Re: [apps-discuss] Working Group Last Call draft-ietf-appsawg-sieve-duplicate

[Hector Santos <[e-mail]>]

On 1/2/2014 1:24 PM, t.petch wrote:

> "   Duplicate messages are normally detected using the Message-ID header
>     field, which is required to be unique for each message.  "
>
> REQUIRED maybe, but I seem to recall the malformed-mail I-D raising the
> possibility that it was not.  In which case, ...?

Small input:

When we dealt with dupe issues in Fidonet, the Message Id was REQUIRED 
to be unique only at the PER DOMAIN basis. Because you could not 
guarantee that it would be unique at other domains nor guarantee that 
it would exist, the first main goal is to make sure you stop DUPES at 
the local domain level first.  I don't see this as any different as we 
migrated to RFC 822/2822/5322 since the same or similar dupe problems 
(and design requirements) applies.

That said, in my experience, dupe methods can range to include other 
factors in an dupe algorithm.  In addition, it depends on the type of 
mail.  For example, in a 1 to 1 (private direct mail), we had to offer 
an option to DISABLE dupe checking for private netmail communications. 
  However, for 1 to Many (public, list, echo, news), there are other 
means included.   Fidonet had "seen-by" lines.  NNTP News Articles has 
the path statement as well to detect dupes.   Our 30+ year old system 
still has a proprietary hashing method that considers TO, FROM, DATE 
and SUBJECT and the body and that normally catches what any RFC MSGID 
methods misses.

-- 
HLS