[attn aaron] differences between RevisionTree and WorkingTree

[Aaron Bentley]

Robert Collins wrote:
> On Thu, 2006-07-20 at 10:38 -0400, Aaron Bentley wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Robert Collins wrote:
>>> Looks to me like the cleanest way to fix this is for
>>> inventory.iter_entries() to yield self.root as well as the other
>>> elements.
>>>
>>> That sound ok to everyone ?
>> I think you'll wind up duplicating a lot of the effort in nested-trees.
> 
> If you can guide me such that I do not do it differently, that would be
> great. However, to write the tests for the new code paths I'm bringing
> in, I need RevisionTree and WorkingTree to be more homogenous than they
> are at the moment :(.

I overreacted a bit.

I think what can happen if you alter iter_entries is that compare_trees
will start behaving differently, because it will include the root entry
in its iteration.

However, the root entry will only show up in the list of unmodified
files, so that's fairly easy to paper over.  It's not as nasty as it was
when the root id was showing up as a new entry in some places.

Aaron