[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Extending 'mbox:' URL

From: Sergey Poznyakoff
Subject: Extending 'mbox:' URL
Date: Tue, 15 Jan 2002 15:43:40 +0200


I'd like to propose an extension to the existing 'mbox:' url format
that will allow to use mailspool hashing schemes. Such schemes reduce
the mailbox access time for sites with a big number of mail boxes.
Widely deployed hashing schemes are:

 1. Indexing by first 'n' letters of a login name. For example, when
    n=2 the mailbox for user 'gray' is '/var/spool/mail/g/r/gray'.
    When n=0 this produces usual mailspool names.
 2. Indexing by last 'n' letters of a login name. This provides a
    better average distribution than the first method.
 3. Hashing first 'n' letters of a username.

Here is the proposed url format:


Keywords are:
 type  -- The hashing scheme type. E.g.: "index", "rev-index" and "hash"
          for the three schemes described.
 param -- Number of letters from the username to be used in hashing scheme.
 user  -- A username.

For example, the url


produces the following mailbox path:


When no keywords are used, the url coincides with the format currently in
use by mailutils. 
If user= is used as the last keyword, the url allows for creating the mailbox
names by simply appending a username to the url, so any existing code like

    sprintf (buf, "%s/%s", _PATH_MAILDIR, username)

could simply be rewritten as

    sprintf (buf, "%s%s", maildir, username)
(Assuming maildir is in mbox: format).

I have prepared an implementation of this extension. Any objections
if I commit it?


reply via email to

[Prev in Thread] Current Thread [Next in Thread]