Re: RDF Lists

On Fri, 2012-05-04 at 10:08 -0400, Manu Sporny wrote:
> On 05/04/2012 03:47 AM, Andy Seaborne wrote:
> > On 04/05/12 05:11, Manu Sporny wrote:
> >> In summary - RDF Lists are difficult to implement, even for people
> >> that know quite a bit about RDF. They are fantastically difficult
> >> to grasp for Web developers. They are really hard to author in many
> >> of the RDF syntaxes.
> >
> > but not Turtle :-) and your other messages suggests Turtle
> > everywhere.
> 
> Yes, TURTLE got it right. :)
> 
> >> I'd like to propose something that the group should seriously
> >> consider:
> >>
> >> 1. Add lists as a first-class citizen for all RDF serializations -
> >> deprecate all serializations that don't support lists as
> >> first-class citizens.
> >
> > This is the only complete solution -- anything that encodes in
> > triples means that the triples view will show through to developers.
> 
> Yes, to be more precise, an object can  now be:
> 
> * a plain literal (with optional language)
> * a typed literal
> * an IRI
> * a list (with optional type)

My proposal was that for RDF 1.1 we highlight the notion of a
"Well-Formed List" as a list that can be losslessly serialized in
Turtle, and suggest that Best Practice is to use Well-Formed Lists in
preference to all the other RDF list/collection/container mechanisms. 
This seems to me be a good transitional stage.

Personally, I think we should put all that other stuff on notice that it
might go away in RDF 2.0, but I'm not sure we can get consensus on that.

   -- Sandro

> > But IMHO making these changes as part of an incremental update of RDF
> > is not a good idea. RDF 2.0, or more realistically as part of a
> > planned migration from where we are today to where we want to be.
> > Simply replacing one approach with another one without looking at the
> > deployed base of software and published data is not a planned
> > migration.
> 
> I agree - just getting this very strong desire for change into the minds
> of this group.
> 
> >> 2. Get rid of the the Seq, Bag and List classes - replace with two
> >> datatypes - rdf:ordered and rdf:unordered. All "lists" in RDF are
> >> ordered by default.
> >
> > Personally, I don't see the need to have unordered as well. This
> > overlaps with the property definition of the property pointing to the
> >  list value.
> 
> An important detail that can be discussed after there is broad agreement
> that lists need to be first class citizens in RDF.
> 
> >> So, N-Triples and N-Quads could look something like this:
> >
> > Yes, NT and NQ will need list syntax.
> 
> In my follow-up post, I fold NT into NQ, and NQ into TURTLE Lite. There
> is only TURTLE Lite, and it supports the current list syntax (or a
> modified list syntax).
> 
> -- manu
>

Received on Saturday, 5 May 2012 00:25:54 UTC