On Wed, Oct 27, 2010 at 09:43:50AM +0200, Dirkjan Ochtman wrote:
> On Wed, Oct 27, 2010 at 09:34, Robin H. Johnson <robbat2@g.o> wrote:
> > For comparison, what would the overhead costs of hg be above?
> Mercurial would be about O(f + d + 10r).
Thanks.

> But wouldn't the real git inode space be O(35r) directly after
> repacking, but grow O(rev) until the next repack?
Correct, I explicitly noted "bare minimum number of inodes" for that
reason.

Git's more common case is:
O(35r + 2c)
where 'c' is the number of unpacked objects. The default repack kicks in
around 6700 unpacked objects (where an object is a commit, tree, or file
blob, and every commit has at least 3 objects). So it would be an extra
27MiB worth of inodes until the repack kicks in.

I'm also leaving out the fact that there might be a couple of pack
files. In these cases, the base notation I've noted gives a good idea of
the minimum number of inodes consumed as overhead.

-- 
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail     : robbat2@g.o
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85