>From: "James Ebright" <[EMAIL PROTECTED]> >> If he knows for a fact that he wants to reject those emails >> outright, then rejecting them in filter() and avoiding a call to SA >> is the more prudent choice. > >Well, while true it would be more efficient, I would not call it more prudent, >to me allowing the end user the choice to decide what is and is not spam when >the only positive hit might be the charset is more prudent, you can easily >allow for that in SA. Basically you loose the flexibility SA allows for a bit >more efficiency (probably a moot amount unless you are slammed with these >types of messages). But, as you pointed out, it is his call.
I have weathered amazing, tho brief, storms of SPAM. I agree that universally rejecting based on the 6 final charsets (I eliminated the "windows_125x" sets from my list) does lose some of the flexibility of SA, but like I said, in my 10 year mail archive, none of those 6 character sets were associated with HAM. My analysis of my individual situation (vis-a-vis the services I provide to the population I support) is that rejecting E-Mail using those character sets is a reasonable balance against imposing my SPAM rules on the user population and efficiently distributing my resources. >Also, Dirk, if you have a spam corpus of 10 years then you need to read the >thread about prior art and some recent patent debates, your information may be >usefull... [or were you exagerating a tad ;-) ] Got a link? Actually, I confess that my SPAM corpus is not that old. Probably only 2 years, 3 at most, and then much of that resides on backup media as opposed to online. My HAM corpus does indeed go back 10 years. I confess to being an E-Mail packrat. Dirk _______________________________________________ Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang