Re: [Json] Nudging the English-language vs. formalisms discussion forward

[Phillip Hallam-Baker <[e-mail]>]

On Mon, Feb 17, 2014 at 8:16 PM, Nico Williams <[e-mail]>wrote:

> My advice:
>
> a) welcome WG work items for specifying one or more Informational JSON
> schemas,
>
> b) recommend (not RFC2119 RECOMMEND) the use of a JSON schema when
> specifying applications using JSON, if appropriate -- a schema of the
> application spec authors' choosing.
>

+1

I would prefer standards track to be limited to actual protocols and
mandatory to implement features of protocols. We don't need to agree on the
schema unless it goes over the wire.

I really want to avoid getting to the point where we pass JSON schemas or
references to schema over the wire like we do with .xsd. That does
absolutely no good in any situation I am aware of.


One other point, I was needing a quick HTTP client for a part of an app so
I just realized I can adjust my JSON schema tool to generate code to parse
headers in a HTTP response. It is a little messier than JSON parsing as
HTTP headers are not quite as regular as JSON. But it is quite easy to
avoid messing with EBNF with a few hints to give the variant of the
encoding.

[Of course I since my Web Services are JSON data in a HTTP wrapper, having
JSON data in a JSON wrapper is quite attractive. More attractive still is
JSON-C encoding on the inner and outer wrappers. Thus finessing both
HTTP/2.0 and CORE]

-- 
Website: http://hallambaker.com/