i've digged a bit deeper into that problem, and consulting some people more closer to the source (experienced with postfix and their ml and so on) i was told, that there is a tcp - user dictionary, which would be the prefered way to make postfix backcheck if citadel knows the user. see
I think we should implement a separate module for this, which listens on a
different port, and is disabled by default. Anyone who wants to use it
can enable the service by setting the port number. This would allow
firewall rules to be constructed.
would you make this a compiletime, starttime, or webcit configurable option? btw, using usermaps with different targets could fork spam-default victim users directly to a spamassesin, even before citadel gains it, i think. The firewall issue is true, it would be better that way. where would one
I'm guessing that it would/could be set via text client or webcit, based
on the fact that setting the other port related options are handled from
those interfaces.