Re: Test suites and RFC2119: meaning of language conformance

On 8/6/2011 4:52 PM, Larry Masinter wrote:
> This seems related to the difficulty with the "normative language reference" issue the TAG raised for HTML5.

Maybe I'm missing something, but I think only in the most indirect way. The 
"normative language reference" issue raised by the TAG, and now 
satisfactorily resolved, had primarily to do with the TAG's concern in 2008 
that the specification for the HTML5 language syntax and semantics was 
available only as part of, and indeed deeply intertwined with, a 
specification for the behavior of a certain class of user agents (browsers).

Since the resolution adopted by the HTML5 working group was to have certain 
information specified in duplicate, I.e. in the original larger 
specification, and in the so-called authoring specification, a need arose 
to deal with the question of which would be "normative", particularly in 
the case of unintended disagreement between them. That question was resolved.

I think the question on the table here now is different, and is focused 
primarily on the use of terms such as RFC2119 MAY, MUST, SHOULD, etc. I 
don't think that was ever directly the subject of disagreement in the 
earlier issue.


Received on Saturday, 6 August 2011 22:25:09 UTC