Re: [Json] What is a JSON-text? [REVISITED]

[Nico Williams <[e-mail]>]

On Thu, Oct 3, 2013 at 12:34 PM, Peter Patel-Schneider
<[e-mail]> wrote:
> The proposal is to *change* JSON in a way that would break some current JSON
> parsers and, worse, break some current consumers of JSON, to get closer
> syntactic conformance with the ECMAScript JSON implementation.

Not really: you can have two definitions of top-level and describe
their interoperability characteristics (the original array-or-object
interops best).

In practice many, many JSON parsers already support at least an option
for top-level value types other than array/object.

> If there is a goal to align JSON with the ECMAScript JSON implementation,
> then I suggest that requiring EMCAScript numbers is much more important and
> that even officially requiring support for unpaired UTF-16 surrogate code
> units is more important.

I think the OP wanted one cheap ECMAScript compat improvement.  I
agree that this is the "cheapest" such improvement.  All the others
you propose have significantly more impact on existing implementations
for them to be able to support the ECMAScript variant.

I'd settle for noting the differences between IETF JSON and ECMAScript.

Nico
--