Re: [Json] Bug in RFC7493 (I-JSON)?

The question is not whether a number makes it through a round-trip (lots more do so) but whether the result of deciding is unambiguous. If you get 2**53, you don't know whether the sender sent 2**53 or 2**53+1, so you shouldn't be accepting either number as if it were unambiguous. 

Sent from mobile

> On 6. May 2018, at 08:21, Anders Rundgren <anders.rundgren.net@gmail.com> wrote:
> 
> During testing of the JSON canonicalizer [1,2] I tried with an integer just above the specified limit (2**53) + 1 and to my surprise it didn't fail.
> A short run with an IEEE-754/ES6 debugger seemed to confirm that +-2**53 is the actual limit for integers.
> 
> Input floating point: 9007199254740991
> Output floating point: 9007199254740991
> Hex value: 433fffffffffffff
> Binary value: 0 10000110011 1111111111111111111111111111111111111111111111111111
> 
> Input floating point: 9007199254740992
> Output floating point: 9007199254740992
> Hex value: 4340000000000000
> Binary value: 0 10000110100 0000000000000000000000000000000000000000000000000000
> 
> Anders
> 
> 1] https://github.com/cyberphone/json-canonicalization#json-canonicalization
> 2] https://cyberphone.github.io/doc/security/draft-rundgren-json-canonicalization-scheme.html
> 
> 
> _______________________________________________
> json mailing list
> json@ietf.org
> https://www.ietf.org/mailman/listinfo/json
> 