[Mailman-Developers] New RFC on using DKIM with MLMs

C Nulk CNulk at scu.edu
Mon Nov 28 20:01:58 CET 2011

Hello Ian,

Sorry I am a bit late in responding but with our Thanksgiving holiday
and me taking a few days of vacation I was away from email.

I understand what you are saying in your message but I think you
possibly missed my point.  Given the definitions previously proposed for
the "Mediator" header, I believe it is far too generic to be useful.

To paraphrase the previous definitions for the "Mediator" header,

   "any program/instance/etc. that modifies, adds content, regulates
Users, etc. can use it [the 'Mediator' header]".

To me, a reasonable reading and understanding of the definition[s] tells
me the MUA, the MTA, any appliances, any other program/instance
[including devices]/etc. can add the "Mediator" header REGARDLESS of
whether or not that appliance/program/instance/device has added a
"Received" header, a "User-Agent" header, or any other pertinent header(s).

Since I think the "Mediator" header is, at this time, poorly defined and
the initial point of the discussion was about having a "List-Agent"
header for MLMs, I say "List-Agent".

As for my example you pulled apart :), I deliberately made it a worst
case scenario just to highlight how the "Mediator" header was poorly
defined.  In fact, given the "Mediator" header definitions and your
comments, you also highlight how poor the definition is since not only
will the "Mediator" headers be added but also all the additional
"Received" headers.  I may be wrong in that you assume that if
"Received" headers are used then the "Mediator" headers won't be used.
However, the definition does not precluded having both, thus doubling
(or more) on the number of headers.

Again, I say stick with the "List-Agent" header for MLMs.  It is
narrowly defined and limited in scope, precisely what a header should be
to be effective.

Thanks for listening/reading :),


On 11/23/2011 5:16 AM, Ian Eiloart wrote:
> On 16 Nov 2011, at 16:10, C Nulk wrote:
>> I can't help but consider the rash of useless Mediator headers.
>> Consider the following example:
>> Person 1 sends a message to a list which is then sent to Person 2.
>> Person 1's site has separate appliances for the MTA, Spam checking,
>> Anti-Virus, and Spy-ware, likewise for the receiver's site and the site
>> the MLM is for the list.  My reading of the above definitions tells me:
>> 1. Person 1's MUA adds a Mediator header (creating a message is a simple
>> special case of modifying/adding a message,
> But, there are already headers to identify the creator. There's no need to use a mediator header to duplicate other information.
>> 2. Person 1's site adds 4 additional Mediator headers (one each for the
>> MTA, Spam, Anti-Virus, Spy-ware since each modify/forward/add to the
>> message,
> The MTA would use a "received" header, not a mediator header. If the message is routed through the other systems, then they should already add received headers. If the message isn't routed through those systems (e.g., the systems are plugged in to the MTA, then there's no need to add mediator headers).
>> 3. The MLM's site adds 9 additional Mediator headers (4 inbound [item
>> 2], 1 for the MLM [maybe more], and 4 outbound),
>> 4. Person 2's site adds 4 additional Mediator headers (4 inbound [item 2]),
>> 5. Person 2's MUA may or may not add a Mediator header depending on any
>> rules/filters Person 2 has in place.

More information about the Mailman-Developers mailing list