Re: [Json] Encoding Schemes

[Tatu Saloranta <[e-mail]>]

On Tue, Jun 18, 2013 at 7:54 PM, Joe Hildebrand (jhildebr) <
[e-mail]> wrote:

> On 6/18/13 12:39 PM, "John Cowan" <[e-mail]> wrote:
>
> >Joe Hildebrand (jhildebr) scripsit:
> >
> >> "When serialized to an octet stream, JSON text SHALL be encoded in one
> >>of
> >> the following Unicode encoding schemes: UTF-8,  UTF-16BE, UTF-16LE,
> >> UTF-32BE, and UTF-32LE.  The default and RECOMMENDED encoding is UTF-8.
> >
> >Oh no, that SHALL will not fly.  Any encoding (which means "encoding
> >scheme" in a media-type context) can be used to represent JSON.
> >Including an EBCDIC variant.
>
> I assume you're talking about UTF-EBCDIC from tr16?  I don't see how you
> could auto-determine the encoding for that.  How about this:
>
> """
> Without an external mechanism that specifies encoding, when serialized to
> an octet stream, JSON text SHALL be encoded in one of the following
> Unicode encoding schemes: UTF-8,  UTF-16BE, UTF-16LE, UTF-32BE, and
> UTF-32LE.  The default and RECOMMENDED encoding is UTF-8.
>
> Note: the MIME type registered in section 6 does not specify a mechanism
> to specify the encoding scheme, so when used in a MIME context, one of the
> above encoding schemes MUST be used.
>
>
> Other Unicode encoding schemes MAY be used, but such octet streams cannot
> have their encoding scheme automatically detected and SHOULD NOT be
> assumed to interoperate with existing software.
> """
>
>
I like this wording, because it is somewhat compatible with the current
state of affairs.
Unlike XML, where most encodings can be detected from content (including
even EBCDIC, given xml declaration), JSON decoders have to rely on small
subset of encodings for auto-detection, and external encoding information
for others.

-+ Tatu +-