[asterisk-dev] RFC4733 SRTP option

[Lawrence Conroy]

Hi Nick, folks,
  as someone who has had to go through this meditation,
the way I have been "guided" to think about these is:
SHOULD == you should do this unless there's a **really** good reason  
not to.
MUST == you must do this or it is broken.

[I wanted to put a MUST in an RFC and the IESG pushed back,
  saying MUST unless... was not appropriate]

Yes, there are large companies in Santa Clara that won't implement  
something unless it's a MUST.
They can and do claim compliance, but ...
That's one reason why * (and others) are in wide commercial use.

all the best,
   Lawrence

On 27 Nov 2009, at 09:20, Nick Lewis wrote:
>> Now, the authors of the rfcs says that implementors should treat  
>> SHOULD as if it was a
>> MUST which sounds funny, but that's the message to all coders.
>
> As long as they do not say that implementors must treat SHOULD as if  
> it
> was a MUST we can claim compliance even without SRTP