Re: keep conformance objective (detailed review of section 1. Introduction)

Philip TAYLOR (26 sept. 2007 - 17:43) :
> That is a serious bone of contention which I believe
> must be addressed before the specification can be
> written.  But my idea of using two words instead of
> variants of one is that we are trying to communicate
> two quite different ideas : what makes a document
> (syntactically) valid, and what extra steps are needed
> before a valid document may also legitimately
> claim to conform to a formal-but-not-machine-verifiable
> specification.

The more the discussion goes on, and the more the terms get blurred.
Validation and Conformance checkings are the same depending on what  
you define for those terms.

* Validation is often associated for Web technologies with checking  
the syntactic constructs of a document with regards to a formal  
schema language.
	Note: This leaves the door open to a very wide spectrum of  
possibilities. The schema language could be an SGML DTD, an XML DTD,  
an RDF schema, a relax NG, etc. It could be whatever as long as  
there's a formal grammar.

* Conformance checking is often associated for Web technologies with  
checking the requirements defined by the spec.
	Note: A specification could have one only basic requirement of being  
valid with regards to a defined grammar. Then validation and  
conformance checkings would be exactly the same.
	A conformance section in a specification is just  a tool for  
different class of products (authors, authoring tools, other specs,  
user agents, etc.) to give guidance on how the technology must be  
implemented and how to achieve a basic level of interoperability.


Prose somehow could be defined as a schema language, if there was a  
regular way of defining the things in the spec, which is true for  
part of the spec. For example, look at the title section or any kind  
of elements definition.

http://www.w3.org/html/wg/html5/#the-title0
3.7.2. The title element
Metadata element.
Contexts in which this element may be used:
     In a head element containing no other title elements.
Content model:
     Text (for details, see prose).
Element-specific attributes:
     None.
DOM interface:
     No difference from HTMLElement.

  Ian Hickson uses a regular pattern to define some of the elements  
of the spec.  A program with regular expressions could extract the  
rules for each elements and that would define a minimal grammar,  
without being an XML or RNG or SGML grammar, but still a grammar.

Yes it would be a lot easier to have an already defined schema  
grammar. It is often said that all these grammars are not expressive  
enough for the purpose of HTML conformance, and that in the end, we  
still have to implement a conformance checker by reading the  
specification prose. I have the feeling that an RDF schema could  
express almost all requirements, but that would be only a small part  
of the job. It would then require a parser to analyze these  
conformance requirements, and then a conformance checker applying  
them against a document.

It would be beneficial only if more than one products were using this  
engine, but if people ends up implementing each time a different  
engine then it would be not be very useful. We have to find the right  
balance, there is not only one answer to each issue.


-- 
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager, QA Activity Lead
   QA Weblog - http://www.w3.org/QA/
      *** Be Strict To Be Cool ***

Received on Thursday, 27 September 2007 00:54:53 UTC