[MERGE] Accept NULL_REVISION everywhere

[Martin Pool]

Hm, bb didn't catch this yet.

Anyhow: clarifying that None is not the null revision is a worthwhile
improvement, i'm happy to have that come in.

Could you please also add something about this to the HACKING
document, or somewhere like that?  Should this also be mentioned in
the api section of NEWS?

=== modified file bzrlib/repository.py // last-changed:abentley at panoramicfeedba
... ck.com-20070710211854-93ofaa70day2p8d3
--- bzrlib/repository.py
+++ bzrlib/repository.py
@@ -498,6 +498,9 @@
         for revision in revisions:
             required_trees.add(revision.revision_id)
             required_trees.update(revision.parent_ids[:1])
+        if ('null:' in required_trees):
+            import pdb; pdb.set_trace()
+
         trees = dict((t.get_revision_id(), t) for
                      t in self.revision_trees(required_trees))

That should change to be an assertion, if you still need it.



=== modified file bzrlib/tests/__init__.py // last-changed:abentley at panoramicfe
... edback.com-20070711194451-3jqhye1nnd02a9uv
--- bzrlib/tests/__init__.py
+++ bzrlib/tests/__init__.py
@@ -100,7 +100,7 @@
 # Mark this python module as being part of the implementation
 # of unittest: this gives us better tracebacks where the last
 # shown frame is the test code, not our assertXYZ.
-__unittest = 1
+#__unittest = 1

 default_transport = LocalURLServer

Why?

Aside from those it looks all good to merge.

-- 
Martin