Re: JSON-LD bnode canonical naming algorithm

>
> Everyone on here has use cases that are important to them. It is
> important that we look at each of them. We need to discover which ones
> we can address while keeping the JSON-LD spec under control.
>

If the audience for this spec is just the people who show up here, then yes.
But I think we're trying to do more than that. JSON-LD seems like an
opportunity to try to push LD into a form where it could take off among the
vast numbers of web/data developers who don't currently give a crap about
RDF. The road to that level of approachability is, I think, not a matter of
simply trying to accommodate people just because they show up on this
mailing list with expectations.

We should be gathering use cases from the community. If JSON-LD doesn't
> address many of Needle's use cases, then that's not good either.
>

No, it's the other way around: Needle's model is much simpler. I'm not here
trying to lobby for my personal use-case, I'm here as a system-designer
arguing for a simpler system-design. As it happens, Needle is a horizontal,
domain-neutral graph-database. So I'm not advocating my Needle-like proposal
because that's how we do it in Needle, we do it that way in Needle because
that's my solution, as a designer, to the problem of modeling and
serializing a graph database. And thus my conviction that it is possible to
do without a bunch of this stuff, because in Needle we *do* do without it.
No blank nodes, no CURIEs, no inline context, no inline datatypes, no inline
literals, no chaining, no rdf:lists (no RDF). Schema is just more nodes. I
have so far seen no graph that can't be represented this way, so I feel
pretty confident about its expressiveness.

glenn