commit in brisbane core?

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Robert Collins wrote:
> Could we remove this change please :) its a bogus condition to test for.
> 
>          if self.new_inventory is None:
> +            # XXX: Using these asserts causes test failures. However, at least
> +            #      "self._recording_deletes" seems like a useful check to do,
> +            #      as it ensure the delta is completely valid. Most likely this
> +            #      just exposes that the test suite isn't using CommitBuilder
> +            #      100% correctly.
> +            # if (not self.repository._format._commit_inv_deltas
> +            #     or not self._recording_deletes):
> +            #     raise AssertionError('new_inventory is None, but we did not'
> +            #         ' set the flag that the repository format supports'
> +            #         ' partial inventory generation.')
> 
> 

I'm fine removing it if you are sure it is okay.

However, if self._recording_deletes is not set, then the calling code
hasn't guaranteed that we will properly have 'delete' called. I suppose
for 'record_iter_changes()' it is new code, but it seems like it should
conform to the existing requirement of informing about "recording_deletes".

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAknVMuEACgkQJdeBCYSNAAMXEQCgjjiUPCtDOgdKyHUNMuD0mt8D
acIAn2BeGxv0/fQUdOaZFt/32lroa5xo
=1eD6
-----END PGP SIGNATURE-----