Re: [Json] Proposal for strings/Unicode text

[Paul Hoffman <[e-mail]>]

On Jun 16, 2013, at 10:02 PM, "Manger, James H" <[e-mail]> wrote:

>>> The point is that if JSON is encoded in UTF-8, any surrogate code
>>> points MUST be escaped, even though the grammar does not say so.
>> 
>> What about changing the grammar to make that clear?
>> 
>> unescaped = %x20-21 / %x23-5B / %x5D-%xD7FF / %E000-%10FFFF  ; Any
>> unicode code point except control characters, QUOTATION MARK,  ;
>> REVERSE SOLIDUS, or code points reserved for UTF-16 surrogates
> 
> +1
> Unpaired surrogates cannot be interchanged reliably so they should be dropped from this ABNF. I don't mind a note saying how some implementations handle them (or their escaped form).
> 
> Fixing typos and tweaking the comment:
> 
>  unescaped = %x20-21 / %x23-5B / %x5D-D7FF / %xE000-10FFFF
>    ; any Unicode scalar value, except those that must be escaped
>    ; (control characters, quotation mark, and reverse solidus)


<no hat>

Making comments in ABNF that disagree with the ABNF itself seems like a completely terrible idea that will lead to lack of interoperability. Complicated ABNF is better than simple ABNF that has contradictory comments.

--Paul Hoffman