These are chat archives for rdfjs/public

2nd
Mar 2016
elf Pavlik
@elf-pavlik
Mar 02 2016 00:02
also rdflib.js doesn't currently use .value on NamedNode
Dmitri Zagidulin
@dmitrizagidulin
Mar 02 2016 00:02
it doesn't. (which is part of the argument above)
elf Pavlik
@elf-pavlik
Mar 02 2016 00:06
unless Tim already considers .value on NamedNode reserved for somethign else in rdflib.js, we wouldn't have colision and rdflib.js could alias .value proposed by TF to .uri or .iri or both
Dmitri Zagidulin
@dmitrizagidulin
Mar 02 2016 00:10
I think it's more a protest on philosophical grounds rather than a collision.
elf Pavlik
@elf-pavlik
Mar 02 2016 00:41
SPARQL 1.1 Results JSON also uses .value for URI https://www.w3.org/TR/sparql11-results-json/#select-encode-terms
@kidehen maybe you have some helpful insights on IRI vs. NamedNode vs. MergedNode discussed here in last hour ( from https://gitter.im/rdfjs/public?at=56d5fb2bf760676329bab660)
elf Pavlik
@elf-pavlik
Mar 02 2016 01:51
@RubenVerborgh have you looked at https://github.com/linkeddata/rdflib.js/blob/master/term.js ? Besides RDF it also has some N3 implementation... Since you work on N3.js, do you see current RDFJS representation draft fit to extend and implement N3 based on it?
elf Pavlik
@elf-pavlik
Mar 02 2016 02:01
what do you think about this N3 example of a node with multiple IRIs and numeric value https://gitter.im/rdfjs/public?at=56d61e05b01413547d894cff
elf Pavlik
@elf-pavlik
Mar 02 2016 02:08
Tim Berners-Lee
@timbl
Mar 02 2016 10:06
@RubenVerborgh — how much of N3 the langauge does N3 the module handle? Nested graphs? @foreach and @forsome ? @keywords ?
Ruben Verborgh
@RubenVerborgh
Mar 02 2016 14:56
@timbl Nothing yet, I must shamefully admit. N3.js started out as with the goal of supporting N3, but along the way, became more focused on Turtle and the rest of the family.
It is, however, still my goal to get to N3 compatibility. At the moment, what holds me back most is the relation between Turtle and N3. Is N3 a proper superset of Turtle, or is N3 the spec as defined there (with some lexical and grammatical nuances from earlier Turtle drafts)?
That said, it should not be too much work, it's only a couple of extra tokens.
Ruben Verborgh
@RubenVerborgh
Mar 02 2016 15:04
@/all The NamedNode discussion has been revived at rdfjs/representation-task-force#50. Perhaps we can add structured arguments there, and alternative solutions such as aliasing. Note however that, in reality, the discussion comes down to what the value of .termTypewill be for <a> (currently: "iri"). The interface name itself is not be exposed and is up to individual implementers.
Thomas Bergwinkl
@bergos
Mar 02 2016 16:32
yes, we should collect all arguments in a structured form, before we start the discussion again.
@RubenVerborgh the factory uses the interface names as method names, so it's not only about the .termType
Tim Berners-Lee
@timbl
Mar 02 2016 17:39
@RubenVerborgh The objective was that turtle was a formal subset of N3.
The fact that a different spec was prodcued and no process to keep them together was a shame.
Certainly I always use an N3 parser to parser turtle.
If theer are any differences then they are obscure.
and they should ideally be fixed.
Ruben Verborgh
@RubenVerborgh
Mar 02 2016 17:56
@timbl Alright, I'll implement it as such then. Hope to get to this during summer, and finally release N3.js 1.0.
The differences are indeed minor, but they include things such as single quotes for literals, and slightly different regexes etc. More or less everything that was introduced quite late Turtle. I.e., the N3 spec started as a kind of fork of what Turtle was back then, but it still needs to be merged/rebased on HEAD :smile:
Tim Berners-Lee
@timbl
Mar 02 2016 19:33
:-)
I tell you what — it would be worth collaborating if we could get them parallel and also allowing literal dates in turtle 1.1 — my parsers both support them
they are really nady for a lot of data
like 2016-03-01
and 2016-03-01T23:23Z
Ruben Verborgh
@RubenVerborgh
Mar 02 2016 22:41
I definitely want to help getting the N3 spec even with the Turtle spec. Would help a lot of potential implementers, I think. Where do we get this started?
elf Pavlik
@elf-pavlik
Mar 02 2016 23:41
@bergos @RubenVerborgh I see in Data Factory you use parameter names for different terms: iri (for IRI), value (for Literal), name (for Variable) but than access them all using .value no matter of term type
Tim Berners-Lee
@timbl
Mar 02 2016 23:43
@RubenVerborgh good question — ask me again tomorrow
elf Pavlik
@elf-pavlik
Mar 02 2016 23:50
@timbl what do you think about Default Graph Term proposed in current draft?