On the tenth anniversary of Git, Linux.com has a nice interview with Linus. In it they explore the reason for Git’s creation, its supposed opaqueness to ordinary users, its success, and its future.
Although many already know the story, Linus recounts how he was forced to give up using BitKeeper when one of the kernel hackers, Tridge, started reverse engineering its protocol in violation of the agreement that allowed its free use for kernel development. Add to that the fact that many developers objected to BK because it wasn’t open software and Linus pretty much had to change. The problem was that there was nothing else that came close to meeting his requirements. As a result, he spent a few days hacking together the first version of Git. What I didn’t realize was that it was self-hosting Git development after the first day. The first kernel commit happened at about day 10. That’s pretty amazing for any non-trivial development effort.
Linus admits that Git was difficult to use for the first six months because all the effort went into getting it running and adding features. Now, he says, that’s no longer true. Its one weakness, he says, is that there are several ways of accomplishing some tasks.
As for Git’s success, Linus thinks it’s mostly because all the other systems are so bad. First, you have to have a distributed model for a successful VCS. Even the distributed systems that came before were lacking. All that said, I think it’s pretty clear that Git has a simple, easily understood model that contributes greatly to its success.
Linux.com asked Linus if he foresaw some other system replacing Git in the next ten years. Linux replied that one thing for sure is that he wouldn’t be the one to write it. He also noted that if something did replace Git, it would have to be Git-like.
It’s an interesting interview and well worth a few minutes of your time.