Re: spam category for Mail::DeliveryStatus::BounceParser

2009-11-24 Thread Michael Stevens
2009/11/24 William Yardley p...@veggiechinese.net

 On Fri, Nov 13, 2009 at 11:03:10AM -0500, Ricardo Signes wrote:
  * Michael Stevens michael.stev...@dianomi.com [2009-11-13T10:20:58]

   I'm pondering adding a spam category to std_reason in
   Mail::DeliveryStatus::BounceParser.
  
   This would record when a message was bounced because the recipient
   considered it spam.
  
   Does this sound reasonable?
 
  Yes.

 I think there had been some talk about this a long while back, when I
 was using MBP for a project  working on this a little more actively.

 The only issue to me is that calling it spam doesn't make it super
 obvious whether the message itself is spam (i.e., not a bounce at all,
 but spam), or whether the bounce parser thinks that the message was
 rejected because the intended recipient considered it spam (I think it
 might actually be useful to have a spam classification for messages
 tagged as spam by an external filter, but passed along to the
 bounce-parser).


It's a potential problem, with the name spam, although if they read the
documentation they'll see it explains what's considered spam (see my tree on
github).


 On a more technical level, I think it may be difficult to do this
 reliably. Presumably, dnsbl based rejections (e.g., 554 10.0.0.1 blocked
 by
 dnsbl.example.com), anything containing the word spam (554 message
 appears to be spam), and other low hanging fruit would be easy
 enough to match, but I think any such logic should definitely err on the
 side of caution. Plenty of policy based rejections aren't due to spam.


At the moment the classifications I've done are only on fairly obvious you
have been blackllisted / are considered spam examples, so hopefully we're
okay here.

I don't have as much time to spend looking at bounces as I have had in the
past few weeks, but I still hope to keep tweaking things on and off.

-- 
Michael Stevens
Dianomi Ltd
18 Buckingham Gate
London SW1E 6LB

Tel: 020 7802 5530
Fax: 020 7630 7356
www.dianomi.com

The information in this message and any attachment is intended for the
addressee and is confidential and may be subject to legal privilege. Dianomi
Ltd, Registered Office: One America Square, Crosswall, London. EC3N 2SG.
Registered in England and Wales with Company Registration Number 4513809.
VAT registration number: 809754988


Re: spam category for Mail::DeliveryStatus::BounceParser

2009-11-23 Thread William Yardley
On Fri, Nov 13, 2009 at 11:03:10AM -0500, Ricardo Signes wrote:
 * Michael Stevens michael.stev...@dianomi.com [2009-11-13T10:20:58]

  I'm pondering adding a spam category to std_reason in
  Mail::DeliveryStatus::BounceParser.
  
  This would record when a message was bounced because the recipient
  considered it spam.
  
  Does this sound reasonable?
 
 Yes.

I think there had been some talk about this a long while back, when I
was using MBP for a project  working on this a little more actively.

The only issue to me is that calling it spam doesn't make it super
obvious whether the message itself is spam (i.e., not a bounce at all,
but spam), or whether the bounce parser thinks that the message was
rejected because the intended recipient considered it spam (I think it
might actually be useful to have a spam classification for messages
tagged as spam by an external filter, but passed along to the
bounce-parser).

On a more technical level, I think it may be difficult to do this
reliably. Presumably, dnsbl based rejections (e.g., 554 10.0.0.1 blocked by
dnsbl.example.com), anything containing the word spam (554 message
appears to be spam), and other low hanging fruit would be easy
enough to match, but I think any such logic should definitely err on the
side of caution. Plenty of policy based rejections aren't due to spam.

w