[Mimedefang] questions about filter parsing, message mod'ing, and bouncing messages

TLD MimeDefang tld.mimed at stimulacra.com
Mon Aug 31 12:03:04 EDT 2009

Thanks for the reply David!

David F. Skoll wrote:
> filter() is called once for each leaf MIME-part, and filter_multipart()
> is called once for each non-leaf MIME part. Keep your SA tests in
> filter_begin or filter_end.

   Makes sense..  I wasn't really sure when those two subs were called,
so that helps to clarify, as well with the question of putting Spam
Assassin into the filter_begin. I'm assuming that I can also put
other file attachment name filtering functions into the filter_begin
as well - we severely restrict the types of file attaches that are
allowed in and out of our office and refuse messages containing
certain prohibited types.

> MIMEDefang does not currently support multiline replies (it's on the
> TODO list.)  There's also no guarantee end-users will see the text;
> some MTAs eat it.

   I guess I'm just curious how some companies do a more robust
reply in a bounce message. It's not really important, only that I'd
like to be able to provide a URL for assistance in false-positives.
But as I said it's not really important.

>> So, I figured I could use the action_notify_sender() to do this,
> Don't do that.  You'll be blacklisted for generating backscatter
> within hours.

   Crud, I hadn't thought of that.  :(  Thanks for the warning.

   Thanks a bunch for the advice, input, and suggestions!


More information about the MIMEDefang mailing list