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

[Richard Damon]

On 10/24/11 10:31 AM, Murray S. Kucherawy wrote:
>> -----Original Message-----
>> From: Ian Eiloart [mailto:iane at sussex.ac.uk]
>> Sent: Monday, October 24, 2011 7:24 AM
>> To: Murray S. Kucherawy
>> Cc: mailman-developers at python.org
>> Subject: Re: [Mailman-Developers] New RFC on using DKIM with MLMs
>>
>>> Isn't "hide" a function of the MUA, not the MLM or MTA?
>> No. "Display" or "Expose" might be a function of the MUA. "Find" might
>> be something that the user does. To hide something is to put it in a
>> location where it's unlikely to be found.
>>
>> As long as you know that people aren't looking there (and they're not),
>> putting information in a header (other than to, from subject and date)
>> is "hiding".
> My point is that if using header fields is the right way to encode this information in a protocol sense, then the issue is really that the MUAs need to expose that information somehow.
>
> I have some trouble with the assertion that making the MLM and MTA do what we all agree is the right thing to do constitutes "hiding".
>
The problem is that the Core Mail RFCs (822/2822) doesn't say that the 
headers are the right place to put this info, and thus it can't be 
assumed that MUA should display such a header. The LIST_UNSUBSCRIBE 
header is only defined in RFC 2369, but not the core RFC 2822 (Internet 
message format), as such the existence of MUA that don't understand it 
can be reasonable expected (as is seen in practice). The conflict in 
needs between DKIM and MLM becomes a legal issue is some jurisdictions. 
The LAW (which trumps any RFC) demands a conspicuous notice, so the MLM 
must provide it. The header provided in RFC 2369 is not sufficient due 
to too many MUAs not complying with it. This makes DKIM suggestion to 
only use it (and not add the link to the message) actually illegal in 
some places.

Since the MLM were around first (and the law), this becomes a defect in 
DKIM.