-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Manvendra Bhangui wrote:
> 1. vdeluser appends the directory being deleted in a MySQL table (for
> file for cdb backend). The latest directory deleted will be at the
> end. Let's say this filename is dir_control_free
This would not work because users can be deleted out of the hash tree
anywhere. It appears your patch assumes a FILO ordering of user additions
and deletions.
> 2. vaddduser locks this file (dir_control_free). Picks up the first
> entry and assigns the new user the first prefix and after creating
> the user and the Maildir, deletes the line. No changes will be made
> to dir_control if a entry exists in this file. When all entries are
> backfilled, dir_control will again get used.
If the hashes, 'a' through 'd' existed, and the 'b' hash directory cleared
out, your method would fail to backfill correctly.
Furthermore, if I'm assuming other details correctly, it seems as if
dir_control would never pick up again because you would never backfill 'all
entries' because dir_control and dir_control_free would never match.
If you have some further details, please let me know. I don't believe that
the current hashing solution will support backfilling without major hacks.
It may make sense to re-implement hashing, keeping backfilling in mind.
Fortunately the dir_control method is quite simple, and in the event a new
hashing method is devised, conversion to a newer method should be rather
straightforward.
- --
/*
Matt Brookings <m...@inter7.com> GnuPG Key D9414F70
Software developer Systems technician
Inter7 Internet Technologies, Inc. (815)776-9465
*/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFJZ2XF6QgvSNlBT3ARAi5XAJ9VFl6ajjM48TyTBnR/6s5/wleU3gCfRl2z
WOnitBBSXhg9P8iI3FB8Xog=
=bhWd
-----END PGP SIGNATURE-----
