Przemyslaw Wegrzyn <[EMAIL PROTECTED]> writes:

> On 7 Aug 2001, Scott Gifford wrote:
> 
> > While we're on this topic, is there a reasonable way to reliably
> > guarantee uniqueness across both the 'mail' and 'mailalternateaddress'
> > attributes?  We finally stopped using 'mail' altogether, because we
> > couldn't find a way to make this work without race conditions.
> > 
> 
> For me it's similar to UNIX hard links - many people think of
> "original" file name and hard links to a file. But actually every
> file's name is a hardlink to a particular inode, even the first one.
> 
> Let's return to qmail-ldap -> what is exact, semantic difference
> between "mail" and "mailAlternateAddress" ?  Can't we just use, say,
> "mailAddress" ? What's the actual difference between mail address
> and alternative mail address ?

There is no functionality difference; deliveries are performed to the
mailMessageStore which shows up when you search for:

  (|(mail=address@host)(mailAlternateAddress=address@host))

Henning makes the point, though, that it's useful to know what the
user's "primary" address is.  See below, however...

> I'm not familiar with LDAP so well yet, don't blame me if I'm totaly
> wrong. If we use just one field, we can make it unique by schema
> definition, and simply "schemacheck on" saves us much trouble,
> mentioned in Scott's post ?

Right, that's what we did; we had all email addresses in
mailAlternateAddress, then the user could select *one* of those
addresses as their primary email address, and that one went in their
mail attribute.  The interface wouldn't let them add a mail attribute
unless it was already one of their mailAlternateAddresses, and a
uniqueness setting on the mailAlternateAddress guaranteed uniqueness.

We could have taken this to the next level, and asked qmail-local to
just search for

    (mailAlternateAddress=address@host)

but we never got around to it.

Anyways, sorry for taking this thread off on a tangent; I was just
wondering if somebody knew a way to tell an LDAP server "guarantee
uniqueness across both of these two attributes", and it looks like you
can't; you just work around it in your interface.

----ScottG.

Reply via email to