Dataset: Deleting default graph

[uholzer]

I am currently working on support for the SPARQL 1.1 Graph Store Protocol in rdfextras-web. I use rdflib.Dataset and I like to discuss this issue:

The following tries to delete the default graph, once using Dataset.remove_graph and once using the SPARQL update DROP DEFAULT. In both cases I expect all triples to be removed from the default graph, while the default graph itself remains. It turns out, that Dataset.remove_graph(Dataset.DEFAULT) does nothing:

from rdflib import Dataset, URIRef
d = Dataset()
d.add((URIRef('a:a'), URIRef('b:b'), URIRef('c:c')))
print("Before:")
for q in d.quads((None,None,None,None)): print(q)
d.remove_graph(Dataset.DEFAULT)
print("After remove_graph:")
for q in d.quads((None,None,None,None)): print(q)
d.update("DROP DEFAULT")
print("after DROP DEFAULT")
for q in d.quads((None,None,None,None)): print(q)

This yields

(rdflib.term.URIRef(u'a:a'), rdflib.term.URIRef(u'b:b'), rdflib.term.URIRef(u'c:c'), None)
After remove_graph:
(rdflib.term.URIRef(u'a:a'), rdflib.term.URIRef(u'b:b'), rdflib.term.URIRef(u'c:c'), None)
after DROP DEFAULT

[joernhees]

i agree, this is weird

[gromgull]

@iherman : you contributed the dataset class - could you have a look at this?

[gromgull]

This was fixed by the merge on #309, there is also a test for it here:

rdflib/test/test_dataset.py

Line 98 in 06be678

# removing default graph removes triples but not actual graph