[MERGE] documentation from the london sprint

[Robert Collins]

On Sun, 2007-06-03 at 23:55 -0400, Aaron Bentley wrote:
> 
> > The new enough bit is that the hash for the subtree1 (assuming the
> > hashes are directory aligned) can be stale (still be the one from
> the
> > time of checkout, for instance), as long as now
> > moves-into-or-out-of/adds/deletes have taken place within it. 
> 
> So how do we know that the fileid for new.c is unique?

Bah. it must have been something else. let me see now...

Oh right, revert.

So for revert, you just need to know that nothing in a subtree not
matching the paths given by the user could have been moved into/outof
the subtree the user selected.

So given two trees where one is historical and one the working tree, we
can avoid scanning all the elements of the working tree for a path, if
we have a hash of that path at some point in time no *older* than the
last add/delete/move-into-outof that subtree. By using that hash rather
than the realtime hash we avoid statting and or reading all the contents
of the subtree the user did not select.

-Rob
-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070604/494a1398/attachment.pgp