Re: Formal definition of HTML5 (was Re: Version information)

Le 17 avr. 2007 à 17:04, Ian Hickson a écrit :
> HTML4 has an official schema, and in the YEARS of HTML4 being the  
> latest
> and greatest, no serious usable conformance checker was made that used
> more than the official schema.

The fact that no conformance checker was built on top a validator  
doesn't mean it was impossible to do. So to frame exactly the history  
of the W3C Markup validator.

It has been built as a validator and only a validator. It means a  
tool that will help someone to fix some constraints of the document  
with regards to the grammar (a DTD) used to write it.

The W3C Markup validator lives by its community too. There was a  
strong push for a while that it stays only a DTD validator, because  
it was easier to maintain. The community has pushed for more helpful  
messages, more help, etc. Then the validator has evolved with the  
dedicated time of Olivier and the valuable participation of the  
community (too limited sometimes.)
For the last update see
http://www.w3.org/QA/2007/04/validator_080_upgrade_notes.html

Henri Sivonen has created a very interesting work (still incomplete,  
as he said it) for his Master Thesis by creating a conformance checker.

I think there are at least 3 levels.

1. Requirements defined by the formal language used (within its own  
limitation)
2. Requirements defined by the english prose used (within its own  
limitation)
   and can be automatically verified
3. Requirements that can be checked only by humans
   appropriate content of an "alt" attribute for example

Nothing is perfect, but each of them has benefits in some circumstances.
Having a formal language for example, can speed up and make it easier  
for tools developers. Not having to develop their own tools. It means  
it lowers the barrier of entry for participating.

The fact to have a formal language or not and the way you define your  
conformance section in a specification is completely orthogonal.

Related to this thread:
Use formal languages when possible
http://www.w3.org/TR/qaframe-spec/#formal-language-gp


-- 
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 Wednesday, 18 April 2007 02:50:18 UTC