Re: [Json] I-JSON vs. JSON-S

[Nico Williams <[e-mail]>]

On Mon, Jul 8, 2013 at 7:17 AM, Carsten Bormann <[e-mail]> wrote:
> JSON-S is just the syntax.  The data model being used may be related
> to that of JSON, but ultimately is application specific.  So, Solr can
> have their repeated keys, the "array of general 16-bit values"
> faction can have their unpaired surrogates, etc.

I think even for JSON-S we should leave the duplicate name text that's
in RFC4627.  Solr can still have their dup names, and strings would be
arrays of 16-bit code units.

> When people talk about "not breaking things", it is just not clear
> whether they mean "carry along all usages of JSON-S" or "make sure all
> I-JSON implementations stil interoperate flawlessly".  That's why the
> current charter may have less of a defined meaning than it appears.

I think "don't break me" mostly means: allow my non-I-JSON usage.
I-JSON is about guaranteeing interop because we [will] have a standard
with sufficiently normative language.  So, mostly agreed.

> Unless the two levels of interoperability are clearly identified, it
> is not possible to break neither.

Interop will require specifying which JSON you're using, or actually
sticking to a subset of JSON that happens to interop (which will be
very close to I-JSON, if not exactly I-JSON).  Seems fine to me, given
that we have no other way forward.

Nico
--