Re: [Json] I-D Action: draft-ietf-json-i-json-01.txt

"For applications such as cryptography, where exact interchange of
much larger numbers is required, it is RECOMMENDED to encode them in
JSON string values."

The phrasing of "it is recommended to encode them" seems awkward to
me. Maybe "it is recommended that such values be encoded as"?

"encoded" also seems off. Maybe "written", "stored", "represented"?

I am not sure that cryptography is the right example. The most common
surprising case will be people trying to store arbitrary 64-bit
integers from commonplace types like Java longs and finding that most
of the value range can't be read & correctly represented in
Javascript. Those numbers might come from 64-bit hashes or storage
systems that assign evenly-distributed 64-bit IDs.

Cryptographic applications deal with much larger numbers, and are
already accustomed to exchanging them as binary, hex, or base64. I
doubt anyone would really even think of trying to send a 2048-bit key
as a JSON number.

On Tue, Jun 17, 2014 at 12:30 PM, John Cowan <cowan@mercury.ccil.org> wrote:
> Tim Bray scripsit:
>
>> > s2.2, "For applications such as cryptography, where exact interchange of
>> > much larger numbers is required, it is RECOMMENDED to encode them in JSON
>> > string values", I suggest adding some guidance on encoding if we have a
>> > quick consensus on decimal|hex|base64url|etc.  Otherwise leaving it as-is
>> > works.  Note that JOSE chose base64url for good reason.
>>
>> Not sure I get this... for big numbers, it’s just digits, right?
>
> Yes, but digits of what base?  Decimal, ça va sans dire; but it goes
> even better when it is said.
>
> --
> John Cowan          http://www.ccil.org/~cowan        cowan@ccil.org
> You cannot enter here.  Go back to the abyss prepared for you!  Go back!
> Fall into the nothingness that awaits you and your Master.  Go! --Gandalf
>
> _______________________________________________
> json mailing list
> json@ietf.org
> https://www.ietf.org/mailman/listinfo/json