Re: [Json] Call for Consensus: Proposed Text for "8.1 Character Encoding"

[Pete Cordell <[e-mail]>]

On 10/05/2017 14:08, Julian Reschke wrote:
> I believe we should have separate names for JSON represented as a
> sequence of characters (such as in a string variable in a programming
> language) and for a JSON-shaped octet sequence inside a
> "application/json"-typed (HTTP) message. For the latter, enforcing UTF-8
> IMHO is attractive.
>
> I think it's ok for the spec to talk about both, but it really needs to
> be clear what we are talking about in each section.

It's an interesting thought on JSON in programs.  It would be strange to 
be able to say it was not valid JSON if it was encoded in a string 
inside a Shift-JIS encoded Ruby program for example.

My ISO layers are rusty, but it looks like we can talk about JSON 
character sequences somewhere above the transport layer, and JSON 
encoded messages somewhere below the transport layer.  The latter 
possibly being transport specific.

To me it would seem discussion of "JSON" (without any further 
refinement) ought to be independent of the lower layer transport 
encoding aspects.

"application/json" would be one transport specific encoding (for which I 
think most are happy with only UTF-8).  JSON inside a Shift-JIS encoded 
Ruby program is in effect another form of transport for a JSON message.

Cheers,

Pete.