W3C home > Mailing lists > Public > www-rdf-interest@w3.org > May 2002

Re: Query and storage

From: Dave Reynolds <der@hplb.hpl.hp.com>
Date: Fri, 24 May 2002 14:27:04 +0100
Message-ID: <3CEE3FA8.E51FEB28@hplb.hpl.hp.com>
To: "Thomas B. Passin" <tpassin@comcast.net>
CC: www-rdf-interest@w3.org
[Thomas B. Passin]
> I think this is a matter of what your query asks for.  Your example is not
> well posed.  In sql you would proabably do it with three queries that get
> UNIONed together.  You don't really want one tuple where c,a, and r all
> appear in the same tuple.  You may in fact only want to get back ?X,
> depending on your needs.  If you pose the question correctly, the
> combinatorial explosion goes away.

Agreed. If the query language supported UNION then that would handle it.
Alternatively I can simply issue three queries and union the results together in
the application, so long as I can batch up queries and responses to avoid the
three network trips. 

> This scenario could provide good guidance in the design of the query
> language, but I don't see it as having anything to do with variable binding
> vs subgraph results.

Fair comment. 

Dan's post, which I was originally responding to, was suggesting  "a simple
'graph match, return the bindings' query protocol". I was just pointing out that
that combination is an imperfect match to a simple use case that affects my
work. Enriching the query language would be one way to address this - though
perhaps SQL with UNIONs is less simple than Dan had in mind.
Alternatively, as Andy points out, making a clearer separation between query
language, results format, a over-the wire protocol would address it.

Dave
Received on Friday, 24 May 2002 09:28:21 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:54 GMT