Re: [JSON] I say again, what *is* JSON?

Ivan Herman wrote:
> One more point to this discussion. I understand Peter's criticism, and it is good that he provokes the group in making some of the issues clear. That being said, if the WG decided not to do a JSON serialization _solely on ground of a missing 'standard'_, I think the community would simply not understand this and would see this as yet another argument why W3C as a whole, and the Semantic Web in particular, should be regarded as irrelevant for the Web out there. I would prefer not to go there.

+1, we are (hopefully!) defining a JSON compatible standard, for all 
it's lacking an internet standard itself, it is well defined and used 
heavily by the industry, some nuances such as avoiding particular 
multiple "keys" with the same name are simply not encountered in day to 
day usage of JSON (since you can't create an object which has duplicate 
keys to serialize).

I guess my point is, JSON is well defined by it's usage, the media type 
we need to be compatible with (to create our own +json type) is defined 
by rfc 4627, we need to keep that in mind for IANA/IESG. For practical 
use though, ECMAScript-262 V5 and the many, many, many implementations 
of JSON.parse and JSON.stringify stipulate what we need to remain 
compatible with practically. As in, we don't have free unending scope 
here, but we can clear up any issues and place our own constraints on 
our own specification, such imposing limits on numbers, requiring a 
certain "structure" to the data (sequence of objects only for instance), 
and disallowing duplicate keys.

Personally, I think we'll be just fine - the compatibility test at the 
end of the day is just to JSON.parse what we produce and check it's what 
we expect!

Best,

Nathan

Received on Friday, 25 March 2011 08:51:50 UTC