- From: Rob Shearer <Rob.Shearer@networkinference.com>
- Date: Tue, 6 Apr 2004 12:17:29 -0700
- To: <public-rdf-dawg@w3.org>
The query language must be defined independently of any protocols with which it is used. Notably, it must not enforce the use of a particular network protocol in order to answer queries. I recognize that this proposed requirement is on the periphery of a rather contentious issue that we have wisely chosen to avoid for the time being (the network protocol aspect of the charter), but I think it's quite an important decision to make. I see the primary use case model dictating this requirement being a system that simply has a local RDF repository, either on disk or in memory. It should be possible to write an application which accepts queries in the language we define (i.e. using a text field in a GUI or a text file as input to a command-line tool) and produces results in an implementation-defined manner which may have absolutely nothing to do with accessing a network--it can simply analyze its own local RDF store. Such architectures might also allow our query language to function as an (admittedly potentially heavyweight) analog of XPath for XML: a way of retrieving particular bits of RDF from local stores, making RDF a suitable choice for configuration settings and the like. Such a requirement certainly doesn't dictate that network protocols can't influence language design; only that one must be able to implement the language in environment which include absolutely no networking support of any kind.
Received on Tuesday, 6 April 2004 15:18:31 UTC