JSON

[Kent Karlsson]

Den 2013-03-18 17:14, skrev "John Cowan" <cowan at mercury.ccil.org>:

> Stephane Bortzmeyer scripsit:
> 
>>> It hardly matters, since this is valid JSON:
>> 
>> No, it is not. RFC 4627:
>> 
>>> The names within an object SHOULD be unique.
> 
> SHOULD, alas, is not MUST.
> 
>> The question was discussed a lot of the IETF meeting in Orlando last
>> week. This is because ECMAscript standard does not say exactly the
>> same thing (one of the main differences between the two JSON
>> specifications) and there is a project to make an IETF JSON working
>> group to produce a 4627bis.
> 
> I hope so.  There is another known problem:  by my reading of the
> RFC, "\D800" is not a valid string, because U+D800 is not a Unicode
> character.

I think you mean "Unicode codepoint". Requiring "character" whould require
that is is assigned (to a character), and never something as yet unassinged,
making this highly version (Unicode version) dependent.

> Crockford has said, however, that he intended it to be
> valid: that is, JSON strings, like JavaScript strings, are sequences
> of UTF-16 codepoints

I think you mean "UTF-16 code unit", as codepoints are encoding independent,
while code units are encoding dependent.

    Kind regards
    /Kent K


> rather than of characters.