Re: FYI: Content-MD5(-Origin) hacks

[Martin Hamilton <]

-----BEGIN PGP SIGNED MESSAGE-----


Richard Kettlewell writes:

| [Martin, I don't know if you're subscribed to anti-spam@localhost,
| hence the CC.]

Hi Richard - yep, I'm on the list.

| Actually these two are special cases of the general problem where two
| users at the same site are sent the same message by different paths.

I think there's two sides to it - on the one hand we have hacked 
versions of mail hub programs which tag each message with its 
Content-MD5: hash, and on the other hand we have delivery and filter 
programs which could decide to do things with the hash info.

So, if you try those hacks to Exim or sendmail, all that'll happen is 
your messages will get Content-MD5: headers tagged on to them.

The procmail thing is very much a simple knee-jerk reaction - in
practice I think you'd probably want to have a threshold as well, e.g.
N attempts to send the same message cause it to trigger an alarm.
Before trying to do anything more sophisticated like this, I thought
I'd try to get some feedback from people on the basic idea of using
Content-MD5 by hacking up a quick proof-of-concept.

Having said that, yes, I'm running it on my mail - and filtering out 
duplicate messages with the same contents is something I've wanted to 
have in my mail system for a long time.  Not sure whether I'm the 
exception or the norm in this respect, though!

| * Sending a message with Qmail, which AIUI makes a separate SMTP
| connection for every recipient of every message.
| 
| Of course you might choose to consider this behaviour "pathologically
| broken".

:-))

Ciao!

Martin



-----BEGIN PGP SIGNATURE-----
Version: 2.6.3i
Charset: noconv

iQCVAwUBNSNe69ZdpXZXTSjhAQGehAQAlKaqolIOHeg5+I+gPPFQ409lv6xTaxwV
NOVgS4GFHZYDvOumt0p+6ZuVrKxTIAr4YgDSqk8+5mlAp30wkQKRx32HGGrf+SH7
fWNtiVD6rXR5Traid+28f7kY5OgKHfxSAWC5ibAW+gTfx76f7L4YiULh+Cc4/04r
Yix+netJsJI=
=bRqW
-----END PGP SIGNATURE-----