[Json] Fwd: RFC 4627bis vs RFC 6902 (JSON Patch)

[Francis Galiegue <[e-mail]>]

Sorry, user error, didn't find its way to the list


---------- Forwarded message ----------
From: Francis Galiegue <[e-mail]>
Date: Thu, Feb 20, 2014 at 7:17 PM
Subject: Re: [Json] RFC 4627bis vs RFC 6902 (JSON Patch)
To: Nico Williams <[e-mail]>


On Thu, Feb 20, 2014 at 6:36 PM, Nico Williams <[e-mail]> wrote:
[...]
>>
>> What are your thoughts?
>
> I don't see the conflict.
>
> FYI, text in RFC4627bis was arrived at after very many and lengthy
> (well over 1,000 posts) threads on the JSON WG list.  I'd summarize
> the part about duplicate names as:
>
> The nature of JSON as it is used in the wild is such that RFC4627bis
> couldn't require any specific parser behavior as to duplicate names --
> the canonical example is online parsers, whose very purpose is to keep
> minimal state while parsing, which means that they can't detect
> duplicate names, which means that specific parser behavior regarding
> duplicate names could not be REQUIRED.  (There was also a question of
> whether specific behavior could be required of the parser _and_
> application together, and the conclusion was also "no".)
>

OK, then what about constraints on JSON emission instead? RFC 4627bis
section 10 does mention that any emitted JSON must be conform to the
grammar; can't a condition be added that in the case of JSON objects,
they SHOULD ensure about the uniqueness of member names?

Admittely, I can live with the texts as they currently stands; it is
just that this shadowy area leaves an acrid aftertaste in my mouth,
that's all ;)

--
Francis Galiegue, [e-mail]
JSON Schema in Java: http://json-schema-validator.herokuapp.com


-- 
Francis Galiegue, [e-mail]
JSON Schema in Java: http://json-schema-validator.herokuapp.com