treemanifest: make diff() faster
Containment checking is slower in treemanifest than it is in
manifestdict, making the current diff algorithm O(n log n). By
traversing both treemanifests in parallel, we can make it O(n). More
importantly, once we start lazily loading submanifests, we will be
able to easily skip entire submanifest if they have the same nodeid.
$ . "$TESTDIR/bzr-definitions"
$ cat > treeset.py <<EOF
> import sys
> from bzrlib import workingtree
> wt = workingtree.WorkingTree.open('.')
>
> message, rootid = sys.argv[1:]
> wt.set_root_id('tree_root-%s' % rootid)
> wt.commit(message)
> EOF
change the id of the tree root
$ mkdir test-change-treeroot-id
$ cd test-change-treeroot-id
$ bzr init -q source
$ cd source
$ echo content > file
$ bzr add -q file
$ bzr commit -q -m 'Initial add'
$ python ../../treeset.py 'Changed root' new
$ cd ..
$ hg convert source source-hg
initializing destination source-hg repository
scanning source...
sorting...
converting...
1 Initial add
0 Changed root
$ manifest source-hg tip
% manifest of tip
644 file
$ cd ..