W3C home > Mailing lists > Public > public-html@w3.org > April 2007

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

From: Henri Sivonen <hsivonen@iki.fi>
Date: Sun, 15 Apr 2007 16:00:08 +0300
Message-Id: <94A15360-920E-4BEF-A107-AC2FB7E7EE33@iki.fi>
Cc: public-html@w3.org
To: Henrik Dvergsdal <henrik.dvergsdal@hibo.no>

On Apr 15, 2007, at 14:50, Henrik Dvergsdal wrote:

> Is it possible to specify all  machine-checkable criteria in  
> XHTML5  by means of:
> A RELAX NG schema which refers to
> 1. A datatype library
> 2. A set of formal rules, specifying, for instance, table integrity  
> requirements

A RELAX NG schema cannot refer to a generic "set of formal rules". I  
take it that you meant to ask if XHTML5 conformance can be defined as  
an XML document satisfying a RELAX NG schema and a "set of formal  
rules" (without the schema actually doing the referring).

In theory, yes, but the "set of formal rules" would just be a catch- 
all notion for everything that the RELAX NG schema doesn't cover. To  
implement a conformance checker, you eventually need to have running  
code that checks for all the stuff that the RELAX NG schema didn't  
cover. With the current WHATWG spec model, a human (me) takes English  
as the input and produces a computer program as the output. You seem  
to be suggesting adding an intermediate formalism.

If an intermediate formalism was added, the interesting question  
would be how to get from the formalism to running code. Would the  
formalism be a computer-readable new schema language, so that one  
would have to implement a validation engine for it? Or would a human  
software developer still read the formal stuff as the input and  
produce a program as the output?

If you meant the former, you'd end up inventing a declarative  
programming language as the formalism and have the spec editor  
develop a conformance checker in that programming language. (Or  
alternatively, you'd end up with an XHTML5-specific schema language  
and the purported validator for the schema language would essentially  
contain the kind of code that I have implemented.) If you meant the  
latter, I don't see how the added level of indirection between  
English and running code would help.

Henri Sivonen
Received on Sunday, 15 April 2007 13:00:27 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:18 UTC