On Sat, Jun 28, 2003, Antony Stone <[EMAIL PROTECTED]> wrote:
> On Saturday 28 June 2003 12:13 am, Johannes Erdfelt wrote:
> 
> > On Fri, Jun 27, 2003, Antony Stone <[EMAIL PROTECTED]> wrote:
> > > On Friday 27 June 2003 11:50 pm, Johannes Erdfelt wrote:
> > > > On Fri, Jun 27, 2003, Antony Stone <[EMAIL PROTECTED]> wrote:
> > > > > MailScanner will remove infected attachments from emails and deliver
> > > > > the remainder of the email as it was.   Is that not what you want to
> > > > > do?
> > > >
> > > > No. Like I mentioned further below the quote above, I don't want to
> > > > deliver the body of worms. There's no point to deliver them and will
> > > > just waste disk space and the users time.
> > > >
> > > > That's why I wanted to differentiate between viruses that attach to
> > > > payloads (think infected executable) where the rest of the message
> > > > and/or attachments might still be useful, versus worms which send their
> > > > own emails and as a result, the entire message is useless.
> > >
> > > MailScanner has its own list of such viruses (called "silent viruses",
> > > because it should keep quiet and not inform the apparent sender, because
> > > this is almost certainly a false address), therefore it is possible for
> > > MailScanner to decide what to do with different types of infection, even
> > > if the anti-virus engine (MailScanner supports 15 different ones) does
> > > not supply this information.
> >
> > Ahh, that's good to know.
> >
> > While MailScanner seems to be a fine application, I have various reasons
> > why I want to develop something myself.
> 
> I'm interested to know why that is.   MailScanner is a very capable and 
> well-established application, and it would seem a big task to reproduce this 
> - for what benefit?

Performance is one. Perl isn't as speedy as I want.

It doesn't stop viruses or spam early enough either. I want them stopped
at SMTP time.

I also have a few more features I want to implement.

> > That being said, is there any interest to add similar functionality into
> > the scanning engine? This way the metadata is kept with the signature.
> 
> That does seem like a good idea.   An obvious way to do it might be to have a 
> specific (short) string of characters as the start or end of the virus name 
> which classifies it in this way?

I thought about that at one point, but I don't think it's a reliable way
of handling it.

I wouldn't want to presume that all signatures that start with Worm
are completely bogus and should be dropped.

JE


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


Reply via email to