Re: New version of FLD

On Mon, 06 Apr 2009 09:25:16 +0100
Dave Reynolds <der@hplb.hpl.hp.com> wrote:

> > > I see that both 
> > > schemas for, and calls to, external terms now have a non-optional 
> > > location.  You might want to make the location optional in both cases 
> > > otherwise DTB will need to be rewritten to update all its schemas and 
> > > the existing PRD and BLD syntax for Externals would need to change.
> > 
> > It is optional both in ebnf and in the plain English description (the
> > 1-argument
> > external is said to be a shortcut). Maybe I missed it somewhere. Can you
> > point to a specific place?  
> 
> For schemas I was looking at:
> 
> http://www.w3.org/2005/rules/wiki/FLD#Schemas_for_Externally_Defined_Terms
> 
> which gives the definition for a schema as including an id and I didn't 
> notice any optionality in there which is where I was looking for it.
> 
> However, now you point it out I can see a discussion of shortcuts in 
> http://www.w3.org/2005/rules/wiki/FLD#Terms end of clause 7. Sorry, I 
> must have been blind on first scanning that section.
> 
> So location is always required for schemas but optional in the syntax by 
> virtue of the shortcut, is that right?

Schemas are not part of the language (the actual strings of chars that one
sees), but are only used to define well-formedness and semantics. So,
it is not really necessary to define a shortcut for them.
However, you are right that I should have said how 1-argument externals are
related to these new schemas, which I now did. I also did some other cleanup
around there so things might look a bit better.

> I guess I can see the logic in that but the notion of schemas always 
> having a location will take a little getting used to.

This is not part of the language and has no effect on the user.
DTB needs a small update accordingly. Perhaps, in DTB it makes sense to
just say that we omit the first item in the schema:
e.g.,                 ( ?arg1; xs:double ( ?arg1 ) )
instead of ( xs:double; ?arg1; xs:double ( ?arg1 ) )
because it is always determined by the name of the built-in.


-- 
    -- michael

Received on Tuesday, 7 April 2009 06:57:01 UTC