- From: Graham Klyne <gk@ninebynine.org>
- Date: Fri, 04 Jun 2004 11:48:31 +0100
- To: public-rdf-dawg-comments@w3.org
I like the way this document is presented. Some thoughts about some of the pending issues. I agree with all the points marked "ACCEPTED" (though I'd suggest that the level of XSD Datatype testing support required of any implementation be kept as small as reasonably possible). ... [[ 3.6 Optional Match It must be possible to express a query that does not fail when some specified part of the query fails to match. Any such triples matched by this optional part, or variable bindings caused by this optional part, can be returned in the results, if requested. Status: Pending. 3.6 Optional Match (variant) It must be possible to express a query with optional parts such that the query does not fail to match when one or more optional parts of the query fails to match. Any such triples matched by this optional part, or variable bindings caused by this optional part, can be returned in the results, if requested. Status: Pending. ]] In one of my own query implementations, this is a feature that I have found to be very useful. I have an application that uses query to drive the generation of reports and non-RDF data from RDF input, and I found the introduction of optional (and alternative) query sections made the system much easier to use. I return results as variable bindings, and the optional sections (may) simply result in additional variables being bound. The calling program makes considerable use of the existence or not of a variable binding in response to a given query. ... [[ 3.8 Bookmarkable Queries It must be possible to express a query as a URI. The result of dereferencing the resource identified by this URI will be a representation of the query results. This form of the query is not assumed to be humanly readable. This requirement does not preclude other mechanisms for issuing queries. Status: Pending. ]] It seems to me that this is an orthogonal issue. Given a protocol it should be possible to construct a URI to represent it. (BTW, a few months ago, I had some experience with long URIs -- over 4K. or maybe 8K -- causing a popular web browser to fall over.) ... [[ 3.10 Result Limits It must be possible to to specify an upper bound on the number of query results returned. Status: Pending. 3.10a Iterative Query (variant) It must be possible to handle large result sets of any size by iterating over the result set and fetching it in chunks. Status: Pending. ]] Maybe this should be an application-specific issue? I think it might be useful to have a common way to indicate that not all possible results were returned (which I think is a smaller problem). If you've going to take partial results seriously, then it may also be a requirement to indicate how the results should be ordered? My feeling is that there's a can-of-worms here, and that providing suitable hooks for extensibility may be better than trying to devise a solution to these issues. ... [[ 4.1 Human-friendly Syntax There must be a text-based form of the query language which can be read and written by users of the language. Status: Pending. ]] In practical terms, I think this will be important. Less clear is whether it needs to be standardized. Maybe start with a NOTE covering a proposed human-friendly format, which can maybe later advance to REC if it's sufficiently useful? ... [[ 4.2 Provenance It should be possible for query results to include source or provenance information. Status: Pending. ]] Could this be covered by extensibility hooks? ... [[ 4.3 Non-existent Triples It should be possible to query for the non-existence of one or more triples or triple patterns. Status: Pending. ]] Isn't this just a variation on result limits? ... [[ 4.4 User-specifiable Serialization It should be possible to specify the serialization format of query results; this design objective is meant to be orthogonal to the semantics of query results, whether subgraph, variable bindings, or some other type. Status: Pending. ]] This looks like a possibility for unhelpful scope creep. I'm not sure about variable binding and subgraph results -- I understand there may be significant semantic difference but I'm not sure what -- but I don't think syntaxes should be proliferated. If the data model is clear, private applications may choose to use a different syntax locally (like many people using N3). ... [[ 4.5 Aggregate Query It should be possible to specify two or more RDF graphs against which a query shall be executed; that is, the result of an aggregate query is the merge of the results of executing the query on each of two or more graphs. Status: Pending. 4.6 Additional Semantic Information It should be possible for knowledge encoded in other semantic languages—for example: RDFS, OWL, and SWRL—to affect the results of queries executed against RDF graphs. 4.6a Additional Semantic Information (variant) It should be possible for a query to indicate that the answers should take into account knowledge encoded in RDF semantic extensions such as RDFS, OWL, etc. ]] These look like issues that should be addressed orthogonally from the basic query+access. Essentially, it seems to come down to specifying the graph that is being queried, which I'd expect to be just a URI. If standard aggregation mechanisms are required later, then I could imagine standard forms of URI construction might be introduced. ... #g ------------ Graham Klyne For email: http://www.ninebynine.org/#Contact
Received on Friday, 4 June 2004 06:48:25 UTC