Rev 4563: Update from Martin's review feedback. in http://bazaar.launchpad.net/~jameinel/bzr/1.18-2a-bundle-performance

[John Arbash Meinel]

At http://bazaar.launchpad.net/~jameinel/bzr/1.18-2a-bundle-performance

------------------------------------------------------------
revno: 4563
revision-id: john at arbash-meinel.com-20090804135008-u86i3o3nwkp1axh6
parent: john at arbash-meinel.com-20090730203044-4nj22r3fw1y4d4zg
committer: John Arbash Meinel <john at arbash-meinel.com>
branch nick: 1.18-2a-bundle-performance
timestamp: Tue 2009-08-04 08:50:08 -0500
message:
  Update from Martin's review feedback.
-------------- next part --------------
=== modified file 'NEWS'
--- a/NEWS	2009-07-27 18:03:19 +0000
+++ b/NEWS	2009-08-04 13:50:08 +0000
@@ -33,7 +33,10 @@
 * ``bzr mv`` no longer takes out branch locks, which allows it to work
   when the branch is readonly. (Robert Collins, #216541)
 
-* ``bzr send`` now generates valid bundles with ``--2a`` formats.
+* ``bzr send`` now generates valid bundles with ``--2a`` formats. However,
+  do to internal changes necessary to support this, older clients will
+  fail when trying to insert them. For newer clients, the bundle can be
+  used to apply the changes to any rich-root compatible format.
   (John Arbash Meinel, #393349)
 
 * Fixed spurious "Source branch does not support stacking" warning when

=== modified file 'bzrlib/bundle/serializer/v4.py'
--- a/bzrlib/bundle/serializer/v4.py	2009-07-28 20:21:47 +0000
+++ b/bzrlib/bundle/serializer/v4.py	2009-08-04 13:50:08 +0000
@@ -321,18 +321,32 @@
                                 inv_vf, self.revision_keys)]
         revision_order = topological_order
         if self.target is not None and self.target in self.revision_ids:
-            # Make sure the target is always the last entry
+            # Make sure the target revision is always the last entry
             revision_order = list(topological_order)
             revision_order.remove(self.target)
             revision_order.append(self.target)
         if self.repository._serializer.support_altered_by_hack:
+            # Repositories that support_altered_by_hack means that
+            # inventories.make_mpdiffs() contains all the data about the tree
+            # shape. Formats without support_altered_by_hack require
+            # chk_bytes/etc, so we use a different code path.
             self._add_mp_records_keys('inventory', inv_vf,
                                       [(revid,) for revid in topological_order])
         else:
+            # Inventories should always be added in pure-topological order, so
+            # that we can apply the mpdiff for the child to the parent texts.
             self._add_inventory_mpdiffs_from_serializer(topological_order)
         self._add_revision_texts(revision_order)
 
     def _add_inventory_mpdiffs_from_serializer(self, revision_order):
+        """Generate mpdiffs by serializing inventories.
+
+        The current repository only has part of the tree shape information in
+        the 'inventories' vf. So we use serializer.write_inventory_to_string to
+        get a 'full' representation of the tree shape, and then generate
+        mpdiffs on that data stream. This stream can then be reconstructed on
+        the other side.
+        """
         inventory_key_order = [(r,) for r in revision_order]
         parent_map = self.repository.inventories.get_parent_map(
                             inventory_key_order)

=== modified file 'bzrlib/repository.py'
--- a/bzrlib/repository.py	2009-07-22 19:00:01 +0000
+++ b/bzrlib/repository.py	2009-08-04 13:50:08 +0000
@@ -2190,7 +2190,6 @@
         """
         if ((None in revision_ids)
             or (_mod_revision.NULL_REVISION in revision_ids)):
-            import pdb; pdb.set_trace()
             raise ValueError('cannot get null revision inventory')
         return self._iter_inventories(revision_ids)