[milters] Archive

Lists Index Date Thread Search

Article: 1266
From: Quentin Campbell
Date: 2006-11-17 03:17:30 -0500
Subject: Re: Milter-link whitelisting per recipient problem - debug info & access file included

Removal...........: milters-request@milter.info?subject=remove
More information..: http://www.milter.info/#Support

Anthony Howe said:

>You said you have these entries:
>	milter-link-to:northumbria.ac.uk	SKIP
>	Spam:barbara.tully@northumbria.ac.uk 	FRIEND
>Well I should have remembered that the milter-link-to: lookups come 
>BEFORE the To:, Spam:, and untagged entries and that SKIP 
>would stop the lookup early! (I should have RTFM my own documentation.)
>The complete lookup sequence for this RCPT would normally be:
>	milter-link-to:barbara.tully@northumbria.ac.uk
>	milter-link-to:northumbria.ac.uk
>	milter-link-to:ac.uk
>	milter-link-to:uk
>	milter-link-to:
>	to:barbara.tully@northumbria.ac.uk
>	to:northumbria.ac.uk
>	to:ac.uk
>	to:uk
>	Spam:barbara.tully@northumbria.ac.uk
>	Spam:northumbria.ac.uk
>	Spam:ac.uk
>	Spam:uk
>	barbara.tully@northumbria.ac.uk
>	northumbria.ac.uk
>	ac.uk
>	uk
>BUT the milter-link-to:northumbria.ac.uk SKIP stops this 
>prematurely. So you were originally correct to add:
>	milter-link-to:barbara.tully@northumbria.ac.uk OK

Many thanks for explaining that.

When I re-RTFM I see now where you mention the "lookup order" for
access.db lookups. I vaguely realised any way that lookup order must be
important but missed those words which should have brought ordering to
the front of my thoughts. Perhaps those two words could be in italics to
emphasise the point. 

However the behaviour you describe above is perhaps a little more subtle
than a simple problem of ordering. I found your explanation and example
of a lookup sequence and the effect of the RH value "SKIP" very helpful.

I hope you might consider devising a common "EXAMPLE" sheet for
exploiting access.db lookups in your milters. This could use material
similar to that which you wrote above to help milter users understand
better how lookup sequence and RH values interact and thus avoid the
sort of "gotchas" that confused me.

I may be alone in having missed that and am happy to acknowledge that
other users of your milters may not have the same problems in
understanding all the complexities implicit in the features provided by
your milters!  

Kind regards


Lists Index Date Thread Search