bug? iter_entries() skips from_dir entry

[Aaron Bentley]

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

duchier at ps.uni-sb.de wrote:
| Aaron Bentley <aaron.bentley at utoronto.ca> writes:
|
|
|>| It's intentional, but it's not very clean and should be cleaned up, by
|>| me deciding whether the root entry should be seen as a member of the
|>| inventory or not.
|>
|>I don't see a need to special-case it.  But of course, I can handle it
|>either way.
|
|
| Correct me if I am wrong, but if the root is part of the inventory,
then it can
| be moved (by a topological delta).

True.  I see this as occasionally useful, but moving the root is
complicated, so we can forbid it for now, and implement it when we want it.

An explicit root also allows you to graft trees together in interesting
ways.  You can copy the root of A into B as a subdirectory, and still be
able to

| It has also the further implication that the
| inventory is a tree, not simply a forest, and that all topological
deltas must
| preserve that invariant.

All versioned files, except for the root, are already required to have a
versioned parent directory.  This is so that merging file moves can be
handled sensibly.

| I have not thought about this much at all, but I get the uneasy
feeling that it
| could make things slightly more complicated (unless, you, in one way
or another,
| "special case" the root).

Oh, I agree the root must be special in some ways.  I mean I don't think
it needs to be a hidden entry.  Plus, it's hard to hide the entry
completely, and that way lies complexity.  That's my just my $0.2CAD.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.5 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFCZlHr0F+nu1YWqI0RArMmAJ9ddfF+Ni0S+0l6ytXsuuqfx2B+UwCdGZQq
+xMpEILwmzjQJYeUnRgzkdE=
=OR5Y
-----END PGP SIGNATURE-----