No normative schema (was: Re: toward an issues list/agenda for the HTML WG)

On Mar 22, 2007, at 23:22, Dan Connolly wrote:

> OK, so the IRC office hours experiment went a little
> better than the email brainstorming experiment.
> I'm keeping an issues list/agenda; it covers
> some of the highlights from IRC as well as recent email,
> and has pointers to the full log of the IRC chat.
>    http://www.w3.org/html/wg/il16

I'd like to add a policy suggestion item:
The WG should not designate normative schemata.

Rationale:
Schema languages cannot express all the conformance requirements that  
the WG's spec can be expected to have (see http://hsivonen.iki.fi/ 
thesis/html5-conformance-checker.xhtml#non-schema and
http://hsivonen.iki.fi/thesis/html5-conformance-checker.xhtml#more- 
non-schema for discussion). The experience with HTML 4 suggests that  
designating a normative schema causes people to use it and to ignore  
the machine-checkable conformance criteria that the schema does not  
embody. To avoid that kind of situation, it would be better not to  
designate a normative schema and instead consider schemata  
implementation details just like one would consider particular lines  
of C++ implementation details.

As precedent, the Atom WG of the IETF did publish a schema, but it  
was informative. In addition, not endorsing schemata as normative is  
the WHATWG's current policy. There is a non-normative schema project  
for HTML5.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Friday, 23 March 2007 09:29:05 UTC