2 months ago
I've been wondering whether your servers could stand the whole history of the Linux kernel as mentioned in this post: http://lkml.org/lkml/2007/12/12/450
I didn't dare to add it, but if you like...
I also found something that looks like an import bug. The oldest commit to the Linux kernel can't possibly be from 01-01-1970. Maybe source import should just leave out commits without a sensible timestamp.
2 months ago
note that: " All commits are done with a committer and author git id of "Linus Torvalds", except for the stable kernels which were set to "Alan Cox" or "David Weinehall" etc where applicable."
so it's useless for ohloh...
2 months ago
We've spent some time looking around for better linux histories, and this is the best one I've heard of. However, it sounds like it does have a couple of strikes against it as far as Ohloh is concerned:
It sounds like it expands out to consume a lot of disk space, which would cause some chaos on our servers. During the line counting process we tend to keep the whole tree expanded onto disk, which is ridiculous, but we've never had such a large repository before. It's fixable on our end, but we're not there yet.
Marcin is right, if all the commits get handed to Linus Torvalds, that's not quite honest -- but perhaps it's better than not having the commits at all.
Regarding the 1970 committer dates, yes, I agree that they're not correct, and I did originally think we had an importer bug when I first saw them. But it turns out that Git lets you jam in whatever commit date you feel is appropriate, and someone decided to put some 1970 dates in the repository. So (unless this has changed since I last investigated) the 1970 dates are actually in the repository this way. Sigh.