Re: "Simple Lists" (was Re: ISSUE-77: Should we mark rdf:Seq as archaic (cf ISSUE-24))

On 17 Oct 2011, at 15:32, Sandro Hawke wrote:
> I'm not really comfortable with giving no guidance whatsoever about Seq
> and Lists.  My perception is there's general (if not unanimous)
> agreement that Lists are better than Seq, it's not great to have both,
> and it's best to use Lists in something like the Simple List form (ie
> bnodes, no loops, closed, etc). 
> Before this group was chartered, when it wasn't clear it would be, I
> proposed we simply do a survey of the experts and practitioners on
> questions like this, and make the results public -- then new folks could
> judge for themselves what features were actually being used, and why,
> and by whom.    But I think, given we have a WG, we can do a bit better
> than that.

I think it would be a good idea if this WG published a Note as well that talks about what we've changed and why. It should give use cases for every change. This was mentioned at the F2F.

Some new features will be added to the RDF package in order to address shortcomings of previous features. This will require explanation. For example, why add multigraphs when we already have reification? Why add Turtle when we already have RDF/XML and RDFa and N-Triples? Why add Skolemization when we already have blank nodes and URIs? The Note will probably need some text that contrasts each pair and highlights what each one is good for.

Such a Note might be a good place to also give guidance in other areas where RDF offers multiple features that seem to do the same thing. Like when to choose rdf:Seq and when to choose rdf:List?

The normative specs can contain links to that note.


Received on Monday, 17 October 2011 15:16:27 UTC