[MERGE] InterCHKRevisionTree (brisbane)

[Vincent Ladeuil]

>>>>> "jam" == John Arbash Meinel <john at arbash-meinel.com> writes:

<snip/>

    jam> +    # CHKInventory does not have an InterTree optimiser class (yet).
    jam> +    chk_tree_format = WorkingTreeFormat4()
    jam> +    chk_tree_format._get_matchingbzrdir = \
    jam> +        lambda:bzrlib.bzrdir.format_registry.make_bzrdir('development5')
    jam> +    test_intertree_permutations.append(
    jam> +        (InterTree.__name__ + "(CHKInventory)",
    jam> +         InterTree,
    jam> +         chk_tree_format,
    jam> +         chk_tree_format,
    jam> +         mutable_trees_to_revision_trees))


    jam> ^- I think this is wrong, though. As we *do* have an inter-tree
    jam> optimizer format. And it should be properly set up with the appropriate
    jam> objects for testing. If it isn't, then we should quickly do so.

    jam> I remember the discussion on this, and I thought Vincent had already
    jam> done it. But perhaps it slipped through.

What happened is that I tried to implement the modification along
the lines discussed with you and Robert... but ended up having to
import workintree from revisiontree (ugh) and then either
duplicate mutable_trees_to_revision_trees (Ugh) or import it from
tests.intertree_implementations (UGH) at which point I thought
this was a bit too weird... and indeed it then slipped through.

Then there was the 'we shouldn't use an Intertree object to
detect inventories' remark, but I don't remember how that was
supposed to be addressed (or if it can be addressed without
refactoring/rewriting the tests so that working trees aren't
required anymore).

Finally, we need a 'development5' format which hasn't land in
bzr.dev, so I'm a bit at a loss to how to handle these tweak :)

The best I can think of is to include the
test_intertree_permutations.append in the loop... which I did for
brisbane-core, but until bzr.dev propose a 'development5' format
we just can't activate these tests.

Thoughts ?

         Vincent