record_iter_changes and subtrees

[Robert Collins]

I don't think a 2-phase commit is a good approach, because if a commit
message is done for libfoo that is a nested tree, we should consider it
done. Readers can read it immediately, unless we ask for foreign systems
to have capabilities that most [svn, hg at least] don't have, or only
allow bzr trees to be nested.

As for inline-vs-external iteration, its not significantly more complex
to recurse when you hit the tree rather than go searching for it first,
beyond needing to use the same repository object when it is the same
repository the commit is being made to.

re: iter_changes and what it reports - it already can report unchanged
files, and given the goal of having 'one fast tree iterator' it seems to
be managing very nicely.

-Rob
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20090326/dabfcb11/attachment.pgp