Re: LC-67 leftover -- MUST use MUST?

At 11:11 -0600 2003-06-25, Alex Rousskov wrote:
>Specs SHOULD use RFC 2119, IMO. I am surprised the issue is worth
>debating. Clearly, some specs may want to have more conformance levels
>than a single level offered by RFC 2119. In RFC 2119 "SHOULD" means
>"MUST implement -or- MUST have reasons not to implement", and "MAY"
>means "MUST implement -or- MUST NOT implement".

Some people will argue that the RFC 2119 is made for software and 
should be used only when you have implementations interoperability 
issues.

At W3C, we do not only have specs which defines software 
implementation but also semantics or concepts. For example, do you 
think it has a sense to say:

	1. The "address" element MUST contain address information.
or
	2. The "address" contains address information.

And so how do you define the right usage of the tag in the case of a 
Test Suite.
Some people will say that the 1st case is an abuse of RFC 2119 as 
defined by IETF.

Conform?
<address>
	vegetables<br />
	milk<br />
	coffee<br />
</address>

Conform?
<address>
	Mr Semantic<br />
	HTML Road<br />
	Utopia<br />
</address>



-- 
Karl Dubost / W3C - Conformance Manager
           http://www.w3.org/QA/

      --- Be Strict To Be Cool! ---

Received on Thursday, 26 June 2003 10:57:41 UTC