Conference noted::seal

    OK everybody,
    
       That sounds like a good service item. Let's design together
    an algorithm for assigning a number to the pornographic level
    of a message (up to assigning numbers to images included in
    MIME messages). Then, we just need an rc.config variable
    called MAX_PORN_LEVEL that the mail transfer agent (sendmail
    or otherwise) can compare against... :-)
    
       Marc Chatel
    
    P.S.: In practice, you would want to use technology such as PICS,
          but that type of technology is certainly not easily applicable
          to e-mail messages (imagine forwarding all user e-mail to
          a "rating" body. Good grief, forget Orwell, this is better!!!)
    

    Gentlemen,
    
    	please show some (self)control :-).
    
    From a theoretical point of view and also technical, an interesting
    question. A (potential) customer here in Thailand wants some filtering
    as well, but out of fear for virusses.
    
    I guess no solution will ever be 100% as it's still possible to encrypt
    messages of any kind. Even with known encyption methods you would still
    need 'a bit of power' to decrypt, check and decide to forward. With
    unknown encryption method just trying may delay your mail a bit :-)
    
    Probably you could do simply checks on Subject, Contents type if ascii,
    etc. Maybe a VirusWall should include this type of functionality, let
    it seach for type = HIV
    
    Cheers,
    	rudi
    

Just a thought:

Given that AVFW uses sendmail as its local MTA, one could perhaps install
procmail and run it before sendmail? That would give one a modicum of control
over what gets delivered (on eg. a domain name basis) - I'm using procmail to
keep spammer sites off my system, but haven't done any research as to how it
would work in a firewall environment.

Rgs,
	[email protected]

There are several ways of doing this.

Currently, smtpxd drops things into a queue and then invokes sendmail. You could 
change things so sendmail looked in a different queue, and your virus/porn 
checker looked in the first queue and only transferred allowable messages.

Another way is to use a checker such as Mimesweeper, which pretends to be the 
internal mail hub (relative to the firewall).

The only real way is to have a person read it. How is a mechanical checker to 
distinguish between "the government is bad" in the normal sense of the word, and 
"the government is bad" in the Michael Jackson sense of the word?

Either way, it's not going to work. As .2 says, it's all too easy to slip 
something past a mail checker. And if it did work: AOL (?) tried putting a rude 
words filter on their discussion groups, which if I remember correctly, forced 
completely legitimate discussion of "breast cancer" to become "hooter cancer" 
until somebody came to their senses.

Ah, Singapore. A friend of mine landed there a few years back and was forced to 
have a haircut before he was let out of the airport. Chewing gum is banned, too. 

The ISP wouldn't be CyberWay or SingNet, would it?

PJDM

Take a look at MIMEsweeper from Integralis.  That can make "go-no go"
decisions based on sender/recipient/content, or a combination of these.

T