Re: review comments on rdf-syntax-grammar (version of 25 Mar)

On Sun, 24 Mar 2002, Dave Beckett wrote:

> >>>Graham Klyne said:
> > At 07:11 AM 3/24/02 -0500, Dan Brickley wrote:
> > >I've been thinking about strategies for dealing with 'unserializable'
> > >graphs. These are (AFAIK) all (or nearly all?? would be good to be clear)
> > >to do with having the edges in some RDF graph be labelled with a URIref
> > >that doesn't split conveniently into  namespace name and local name.
> >
> > Other such graphs are literals in subject position,
>
> Not allowed at present in the model^Wgraph.

I must have missed discussion/decision on this, presumably: reading the
Syntax doc (and comments such as those above, which indicate
that the Syntax is a `syntax for the graph per MT spec not M+S'99)...

Why is the MT spec not a normative reference, while the (broken old) M+S
spec is normative? Should implementors code to M+S or new MT's notion of the
graph? Presumably the latter, since we talk of bNodes etc.

Is the lack of a normative ref caution stemming from the fact that the MT
includes stuff like RDFS closure rules, and we don't want to give
impresssion that syntax can't be implemented without that stuff? (eg.
see our exchange earlier re my idea for using subPropertyOf to serialise
XML-unfriendly predicates). It should somehow be possible to make
normative ref from syntax to MT without parsers having to have inference
engines, shouldn't it? (for some sense of 'normative reference').

Dan

ps. typo in abstract: specifyed

also, the abstract references (by direct hyperlink not footnote) MT
version of sept 2001 while the footnotes reference the valentine's edition.

Received on Sunday, 24 March 2002 15:24:22 UTC