On Fri, 2002-04-26 at 17:48, Scott M Likens wrote: > --On Friday, April 26, 2002 1:25 PM -0700 julesa <[EMAIL PROTECTED]> wrote: > > From a technical/efficiency standpoint, sieve should not have to check > > an external source of information before delivering a message. Inserting > > a header, and then letting the user use Sieve to either reject the > > messages, accept them, or file them into a Spam folder is the correct > > approach, IMHO. > > Honestly i forsee the thought of Sieve having to check each message as it > comes in, as a huge process and having to resolve the hosts and everything > just compounds things. Sieve is not a MTA, and i wish people would think > about the processing power it might take for their choice. That's all.
Sorry, I wasn't very clear. I meant to say that the MTA should check to see whether a particular message looks like spam, then insert a header marking the message as such, and pass the message to Cyrus. Then the user can use Sieve to do whatever they want with the message. I agree with you. I think putting the spam detection procedure inside of Sieve (instead of letting Sieve check for a header placed there by the MTA) is a bad idea. But I don't see it as a problem with processing power so much as a problem with efficiency and reliability. Problems with processing power are easy to fix: buy hardware. But there are a lot of other reasons why you don't want a Sieve process waiting around for the spam-check logic to finish what it's doing every time Cyrus receives a message. Better pray you don't get mail-bombed, and have lots and lots of RAM. :-)
