Re: OWL Abstract Syntax

On Fri, 2002-10-04 at 11:12, Peter F. Patel-Schneider wrote:
> 
> Sean raises a number of points here that are somethat more than editorial
> changes, but are not big changes.
> 
> I have made changes to the abstract syntax document to address his
> concerns.  The changes amount to:
> 1/ Using keyword(...) uniformly throughout the document.
> 2/ A few other minor changes.
> 
> Parsing the revised OWL Abstract Syntax does require a bit of lookahead,

???

Isn't that a contradiction in terms?

I thought with an abstract syntax, you just got handed parse trees.
There is no parsing.

> but it does fit within yacc,

Now you're talking about implementing the abstract syntax,
and you're treating comments on it as "more than editorial".

This is taking on the character of a presentation syntax,
not an abstract syntax.

I'm concerned that this distracts readers from understanding
the actual syntax of OWL, which is RDF/XML syntax.

"RESOLUTION: The meaning of an OWL document is conveyed in
the RDF graph"
"RESOLUTION: The exchange language for OWL is RDF/XML "
	-- minutes Apr 2002 ftf
	http://www.w3.org/2001/sw/WebOnt/ftf2.html

I note that the WG hasn't taken any decisions about the
role of the abstract syntax in our design (except to
publish a WD to get feedback). From the feedback we're
getting, I'm opposed to the continued use of the
abstract syntax in our design work.


> except for restrictions where it is difficult
> to distinguish between restrictions on datatype and individual properties.
> (I actually have a yacc grammar for OWL now that finesses this.)
> 
> peter
> 
> 
> 
> From: Sean Bechhofer <seanb@cs.man.ac.uk>
> Subject: OWL Abstract Syntax
> Date: Thu, 3 Oct 2002 11:08:38 +0100 (GMT Daylight Time)
> 
> > Some brief comments on the latest version of the OWL Abstract Syntax
> > document (prompted by trying to use the abstract syntax in a tutorial
> > style document).
> > 
> > o The mixed use of "feature=X" and "feature(X)" forms is ugly and
> >   potentially confusing -- e.g. cardinalities use "()" and someValuesFrom
> >   uses "=". Why not stick to one or the other?
> > 
> > o There are places where parsing could be unnecessarily tricky, i.e.,
> >   something like "union ( person restriction (...))" where restriction
> >   could actually be a class called restriction or the start of some restriction
> >   syntax.
> > 
> > Tidying up the syntax a little and adopting some consistent conventions
> > (e.g. keywords always followed by parentheses) should alleviate both these
> > problems I think.
> > 
> > Cheers,
> > 
> > 	Sean
> > 
> > -- 
> > Sean Bechhofer
> > seanb@cs.man.ac.uk
> > http://www.cs.man.ac.uk/~seanb
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 4 October 2002 12:27:33 UTC