On Tue, May 22, 2001 at 03:06:55PM +0200, Sheldon Hearn quothed:
> Although nhash certainly tends to distribute mailboxes evenly across
> subdirectories, people usually ask this question because they need this
> behaviour to interoperate with other applications. Also, the proposed
> splitting scheme has the advantage of being easy for humans to use.
AFAIK, both Courier Imap and SolidPOP use nhash. Since the code is
relativly small (all of 10 lines), it is easy to hack it into any other
source. Nhash has both a perl and a C code that allows you to get the
directory structure given an input. I think that it's still somewhere on
the exim site. If not, I am sure it's not difficult to re-write from
the exim source.
The whole point of spliting directories is to have fast(er) access to
the mailbox. If you split a directory with the first letter(s), then
your directories will be of different size. Unless you choose the format
of the mail addresses.
Otherwise, see crypto theory for a nice frequency of letters, and their
places. From experience, you will get a lot under the
/var/pool/mail/s/e/ tree if you are an ISP. Otherwise, even first names
tend to have a frequency for them.
> The Exim User Guide explains the length and substr expansion items in
> detail.
Isn't that a RTFM? ;>
--
www.kierun.org
Yann@??? Use Pretty Good Privacy.
