proposal: N-Quads as subset of TriG

Thinking about our abundance of RDF syntaxes, I'm wondering if we can 
make N-Quads be a subset of TriG.   Specifically, I'd suggest each line 
of an N-Quads file either be:

    1.  An N-Triples Line
    2.  GRAPH <iri> { an-n-triples-line }

I know this isn't compatible with old N-Quads.  That's a shame. But it 
would make N-Triples, N-Quads, Turtle, and TriG all just be profiles of 
the same language, readable with the same parser.   And that language is 
closely aligned with SPARQL, being the same where one would expect it to 
be. That seems like a very good thing.

It would mean the W3C Recommended RDF languages would be: RDF/XML, RDFa, 
JSON-LD, and TriG (with its profiles, especially Turtle). Those are each 
so obviously different, I see little possibility of confusion or need 
for advice.    And that's a very good thing.

Maybe we should use a different name, since it's not at all like 
N-Quads.   Perhaps "Line-Mode TriG" or "Primitive TriG" or "Line Quads" 
or "TriG Line Dump" or "Dataset Line Dump Format".     (Do we want to 
rename N-Triples, too?  I know 2013 N-Triples isn't exactly the same as 
2004 N-Triples, so maybe a new name would be helpful? Or is it close 
enough that the same name is fine.   N-Triples and N-Quads are rather 
obscure names.)      Maybe "N-Triples with named graphs" or "Dataset 

I guess the problem with this would be if lots of people are using 
N-Quads as is, in the open, and are totally not convinced by this 
argument.  If they're going to keep using non-TriG N-Quads, even if we 
do this, that would be a little awkward.   Is anyone reading this 
potentially in this camp?  Or do you know anyone who is?

If not, can we please align on one language like this?

      -- Sandro

Received on Sunday, 14 July 2013 04:13:47 UTC