- From: Jeen Broekstra <jeen@aduna.biz>
- Date: Tue, 11 Nov 2003 11:52:00 +0100
- To: Dan Brickley <danbri@w3.org>
- Cc: www-rdf-rules@w3.org, eric@w3.org, em@w3.org, sandro@w3.org, massimo@w3.org, connolly@w3.org
Some quick points of feedback concerning the RDF Query charter draft. First of all: the charter seems very open, which I like very much. Considering the wide variety of proposals out there, I'm happy to see that none are being dismissed beforehand. The big disadvantage is that I have very little to comment on :) I'll give it a go nevertheless: I find it somewhat disappointing that RDFS semantics are completely out of scope. I know it's supposed to be baby steps, but in my opinion things like class- and property subsumption are now sufficiently well-described (and there is enough experience in implementing them) to enable support for them in a QL. Moreover, not adding this in the QL formal model, but allowing for different implementations to add it 'on their own' (which seems to be what the draft charter suggests in section 2.1 [1]), will result in different implementations giving very different answers to the same query. I think that this is undesirable, but YMMV. I'm also missing something about result set format and syntax. U think that creating a norm for these is at least as (if not more) important for interoperability. From the perspective of compositionality I would prefer a query language design that returns subgraph matches (or transformed graphs) instead of/next to variable bindings. But I'm running ahead of the charter here. Jeen [1] As an editorial comment: section 2.1 was clear as mud to me, to be honest. For example, what does the phrase "The [WG] may find it trivial to produce a query mechanism that illustrates the semantics of terms in RDFS or OWL" mean? Trivial? -- jeen@aduna.biz Aduna (formerly Aidministrator b.v.) - http://www.aduna.biz Julianaplein 14b, 3817 CS Amersfoort, The Netherlands tel. +31-(0)33-4659987, fax. +31-(0)33-4659987
Received on Tuesday, 11 November 2003 05:50:55 UTC