RE: Scope

Geoff,

I think you have touched on a significant point about the query abilities of
different systems.  I was thinking that defining a level of operations
around query, not assert or retract, is the more useful.

To me, it is better to have something simple "soon" because it helps
application developers.  One way of starting might be to get some short use
cases.  This helps us all get on the same page.

I'm with you on the interchange format, rather than a single syntax.  I can
see that different syntaxes for the same language can be useful for
different application domains.  An RDF/XML syntax might not be the most
readable :-) but is good for a SOAP service.

	Andy

-----Original Message-----
From: Geoff Chappell [mailto:geoff@sover.net]
Sent: 14 November 2001 15:24
To: Dan Brickley; Seaborne, Andy
Cc: Libby Miller (E-mail); www-rdf-rules@w3.org
Subject: Re: Scope


I'm all for it.  My only concern is that we recognize that systems of
varying capabilities might be answering the queries and we allow for it or
explicitly limit the scope of the language. Possibly this just means having
explicit vocab (for example, so 'not' never means different things to
different systems we use 'not' and 'not_known'). A system could support a
particular vocabulary item or not.

What are the likely features of the query language? which of these should it
include?
1. query
2. assert/insert
3. retract/delete
4. functions
5. logical ops (and, or, not, not_known, exists, known)
6. aggregation functions
7. rules
8. variable quantification
9. ...

How about the features of the results language?
1. results as tabular variable bindings
2. results as triples/rdf
3. other results - i.e  messages, warning, errors

Is the expectation that implementations would alter their native syntax or
that we'd define a common, easily parsable syntax that we could all map
into? the later might be easier - especially if it means we could go light
on the human useability concerns (i.e. use an ugly syntax like
and(and(or({}... )

Geoff



----- Original Message -----
From: "Dan Brickley" <danbri@w3.org>
To: "Seaborne, Andy" <Andy_Seaborne@hplb.hpl.hp.com>
Cc: "'Geoff Chappell'" <geoff@sover.net>; "Libby Miller (E-mail)"
<Libby.Miller@bristol.ac.uk>; <www-rdf-rules@w3.org>
Sent: Wednesday, November 14, 2001 8:57 AM
Subject: RE: Scope


>
> On Wed, 14 Nov 2001, Seaborne, Andy wrote:
>
> > Geoff,
> >
> > The style I would suggest is one of many small steps - trying to get a
> > regular flow of useful output, both so that everyone is informed and to
let
> > implementers try out the ideas.  We could start with requirements
gathering
> > and use case examples.  This would also draw on previous and related
work.
> >
> > As to the scope of the work, it is what the people involved make of it.
I
> > would like to see a framework in which query capabilities fit - not a
single
> > set of funconality that every system must implement to be conformant.
[...]
>
> Yes, we're definitely in "what we make of it" territory. Rather than
> launch a big RDF Query (and/or Rules) formal Working Group, we've decided
> to take a different approach for the time being. The www-rdf-rules list is
> here so that implementors can find out more about each other's systems,
> specifications, working assumptions... I've a hunch that a very limited,
> bare bones RDF query language could form the basis for some early interop
> testing amongst existing systems. One lesson I'd like to learn from this
> is how the limitations of such languages relate to practical use cases,
> and where the complexity/shippability tradeoffs lie. It'll certainly be
> more fun exploring this in the context of an Interest Group than in a
> Working Group, I think. Once we've got a better sense for the options
> available, I'd like to start discussions that are more explicitly about
the
> chartering of a proper Working Group in this area...
>
> Dan
>
>
> RDF Interest Group chair
>
>
>
> --
> mailto:danbri@w3.org
> http://www.w3.org/People/DanBri/

Received on Monday, 19 November 2001 06:58:55 UTC