Re: [Json] Limitations on number size?

[Nico Williams <[e-mail]>]

On Tue, Jun 4, 2013 at 4:39 AM, Stefan Drees <[e-mail]> wrote:
> In the wild - and in my little eco system at least - JSON is used more like
> ONS (Object Notation Serialization), isn't it?

Often, yes.

> I regard the "indefinite-length stream of digits to be interpreted as a
> number" as a small price for the unversal usability and acceptance of JSON
> as a format.

Off-topic: Every value type in JSON (including null and booleans) has
indefinite-length encoding.  I think this is a highly valuable aspect
of JSON, except for the need to escape strings.

> In conclusion such a change should IMO a) not qualify as a change being
> covered by the charter of the WG and would b) break several interesting use
> cases, where implementers would then be forced to put potentially(!) non
> IEEE754 64-bit number literals always into quotes instead of as now happily
> producing and consuming "streams" of say:

I agree.

Nico
--