- From: Seaborne, Andy <andy.seaborne@hp.com>
- Date: Thu, 13 May 2004 11:25:21 +0100
- To: "'Rob Shearer'" <Rob.Shearer@networkinference.com>, "'RDF Data Access Working Group'" <public-rdf-dawg@w3.org>
-------- Original Message -------- > From: public-rdf-dawg-request@w3.org <> > Date: 11 May 2004 18:56 > > I think requirement 3.8 brings up major questions about just > what we consider a "query". Does a "query" include a > selection of the RDF source, or can the same query be > executed against different sources? (The latter seems much > more sensible to me...) +1 : I don't see why the query should need to record the execution but I can be persuaded by an example. At the moment, I see it as a protocol matter concerned more with selecting the service access point. A "query" is one of Pat's 'familiar abuse of terminology'. The term is used for both a recipe (a sequence of symbols in the query language) for executing requests and the request (execution instance). A template can be partially instantiated (the source set, still can be repeated used). This multiple use is widespread and so our documents will be read with both meansings in mind - it isn't something a glossary can help with much. C.f. "program" - the file continaing the binary or the one-time execution. Its this overloading that causes the "bookmarkable query" to be a bit confusing. It isn't a term invented in this WG; it is a general web arch thing. > > It makes a lot of sense to make clear that a query should be > encodable as text, and that requirement implicitly means that > a query can be encoded *within* a URI, but that's a long way > from saying that a query > *is* a URI. The requirement as written conjures notions of > new schemes and URLs and protocol dependence and all the rest. > > At the very very least, I'd suggest changing "...a query as a > URI." to "...a query within a URI." > > I'm still not very happy with this requirement, because the > whole URI issue just seems a side point. I don't think we > need to standardize the format for a query URI in order to > declare victory, What I think this requirement really says is that the protocol must be able to be used over HTTP GET as "bookmarkable". Query is nromally a safe operation and web caching also works. The requirement says very little about the query language as x-www-form-urlencoded can be applied, if there is a concrete syntax, it can be put in a URL. The charter says "HTTP and/or SOAP-based protocol" so this makes the HTTP part a requirement. > and if we don't do that then I'm not sure > we'll have grounds for deciding whether we've met the requirement. > > I'd go back to the very simple "It must be possible to > express a query as unicode text." (Better ideas than > 'unicode'?) This actually subsumes the requirement that > queries be able to be embedded in URIs, and yet I think we've > already agreed on it, so it seems like a no-brainer.
Received on Thursday, 13 May 2004 06:29:30 UTC