you can switch to OpenOffice or whatever at any time. The second is valid, because even OpenOffice isn't 100% reliable at .docs yet, so you're locked in.

Refusing to acknowledge the difference in the two types of proprietary software undermines the very argument you try to make. "BitKeeper was proprietary, and it was taken away" is not a valid argument for avoiding proprietary software, because it can instantly be countered by "But it was replaced in no time at all by a superior system that was only possible because BitKeeper was used in the first place."