Michael, > I am completing some testing on new altermime version 0.3.10 for freebsd > (it has already been submitted to ports) > > If you remember, using dkim signing and altermime would add \r\n to > emails if you added disclaimers. > (i have separate plain text and html disclaimers)
Actually just a \r, and not always. > Several emails to [EMAIL PROTECTED] and [EMAIL PROTECTED] > with disclaimers seem to pass now, even with amavisd 2.6.1. altermime 0.3.10 still stumbles with quote-printable encoding. Using amavisd-new-2.6.2(-rc*) avoids the problem with lone \r in altermime disclaimers. > For anyone running dkim, this one should pass. It does. > on seperate note, if SpamAssassin is scoring based on dkim/ pass/fail, > what extra functionality, other then amavisd dkim whitelisting do I get > by using $enable_dkim_verification = 1 ? - invokes DKIM verification regardless of mail size (SA is only invoked for smaller messages - below the configurable limit); - can load a policy bank based on a verified signature through @author_to_policy_bank_maps - and a policy bank can affect most settings of your choice, e.g. whitelist spam or banned contents, add score points, affect quarantine, notifications, mail routing ...; - adjust score through @signer_reputation_maps (using a formula similar to AWL); - add a header field Authentication-Results:, helpful for troubleshooting and potentially useful for (future) MUAs; there is a draft standard on this; - log information about DKIM/DK verification results to facilitate troubleshooting and gathering statistics. If you don't need any of these, by all means, turn off the $enable_dkim_verification to save few milliseconds on nonsigned mail and few tens of milliseconds on signed mail. Also the verification code is not loaded (unless you do DKIM signing), saving a little memory. Regarding a DNS lookup on signed mail, even if you do DKIM/DK verification in both amavisd and in SpamAssassin, it is very likely the DNS result would be cached in your DNS, so you pay the price of a DNS query delay only once. Mark
