Re: A little help on SAEximRunCond & rcpt domain based lookups
Sander Smeenk <[email protected]>
| Newsgroups | gmane.mail.exim.spamassassin |
|---|---|
| Message-ID | <[email protected]> |
Quoting Chirik ([email protected]): > Instead, do the checks during the master check_rcpt acl, and set one of > the 'acl_m#' variables, I did what you suggested. And this works. I can now switch on and off the spamassassin scans from within my policy file for each domain seperately. > Note, this will result in SAEximRunCond being true if *any* domain for *any* > recipient has it set true. You can adjust it to default to run, and set it > to not run if any domain doesn't want SA run, instead, if you'd like. Whoo :) You lost me. If a message comes in with, say, two RCPT TO:'s, and one of them has spamassassin=yes in it's policy, and the other hasn't, the message will get scanned? That's ok I think. Wouldn't the same happen when I default to scan, and set for specific domains not to scan? Wouldn't it be the same? One has exim-sa set to not run, so both RCPT TO's won't get scanned? > Personally, I think using the acl_m# variables for this purpose is better > than using headers - they get reset when a new message is started, and you > don't have to take precautions to prevent SA not being run if they happen to > be set on an inbound message ... unless you want that, of course. Funny is that I didn't come to think of setting acl_m# variables. I thought they wouldn't be available in SAEximRunCond either, because they are ACL variables. But, yes, I want to scan incoming messages. Unless the message was received from localhost / localIP or the 'don't scan' header is set :) But SAEximRunCond has correct {eq{}{}} things for that. > In my case, I use acl_m0 as a signal to whether to run spamassassin, > and whether or not to reject messages or just flag (I set it to > 'do-not-reject' in acl_check_rcpt, and that is overridden by > 'do-not-run' in acl_check_data) I use acl_m1 as a count of recipients > I accepted, or would have accepted, were I not lying and telling the > remote site that anything they tried was invalid because they were > just guessing at usernames. ;-) I'll first see how this works out before I start doing kinky stuff like making rejecting optional ;) > I really like exim, and it works great for my home systems - I'm looking for > alternatives to deploy at my company, and I'm not sure exim is appropriate > there. :-/ Hmm, with exim4's ACL's I have the idea anything is possible ;) Thanks alot for helping me out this much... I'll continue doing tests now! Sander. -- | I doubt, therefore I might be. | 1024D/08CEC94D - 34B3 3314 B146 E13C 70C8 9BDB D463 7E41 08CE C94D