Re: [Json] -0.0

[John Cowan <[e-mail]>]

Paul Hoffman scripsit:

> The additions that we are making about interoperability are about
> things that a significant number of JSON users might expect.

I don't actually think so.  They are edge cases (unpaired surrogates,
too-large or too-precise numbers, duplicate names), but edge cases wind
up being important to somebody.  Negative zero is just another edge case.

> The likely-forthcoming "implementation advice" document can take care
> of telling implementers of JSON encoders and parsers what they should
> consider.

Oh, I agree.  But all the things above, including negative zero, are about
what users can and cannot expect to survive when transmitted by JSON.

> "Don't boil the ocean." (Lots of people, recently.)

In the Middle East, that's where the drinking water comes from.  What's
more, if you have some high-temperature superconductor and need to kill
off a field of sunflowers, boiling the ocean may be just the thing.

> Number geeks^wspecialsts might expect certain things about -0, but the
> vast majority of normal JSON users won't. I think it is fine to remove
> the paragraph and assume that a more in-depth discussion will happen
> in the companion document.

You could make the same argument about Unicode geeks and unpaired
surrogates.  Are you doing so?  If not, why not?

-- 
                Si hoc legere scis, nimium eruditionis habes.