- From: Jan Grant <Jan.Grant@bristol.ac.uk>
- Date: Fri, 5 Oct 2001 16:45:27 +0100 (BST)
- To: Pat Hayes <phayes@ai.uwf.edu>
- cc: w3c-rdfcore-wg <w3c-rdfcore-wg@w3.org>
On Fri, 5 Oct 2001, Pat Hayes wrote: > >On Thu, 4 Oct 2001, Aaron Swartz wrote: > > > >> On Thursday, October 4, 2001, at 09:34 AM, Jeremy Carroll wrote: > >> > >> > I would prefer test1.nt to only have one line if they are > >> > identical. The > >> > graph is a set. > >> > A comment explaining the deletion would then be helpful. > >> > >> Whoa, whoa, whoa, I don't think we ever agreed to this. It was > >> my understanding the output was a bag (there wasn't harm in > >> doing so), but could be interpreted as a set. > >> > >> When was this changed? > > > >The MT would give the same interpretation for equivalent arcs in a > >multigraph, wouldn't it? > > Yes, in fact RDF graphs are multigraphs. However, that means that the > MT works just as well if they are sets or bag. Bags is semantically > harmless and puts a lighter burden in implementers, so I would vote > for bag. Sets are semantically harmless too and put a lighter burden on me as an implementer :-) , so I'd rather do the following: - treat rdf storage systems as sets (maybe, unless they do something groovy with provenance) - warn consumers of TripleIterator streams that they may see the same statement more than once - and so on. In other words, the behaviour of an application over time as assertions and retractions are made is a decision that the application-writer has to make and advertise to his users. jan -- jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/ Tel +44(0)117 9287088 Fax +44 (0)117 9287112 RFC822 jan.grant@bris.ac.uk Ceci n'est pas une pipe |
Received on Friday, 5 October 2001 11:47:53 UTC