Re: [rtcweb] Transports: RFC 6062 support

[Harald Alvestrand <[e-mail]>]

On 03/11/2014 10:23 PM, Bjoern Hoehrmann wrote:
> * Harald Alvestrand wrote:
>> On 03/11/2014 09:26 PM, Bjoern Hoehrmann wrote:
>>> * Harald Alvestrand wrote:
>>>> In RFC 2119 theory, MAY and not mentioning it at all have exactly the 
>>>> same weight.
>>> (That depends on whether the default is "allowed" or "prohibited" and
>>> there are plenty of specifications where explicit MAY overrides a de-
>>> fault of "prohibited".)
>> Probably true. But I can't think of any at the moment - do you have
>> something specific in mind?
> (A typical example is when RFC 2119 keywords are used to describe how a
> protocol element is constructed, say, "<foo> elements MAY carry bar=''
> attributes"; you wouldn't assume you can use <foo bar=''> without that.)
Any specific examples?

Without context, this seems a strange way to specify protocol elements;
more common is to use ABNF, and not 2119 keywords.

And ABNF even allows "free" extensibility by having other specs use the
/= operator....


-- 
Surveillance is pervasive. Go Dark.