[Mimedefang] Whitelists in mimedefang

[David F. Skoll]

On Thu, 12 Dec 2002, Dan Tulovsky wrote:

> I am looking for a way to have separate whitelists for each domain.  We
> went through this a while back and it turned out that since MimeDefang
> uses a constant spamassassin object, it is not possible to reload it
> half way through.

You can use stream_by_domain to split tests for each domain.

> Now, my question is this.  Will reading in a file every time a message
> comes in slow down the system, or simply make it too slow to be useable?
> Can anyone recommend a better way?

Well, you could use a Berkeley DB, or use a hash-of-lists to read in the
whitelists for each domain into memory.  Reading a large file each time
a message comes in might be slow, but SpamAssassin is already slow, so
you might be OK.

--
David.