Hi Jason "Jason R. Mastaler" <[EMAIL PROTECTED]> writes:
> > The problem with this, as you've explained, is that keyword and > > sender tagged messages may be held if SpamAssassin falsely > > identifies the message as spam. > In your original example, spamc identified the legit mail to the > keyword address as non-spam, so I think you're ok there. Sigh. I think the logic is confusing matters. You have said that filter rules will be applied to messages tagged with sender and keyword addresses even though they should be delivered (as you have said, this is so that you can have a chance to drop keyword addresses that are abused and so on). That I understand. However, as a result, if spamc *falsely* identifies a message to a keyword address (for instance), it will be held and not delivered. I want all keyword and sender tagged messages to be delivered, even if they look and smell like spam according to spamc. > > Is there a way to hook something in once TMDA starts the generation > > of a confirm message, so that it can be stopped if the message looks > > like spam? > I'm not sure what you mean. Can you elaborate? Forget about the pipe rule in my filters/incoming for a moment. Pretend all I have is: from-file ~/.tmda/lists/whitelist ok from-file ~/.tmda/lists/whitelist-confirmed ok with a default ACTION_INCOMING of "confirm". Now, anything that gets as far as hitting the default action will generate a confirmation message (which messages tagged as sender or keyword will never reach). At the last moment, I want to be able to say "but only confirm if the message doesn't look like spam". Is there anything to let me do that? In case you missed it earlier, my objective is a simple one: don't send confirm messages in reply to messages that look like spam. Ben _____________________________________________ tmda-users mailing list ([EMAIL PROTECTED]) http://tmda.net/lists/listinfo/tmda-users
