Re: 2806 parameter optionality

At 12:52 pm -0400 2/6/02, Henning Schulzrinne wrote:
>>Do folks think that:
>>(a) the presence of a 'm-' string in a parameter label puts it into 
>>the first or second category?
>>(b) there should be a way of indicating the other mandatory category?
>
>In the interest of simplicity, I would argue that there are only two 
>categories, defined by externally observable action:
>
>- m-: if the client does not know the parameter, it must reject the 
>URI (by rejecting the SIP request that contained it, for example, 
>presumably in a manner similar to receiving an unknown URI scheme). 
>The action for the parameter can itself be advisory, although that 
>seems a bit odd.
>
>- Otherwise: the receiver can act on or ignore the parameter.
>
>I don't see how a third category can be usefully observed from the 
>outside. SHOULD and MUST are not descriptions of behavior, they are 
>descriptions of implementations.
>
To which I reply:
  Many Thanks, Henning,
Wanted to raise it, but am happy with the answer.
(It simplifies any parameter drafts, so is a Good Idea :).

I thought that using a URI whilst ignoring a m-parameter like rn
may be externally observable, so I misunderstand the above.

Re advisory action being Odd - yes it is, but under certain circumstances
the 'recipient' of such a parameterised URI may not share trust with the
'sender', so may choose to reconstruct the parameter.

all the best,
   Lawrence
-- 
[email protected]: +44 1794 833666::<my opinions>: