[ietf-dkim] Re: t=y

[Dave Crocker]

David Mayne wrote:
> Dave Crocker wrote:
>> 
>> Given that most protocols do not have a 'testing' flag -- and they manage
>> to move into production quite nicely -- a different question might be why
>>  such a flag is needed...
>> 
> Hrm, let's see - the SMTP protocol has EXPN, VRFY, and, well RSET - meaning
> I'm just kidding, reset the session and start over. POP3 has RSET too. What
> about ping, and traceroute, in a more general sense ?

Skipping over the humor, for the moment, these underscore that one can add 
features, albeit as options, without needing a 'testing' bit.  As for ping and 
traceroute, I don't understand.  Traceroute isn't really a hack.  It's a very 
creative re-application of an existing protocol mechanism.


> In the earlier days of DK, one of the hurdles I had to convince our 
> management it was a good idea to experiment with DK signing,  was how would
> receivers handle issues with our implementations,

And that's why I didn't say anything like "a test flag is a silly idea".  It 
isn't a silly idea.  It just turns out that for the Internet, there's not much 
history of needing the flag for functional purposes.  And adding a long-term 
protocol feature for short-term, political expedience seems like rather poor 
engineering (on at least two counts.)


d/

ps. This sub-thread is academic.  I'm not lobbying to change the protocol. 
Just re-issuing my query about whether anyone knows of such a feature being 
particularly helpful in any other Internet protocol.

-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net