I see that github have recently announced their pricing plans. Looks like they’ll not be having many kernel developers using their “open source” plan – my current working Linux .git is 261Mb but you start having to pay at 100Mb. git may be very space efficient but the kernel is an enormous project. Mind you, the whole web service based approach of github is fairly alien to kernel development – a large part of what it’s doing is replacing the combination of mailing lists together with git send-email and git am (or their equivalents) that the kernel relies on for code exchange and review. Interestingly it appears not to have done anything that provides the discussion and review side of things – that seems like a surprising omission for a web 2.0ish thing.
I know I’m a little late to this post, but I thought I’d set the record straight for anyone that happens to read this.
1) When you fork a repo, we only save your differences from the source repo.
Meaning, when you fork github’s mirror of the kernel (http://github.com/github/linux-2.6), the space you’ll use is negligible (< 1MB).
2) We also provide commenting on source files and diffs which works as a really interesting code review tool.
Read more about it here: http://github.com/blog/42-commit-comments
Thanks. For the space usage you probably want to make it clearer how the space is accounted – it’s not at all clear that you only count differences in your billing, especially once you get onto stuff like merging in other repositories. Will this only work if you explicitly fork from within github or will the server figure out relationships otherwise?
I saw the commenting stuff – that wasn’t present at the time I wrote the post.
Having worked with this for a while it actually appears that the duplicate detection stuff doesn’t quite work – it looks like what’s happening is that content pushed up that is also present in the origin repository (eg, merges down of upstream code) are getting counted as space usage.