PING [MERGE] Root entry has a revision id

[Aaron Bentley]

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

John Arbash Meinel wrote:
> Aaron Bentley wrote:
> 
>>I haven't heard anyone's opinion of this, and I think it's a bit too
>>major for me to just merge without review.

> I wasn't commenting on it, because it has too much of the root == None
> stuff, which you need to sort out between you and Robert.

This change doesn't prejudice our decision about whether the origin
revision a root entry.  My motivation in introducing
Inventory(root=None) is to regularize the code.  It reduces the number
of places where we need to special-case the root entry.

So I don't think that this depends on the outcome of Robert's and my
discussion.

> The one thing I didn't like was using 'NewCommitBuilder' as the class
> name. If it is truly meant to be temporary, then I think using
> '_CommitBuilder' would be better.

I can do that.

> Though the way you were writing it, you really should have done the
> changes in NewCommitBuilder, rather than in the base class.

You mean, have CommitBuilder be the new class, deriving from
_CommitBuilder, and overriding the record_root_entry to False?

> Otherwise you are changing a base implementation, away from classes that
> are using it.

Since CommitBuilder is a public symbol, I didn't want to change its
behaviour, if I could avoid it.  The way I've written these changes,
classes that derive from CommitBuilder should continue behaving the same
way, but they can set record_root_entry True when they update to the new
API.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFE4KUK0F+nu1YWqI0RAsKpAJ9W7WfVXqQW/oeBNEWoWV+1Pa+l+gCfcexR
Gh3UlxcmLj7jF4u6/NhUz5I=
=bT21
-----END PGP SIGNATURE-----