[Mailman-Developers] Storing additional user data

[David Birnbaum]

On Thu, 10 Jul 2003, Barry Warsaw wrote:

> > Somewhat expensive, perhaps...BUT it has the advantage of being pretty
> > quick to write and very simple for those who don't have the time,
> > knowledge, or resources (#1 and #3, in my case) to build something more
> > complicated.  And, how many requests per second are really going to hit
> > the server, in any case?
>
> I'm not against a command line interface that meets your needs, if
> that's the most expedient way to hook Mailman up to foreign systems.  I
> /would/ however like to have some requirements for that hookup so we can
> tailor the scripts to those specific needs.  For example, the current
> crop of scripts in bin/ were written primarily to solve a problem that
> the human operator was having.  That may not be the most efficient or
> useful interface for hooking two programs up.

Hmmm...well, for this particular application, we pretty much need a simple
way to link in to our main database when bounces occur.  If I were to
craft a quick and dirty interface, it might look like this:

  [command] -e(mail) 'email at address.com' -l(list) 'listname' -b(ounce) [ soft | hard ]

and when a person gets removed (due to bounces, or anything else, I
suppose):

  [command] -e(mail) 'email at address.com' -l(ist) 'listname' -u(nsubscribe)

People can't subscribe directly to the list, we handle that elsewhere, so
the mailing interface and such does not apply.  So this would take care
of my immediate needs.

The other option as was suggested in the list, would be to write a script
that tails the bounce or other logs, and handles things asynchronously
that way.  I'm considering that as an alternative that's also relatively
straightforward to write.

David.