Re: dataset protocol & service description

On Mon, Mar 14, 2011 at 2:03 AM, Lee Feigenbaum <lee@thefigtrees.net> wrote:
> In the wake of our discussion last week, the lack of subsequent discussion,
> and our time and experience constraints, I plan to propose on Tuesday that
> the SPARQL 1.1 service description cover the SPARQL Protocol _only_, and
> that it not discuss graph stores or the dataset protocol at all.
> Specifically:

> PROPOSED: The SPARQL 1.1 Service Description describes endpoints that
> implement the SPARQL 1.1 Protocol; it does not describe URIs that respond to
> the SPARQL 1.1 RDF Dataset HTTP Protocol.

+1 (in absentia)

Note, however that this will beg the question about how Graph Store
URLs are discovered or known apriori in order to POST to them (per the
dataset protocol)

> Please consider this email a "last call" on this topic -- if you have an
> alternative proposal, now would be a good time to share it.
>
> Even if we do resolve this, we probably need to clarify the relationship
> between Graph Stores and Datasets more, as Chime pointed out last week.

(this is not meant to be an alternative proposal), however I tend to
think of a graph store and dataset as differing only in that one is
comprised of mathematical sets of graph edges and the other is a
mutable resource that can be updated over the web or directly via the
Update language.  Disregarding mutability, there is an isomorphic
correspondence between a graph store and a dataset in the sense that
we want to be able to manipulate a graph store and have a SPARQL
protocol instance managing a dataset comprised of the same RDF
'content' to respond with new answers to queries, for instance.

-- Chime

Received on Tuesday, 15 March 2011 12:20:47 UTC