Re: ISSUE: what do we do about tracking when an address is expanded?

The DSN model is that any time an address is translated to more than one
address (this can be an alias or forward operation) then a "success"
DSN is generated and no further DSN requests are passed.  If an alias or
forward operation only goes to a single address, then the MTA can just
pass the DSN request along.  The basic principle is that no more than
one success DSN should ever be generated.

With DSNs, this is important because otherwise you never know whether to
expect further acknowledgments.  But in the tracking case, you have
hop-by-hop information, so it is possible to build the entire tree.

I'm inclined to think that tracking requests to mailing lists should
not be passed on to the recipients; in fact, a tracking request to the
mailing list exploder should probably refuse to give information about
the content of that list, for the obvious privacy reasons.  Alias or
forward expansions is a different case; perhaps an option to decide
whether the track request should be retransmitted to all recipients
would be appropriate.  Aliases/Forwards to single recipients should
probably be passed regardless, using the same logic as DSNs.

If there is no option, then it seems like the only choice is to not
pass tracking requests -- i.e., do the same as DSN requests do now --
for the reasons you outline.

eric



============= In Reply To: ===========================================
: From:  Steve Hole <[email protected]>
: Subject:  ISSUE: what do we do about tracking when an address is expanded?
: Date:  Thu, 22 Jul 1999 16:52:18 -0600

: The problem:
: 
: When we send a mail message to one or more recipients that are exploded 
: either as the result of an alias expansion or a list expansion (note that 
: this may be the same thing for some MTA's), what do we do about tracking 
: the message?
: 
: My opinion:
: 
: The following discussion is based on the parts of the model that we 
: reached concensus on in Oslo.   Most of this is reflected in Tony's notes,
: but until we have a new draft, some of the detail may be lacking.   Also, 
: concensus reached in Olso is based on a small design team group, and is 
: open to discussion by the general list.   
: 
: For the iteractive (query) based tracking, we follow the path that the 
: message takes from MTA to MTA.   Whether we support a referral or a 
: chaining model for the tracking query (another open issue) one of the 
: tracking entities will still have to follow multiple recipient address 
: paths for the message.   If the message originates with multiple 
: recipients, then multiple tracking paths will need to be followed from the
: start.   If one of the recipients explodes into multiple addresses then 
: presumably we will have the capability to track all the resulting paths if
: we want to do so.
: 
: So, the question resolves to one of policy: do we want to track a message 
: beyond an expansion?   The answer might be different for aliases and lists.
: Particularly for public mailing lists.
: 
: I'll put a stake in the ground by saying that we should do at list and 
: alias boundaries whatever DSN does.   We had good concensus on the 
: relationship between DSN and message tracking -- basically DSN is passive/
: asynchronous/pre-hoc message tracking whereas the new protocol extends that
: to do active/synchronous/post-hoc message tracking.   It makes sense to me
: that they should follow the same rules.
: 
: Cheers.
: ---
: Steve Hole
: Messaging Direct
: Mailto:[email protected]
: Phone: 780-424-4922
: