Getting rid of RootEntry?

[Aaron Bentley]

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

John A Meinel wrote:
> With the future work of having nested trees, we were thinking to have a
> new inventory type inside the parent tree. 

Yes, with by-reference nested trees, the nested root would be a new
type, (e.g. NestedRoot).  A NestedRoot would have a revision, and
possibly a checkout, branch, etc associated with it.  Mind you, it will
only be a NestedRoot in the containing tree.  It would be the same type
in its own tree.

With by-value nested trees, the nested root must be an
InventoryDirectory, which will mean we have to address this issue.

> But that doesn't mean it
> needs to be a distinct entry inside its own tree.
> Because of a desire to be able to merge branches, it seems like it would
> be nice to be able to move the root entry.

Yes, I've come to believe that.  (And yes, I know I dismissed the idea
of moving the tree root as nonsense during my Arch days.)

> We may want #3 anyway, but I'm okay with #1 for now.

I know where you're coming from.  In some ways, it's nicer to permit
type changes, but it does introduce a whole bunch of new edge cases.
I'll note that the TreeTransform code does handle type changes, because
revert must be able to revert a type change.

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

iD8DBQFDy65W0F+nu1YWqI0RAmssAJ9o/JM1NeEelludUgZlrFIvloGbKACfT1uw
eq1Osn/2Lc0pp0Rg9ZmeWTQ=
=3IO8
-----END PGP SIGNATURE-----