SPARQL - A Query Language for RDF from the latest DAWG minutes. Even has a Simpsons reference and possible logo, Mr Sparkle.
Kendall Clark has some ideas about whether FROM is necessary in "Re: [Fwd: FROM keyword unnecessary?]". I wish I had time to consider this more fully.
2 comments:
You'd make a great WG member! Why?
1. Yr too overworked to read DAWG mail -- I've decided I'm write-only from here on out. :>
2. You, too, can mis(spell|type) my first name, which has become something of a shibboleth for our little band. It's been "Kendal" many times, and I think "Kendel", but not "Kendell" -- congrats on the novelty. :>
And, fwiw, we can do without FROM in Sparql but I'm not sure we want to. In the case where there is no network or protocol, rather just executing Sparql queries in a script, you need some way to do target graph selection, or so Andy Seaborne has argued. Though it just occurs to me that in that case it would make sense to pass graph identifiers to the program as command line options or in a config file.
I would like to get rid of it, but another possibility is to keep it in the syntax to be kind to users, but require Sparql clients to sniff the query and put target graphs into the protocol, which is the only thing the server would look at. In which case the server would get queries with FROM clauses, but it would basically ignore them.
Heh, I should have sent this to the WG. :>
Thanks,
Kendall
Sorry about that - I fixed the spelling of your name.
As long as there's a use case I guess that dictates a need - I'd like to see count, having and order by as well.
I think you need to name graphs - irrespective of protocol or network. I don't think it's kind to users, to tie the name of the graph to either though. Users shouldn't have to rewrite queries just because the way you've connected to the database has changed. The name resolves to a protocol and host if required - either through configuration, some lookup service or something else.
Post a Comment