Executable flagging patch

[Aaron Bentley]

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

Aaron Bentley wrote:
> Gustavo Niemeyer wrote:
> 
>>>>I've merged in the two patches - adjusted to fit with the symlink patch.
>>>>I didn't take the generate_cset_optimised->generate_changeset part of
>>>>the change at this point, but all tests are passing.
>>>
>>>
>>>Because the "optimization" is killing entry.metadata_change, which
>>>contains the mode changes, by setting it to None. I have not removed
>>>it because that "optimization" seems so strange that I must
>>>be missing something.
> 
> 
> The name comes from when it was optimizing by comparing the text ids of
> files.  Equivalent functionality is now in ChangesetGenerator.
> 
> The tests should not be passing.  If x-bit changes are now supported, we
> should test that they are merged properly.

Now that I've looked at the changes, I don't think they're enough.
commands like revert and merge should modify the x-bit and that alone,
but it looks like they would also affect the read/write bits.

This is because generate_changeset still uses the ChangeUnixPermissions
object, which affects all permissions.  Instead, I think a new
ChangeExecuteBit object should be introduced that only affects the
execute bit.

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

iD8DBQFDQr6V0F+nu1YWqI0RAgxzAJ9BuoL9j3d467ZhVugX9U6iQY9jtACfaeng
ALf5+Or3JDhhlXQ1L0qMmpo=
=EpWA
-----END PGP SIGNATURE-----