There is also the fact that the protocol states that any default-graph-uri and named-graph-uri parameters included in a request require that the endpoint use the dataset described by those in preference to any described by the query
So it is possible to write a query that specifies a dataset with FROM and FROM NAMED but execute it such that they are ignored in favour of another dataset Rob On 06/02/2015 03:55, "Daniel Hernández" <[email protected]> wrote: >Andy, thanks for your answer. It solve my question. > >Milorad, I'm not sure if I understand your assertion "default graph of a >dataset shouldn't depend on a query", as the default graph can be >described in a query using named graphs. Also, I does not understand >your assertion that a default graph does not depends on an edpoint, >because the SPARQL specification stated that when no dataset description >is provided in the query or in the request parameters (specified in the >SPARQL protocol), then the dataset is provided by the endpoint, i.e., a >default dataset is used. > >Regards, >Daniel > >On Fri, 2015-02-06 at 09:46 +0000, Andy Seaborne wrote: >> On 06/02/15 06:45, Milorad Tosic wrote: >> > Hi Daniel, >> > As you correctly noted, default graph of a dataset shouldn't depend on >> > a query. In other words, default graph depends on the dataset but >>neither >> > on a SPARQL endpoint nor on a specific SPARQL query. Since the second >> > case is a query dependent interpretation it should not be correct >> > interpretation of a default graph. >> > Regards,Milorad >> > >> > >> > From: Daniel Hernández <[email protected]> >> > To: [email protected] >> > Sent: Thursday, February 5, 2015 11:37 PM >> > Subject: A question about service descriptions >> > >> > Hello, this question is about the SPARQL specification. >> > >> > A service description can include a property UnionDefaultGraph. I'm >>not >> > sure what it means. I have two possible interpretations: >> > >> > 1. If the default dataset of the service is {G0, (u1,G1), ..., >>(un,Gn)} >> > then it means that G0 is the union of G1,...,Gn. >> >> Yes. >> >> > >> > 2. If a query with a dataset description FROM a FROM b is sent to the >> > service then the default graph of the query dataset will be the union >>of >> > the graphs resulting of dereferencing a and b, i.e., the service does >> > not do the merge as it is stated in the SPARQL specification. >> >> FROM and FROM NAMED are used to describe the dataset to be queried. >> FROM names all the graph for the default model (unioned together) and >> FROM NAMED all the named graphs (each with a name). This is how it is >> in SPARQL 1.0 and 1.1, not to do with service description. >> >> FROM <a> >> FROM <b> >> describes a graph formed by reading <a> and <b> into the same graph. >> >> UnionDefaultGraph is not referring to this feature. >> >> <a> and <b> are not named graphs and don't show in GRAPH. >> >> Many SPARQL services with a fixed dataset do not support FROM, FROM >> NAMED. The datasets is fixed and can't be set by description. >> >> (( >> In TDB FROM and FROM NAMED are drawn from the underlying datastore, not >> read in from the web. And this is not tdb:unionDefaultGraph. >> https://jena.apache.org/documentation/tdb/dynamic_datasets.html >> )) >> >> Andy >> >> > >> > How I must interpret it? >> > >> > Thanks! >> > Daniel >> > >> >> Andy >> > >> > >> > >> > >> > >
