[Json] comments on Ihttp://tools.ietf.org/html/draft-bray-i-json-01

[Larry Masinter <[e-mail]>]

I have trouble with lots of the normative language in the I-JSON spec.

If you're defining a profile, then go ahead and define the set of values
that are allowed and define values that generators MUST NOT generate
in order to be conforming, and the cases where parsers which are checking 
for conformance to the profile MUST check.

So: numbers: 
>  Implementations which generate I-JSON messages MUST NOT assume that
  >  receiving implementations can process numeric values with greater
  >  magnitude or precision than provided by those numbers.

As written, it sounds like a constraint on the mental state of the author
of the generating implementation.

" MUST NOT assume that
   the order of object members"
I think this is a constraint on receivers of JSON messages, that
they MUST process all permutations equivalently.

I-JSON has no length limits. Security applications probably need to
agree on length limits.

----
  Specifications whose
   messages are specified to be I-JSON messages SHOULD specify the use
   of a media type of the form "application/XXX+i-json", where XXX is
   specific to the specification.

I argued against application/i-json in the meeting, but it's not in the
draft. Is there anyone who still wants it?    I see no value to
application/XXXX+i-json over application/XXX+json

since any application needs to define additional constraaints.

For example, i-json has no length limits, while any practical
implementation will have some, and security applications that
want to insist on consistent behavior will need length limits.

Larry
--
http://larry.masinter.net