Noel J. Bergman wrote:
Stefano Bagnara wrote:
Norman Maurer wrote:
First off all i want to create and VirtualUserTable service
At the moment i created the XMLVirtualUserTable and
JDBCVirtualUserTable as implementations.
+1
We should also make the DefaultJamesUsersRepository to implement this
new service so that we could remove the "AliasingForwarding" mailet
and only keep the VUT service.
Agreed, as we did back in May 2006 (see e-mail archives). I figure that
Norman is getting around to working on what we all discussed back then.
HOWEVER ...
The second step ( and the harder step) is to make it manageable. This
is only the case for JDBCVirtualUserTable at the moment. I want to add
JMX and RemoteManager support for it to create mappings and delete them.
+1
Yes, BUT ... this should have some more careful designing. And we do not
need to rush it. For example, UsersRepositoryVUT should be a higher
priority (IMO), and it already has an administrative interface.
Well, I think both are important. Priorities are imho something we
hardly can discuss: Norman is working on this: he will follow his own
priorities.
Of course there is not always so much freedom, but in this case one
thing does not block anything else, so the new code is welcome.
About "this should have some more careful designing": my idea is that
this project need more work, and less designing. When something has been
implemented and have tests we can redesign and refactor it. It is much
more easy to understand each other when the code is already there and
you can talk of specific changes. Otherwise we'll finish up with long
design threads with no conclusions that are often seen in james
archives: 3 years later no one is aware of that discussions and probably
the discussions are outdated.
Of course this is my opinion, and the cause of my +1 ;-)
I think we should introduce a store for VUTs so we can have multiple
VUTs or otherwise also keep the current mailets.
I don't believe that you mean a "store", and if you do, I probably disagree.
I do agree that we could define the VUTs separately from their caller, and
reference them by name, which is what I believe you mean.
--- Noel
Yes, maybe "store" is not a good word in term of patterns and we have
"Store" objects with different behaviours.
I meant to be able to have multiple VUT service implementation and a
way to associate a specific service to a specific user (mailet/handler
other).
Assigning a name is probably the simpler thing: association could be
done via lookup on a directory (what I previously called store), via a
ROLE specification in a dependency injection specification or any other
idea we can have later.
Stefano
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
