Re: [Json] "best practices" Vs. Profile for i-json

[Nico Williams <[e-mail]>]

On Mon, Aug 04, 2014 at 01:29:03PM +0100, Stephen Dolan wrote:
> Two JSON documents differing only in whitespace outside strings are
> (MUST be? SHOULD be?) considered equivalent by any implementation. If
> you want to specify "use doubles" you should state something similar,
> like:
> 
> "An implementation [MUST NOT/SHOULD NOT] distinguish between two
> numeric literals whose decimal values round to the same IEEE754
> double-precision value (using standard round-to-nearest, ties-to-even
> rounding)."

Is a MUST really needed?  Suppose you have an encoder that produces
numbers (not irrationals) that require better-than-IEEE754 to represent
exactly.  A parser that can only handle IEEE754 should accept such
numbers, but with loss of precision/range -- exactly what we already
knew to expect from JSON as far as interoperability goes.

My concern is that a MUST NOT -even a SHOULD NOT- is too strong and will
either a) cause implementations with better-than-IEE754 numbers to lose
date at encode time even when the next parser in the chain also has
better-than-IEE754 numbers, or b) this will discourage
better-than-IEE754 implementations.  We might end up stuck with IEEE754
for a long, long time.

It should suffice to have a requirement that parsers support at least
IEEE754 (e.g., not 32-bit integers) and a warning that sending numbers
that cannot be represented with IEEE754 may lead to data loss.

Nico
--