I would like to tell you that we are already working with email as login
name without so-called plugin for very long time.
The problem is on the SMTP.
you can build it yourself anyway.
Regards
Patrick
----- Original Message -----
From: "Torsten Schlabach" <tschlabach@xxxxxxx>
To: <cyrus-sasl@xxxxxxxxxxxxxxxxxxxx>
Sent: Thursday, January 11, 2007 6:55 PM
Subject: Sponsoring a canon_user plugin for LDAP lookup
Hi!
We are in needed of a canon_user plugin.
The scope is quite simple:
We use technical IDs for mailbox names in Cyrus IMAPd. Our mailboxes are
called mb00001, mb00002, mb00003, ... You get the concept. In our MTA we
have defined aliases for the actual email addresses that point to a
mailbox, so each user can have multiple email addresses connected to a
single mailbox, for example:
info@xxxxxxxxxx -> mb000002
john.doe@xxxxxxxxx -> mb000002
john.doe@xxxxxxxxxx -> mb000002
someome@xxxxxxxx -> mb000816
This is handles by the MTA (Exim) querying LDAP for object like this one:
dn: mailAlias=someone,ou=...
dc: else.com
mailAlias: someone
objectClass: mailAlias
objectClass: mailDomain
objectClass: top
mailAliasedName: mb000816
We've also configured Cyrus IMAPd to query LDAP via the SASL layer where
we store user objects, such as:
dn: uid=mb000816,ou=.....
uid: mb000816
objectClass: account
objectClass: simpleSecurityObject
objectClass: top
userPassword: xxxxxxx
Now users would like to be able to use one of their email addresses to
login, rather than having to remember their mbXXXXXX id.
I understand that this is what a canon_user plugin would be suitable for,
so it would just take the user's input of a username (in this case
someone@xxxxxxxx), do a lookup for a mailAlias object in LDAP which has
dc={domain part of email address} and mailAlias={local part of email
address} and return the mailAliasedName attribute, which id the Cyrus
IMAPd mailbox id.
From looking at a sample I would guess that this will be 100 lines of code
or less. But we don't have any experienced C programmer and as this is
meant for production use, I would appreciate this done by someone who
knows that he doing. And it would be great if this had a chance to become
part of the codebase.
So as nobody obviously felt the need for that before, we're thinking of
paying someone to do it and donate the code. As I think this is a quite
limited scope, we'd be hoping this could be done for a lower 3-digit USD
figure?
Anyone intersted?
Regards,
Torsten
