I have recently implemented all Obnam features I think I want before I start using it for real, except encryption. The next step is to re-implement the backup store implementation. The current implementation is the simplest, most stupidest one I could get away with. I did not care at all about performance, so it is rather, er, slow.
I said stupid, right? Yes I did.
I may end up implementing the backup store in several ways, to be able to compare them in semi-real-life benchmarks.
I was going to add support for ACLs and extended attributes, but I decided not to: I do not use them myself, and they're just non-obvious enough that I am going to need to find a collaborator to verify I do the right thing. (Ideally, someone who'll also write the code... I promise to show how.)
In other news, B-trees are surprisingly interesting.