This is a public chat log generated from the #semsol IRC channel.
14:58:32
bengee: i think there might be a slight bug in the RDFa parser
14:58:34
http://arc.web-semantics.org/demos/rdfa_tests/extract.php?url=http://sw-app.org/mic.xhtml
14:59:04
see there is a bnode for one of the foaf:interest properties
15:08:32
my guess would be that you could just remove all triples where the object is a bnode that does not occur as a subject
15:08:39
and it would all be cool
15:19:04
hm, no idea where that bnode is coming from
15:20:04
prolly from the empty rel
15:31:42
oh, the extractor actually generates 4 bnodes, one for each interest, but all with the same id, so only one is shown
15:45:48
bengee: hmm, tried against elias torrez's web service but i guess that's old - doesn't generate any foaf:interest triples
15:46:13
kwijibo tries triplr
15:46:46
just writing a test here
15:47:23
hmm, triplr seems to do nothing with RDFa these days ... maybe it never did?
15:47:34
only GRDDL ?
15:48:13
also, i thought RDFa was going to at least recommend the GRDDL approach, but i can't find an official profile uri to use - do you know if there is one?
15:49:43
hmm
15:49:56
my synapses say "fab gandon", lemme
15:51:05
http://ns.inria.fr/grddl/rdfa/
15:51:40
not sure how official it is
15:52:49
so long as it works, it matters not :D
15:53:03
"Authors who want to be certain their documents are transformable by all [GRDDL] processors may also include a profile attribute on the head element that includes a reference to the XHTML vocabulary URI http://www.w3.org/1999/xhtml/vocab."
15:54:28
bengee: btw, what do you think of RDF-EASE ?
15:54:30
can't find a transformation link in xhtml/vocab
15:54:55
I don't find the idea too bad, tbh
15:55:09
yeah - i remember looking for it in xhtml/vocab too
15:56:02
the particular syntax seems a little too powerful (with the cascading functionality), but a subset could be helpful for non-xslt people
15:56:36
yeah - i skimmed the doc: the first simple examples look really cool
15:56:52
and then things start getting complicated ....
15:58:00
I think if you wrote a semantics-sheet that used more than the simple subset, it might be a bit of a head fuck
15:59:22
'k, found the bug
15:59:44
not too easy to fix, though
15:59:58
might be nicer and neater than XSLT as a GRDDL syntax though - it's not so much the separating the semantics from the html markup i like, but the separating the semantics from the programming logic of an XSLT layer
16:00:16
bengee: is it really a bug?
16:00:40
there is a "if (1 || ...)" in my code, and I added the "1" because otherwise I lost certain triples.
16:00:52
ah
16:01:03
I think it's a bug, yes
16:02:01
you could fix it superficially by removing the weird bnode triples afterwards
16:15:27
ah, it's related to a comment danbri made
16:16:47
i.e. whether a bnode chain like rel=next > rel=next > rel=next should generate triples
16:17:45
I think the initial spec said no, but danbri noted that knows > knows > knows might still be useful information, even when all knowees are blank
16:18:29
so, maybe you're right and it's not a bug
16:32:19
no, can't fix it, the spec is too complicated to find out how to skip the bnodes
16:33:05
I'll keep it that way, people will scream louder when triples are missing, than when there are useless triples
16:33:27
and I pass 2 more tests from the RDFa test suite ;)
16:34:42
"some day, maybe"
17:01:50
:)