- From: Steve Harris <steve.harris@garlik.com>
- Date: Tue, 29 Sep 2009 16:55:40 +0100
- To: Lee Feigenbaum <lee@thefigtrees.net>
- Cc: "Seaborne, Andy" <andy.seaborne@hp.com>, SPARQL Working Group <public-rdf-dawg@w3.org>
On 29 Sep 2009, at 16:34, Lee Feigenbaum wrote: > Seaborne, Andy wrote: >> This is a request for your views on starting points for the >> property paths time permitting feature. Please send +1/0/-1 for >> the options (which aren't meant to be mutually exclusive) or we >> might get telcon time. >> I've tried to list the main possibilities in terms of styles and >> approach as starting points: all have variations and areas of >> uncertainty (e.g. ordering of results). >> 1/ Property paths only mention IRIs or prefixed names. >> The most conservative choice. Still need to relate to entailment. >> 2/ Property paths with variables and IRIs or prefixed names. >> (issues include restriction of what can be asked a la ?p* discussion) >> 3/ With access to the length of the path matched >> Issues include how multiple paths between two nodes are handled >> (two lengths possible). >> 4/ With access to the path matched (path-valued variables is one >> possibility) >> Issues as 3 + what is a path value "datatype". >> For all of them: add an option to have >> 5/ A mechanism that will allow a variety of path matching schemes, >> and provide one such system. >> Roughly, this would involve defining syntax so various different >> approaches can at least use common syntax but choose from 1-4 as to >> what the WG describes in this round of standardization and show the >> relationship to the syntax. E.g having a PATH keyword idea in [1]. >> Then >> 6/ Do nothing in this round - too early to standardise. >> Andy > > I'm +1 on #1 and -1 on everything else: I'm afraid of the complexity > of everything else, and think that doing #1 enables a whole range of > uses. It also lays the (syntactical) groundwork for #2, at least, > and I don't think it precludes systems that want to do something > else for #3 and #4. On reflection so am I, I was 0 on 2/, but I think it opens the door to queries with immense complexity. e.g. just SELECT * WHERE { ?a ?p* ?b . } I pretty nasty, and will require a very large working set to avoid loops. - Steve -- Steve Harris Garlik Limited, 2 Sheen Road, Richmond, TW9 1AE, UK +44(0)20 8973 2465 http://www.garlik.com/ Registered in England and Wales 535 7233 VAT # 849 0517 11 Registered office: Thames House, Portsmouth Road, Esher, Surrey, KT10 9AD
Received on Tuesday, 29 September 2009 16:06:34 UTC