Re: the state of ldp-patch, and a procedural proposal

On 10/18/2013 02:54 PM, Alexandre Bertails wrote:
>
> I see emerging several trends:
>
> * 2 kind of syntaxes: SPARQL Update subset or RDF PATCH derivatives
>

I think there are three kinds of syntaxes: SPARQL Update subset, Dataset 
syntaxes, and Custom Patch syntaxes.

> * skolemizing bnodes VS matching bnodes
>

Four options here: match blank nodes, skolemize blank nodes, namable 
blank nodes (as in RDF Patch and GRUF), and dont-patch blank nodes.

> * expressive power, ie. how powerful and complex the BGP can be
>

Two separate issues here: power of the graph pattern for blank node 
matching, and power of the update language for expressing changes (of 
which list-update-operations is one kind).

> * support of specific features, eg. rdf:list
>
> Maybe we can have polls to make progress? I believe that the very
> first question to answer is about skolemization.

Yes, coming out of the last F2F, my sense was the first choice was 
between matching blank nodes and namable blank nodes, and I proposed we 
do both through CR, assuming people were willing to do the work.

It's on today's agenda: 
http://lists.w3.org/Archives/Public/public-ldp-wg/2013Sep/0041.html (and 
of course it's the start of this thread, going back 5 weeks).

I'm slightly less enamored of the work load at the moment, as I find 
myself spread very thin.    If I had to pick a solution today, I'd say 
subset-of-SPARQL-update, with node matching, and I'd try to figure out 
some way to extend that with some list operations primitives that we'd 
want in SPARQL 1.2 any way.

       -- Sandro

Received on Monday, 21 October 2013 13:14:31 UTC