Re: [Json] Proposal: the minimal edit

[Tim Bray <[e-mail]>]

Since there are those who (gasp!) are not native speakers of IEEE Floats,
it’s probably helpful to say something like “which have 53 bits of
precision (2**53 is roughly 9.007 * 10**15)”


On Sat, Jun 22, 2013 at 1:59 PM, John Cowan <[e-mail]> wrote:

> R S scripsit:
>
> > + N. Differences from ECMAScript
> > +
> > + ECMAScript implementations produce and consume primitive JSON values at
> > + the root level of JSON documents.
> > +
> > + ECMAScript implementations generate and consume unpaired Unicode
> > + surrogate code points in JSON documents.
> > +
> > + When there are duplicate names within an object, ECMAScript parsers
> > + overwrite the value corresponding to such names with the value that
> > + appears last in the serialization.
>
> To which I add:
>
> + ECMAScript parsers only accept, and ECMAScript generators only generate,
> + numbers that correspond to the finite values of binary 64-bit IEEE
> + 754:2008 floating point numbers.
>
> We'll need a non-normative reference, too:
>
>     IEEE Computer Society (August 29, 2008). IEEE Standard for
>     Floating-Point Arithmetic. IEEE.  doi:10.1109/IEEESTD.2008.4610935.
>     ISBN 978-0-7381-5753-5. IEEE Std 754-2008.
>
> --
> Time alone is real                      John Cowan <[e-mail]>
>   the rest imaginary
> like a quaternion       --phma          http://www.ccil.org/~cowan
>