Re: [Json] JSON Schema Language

[Anders Rundgren <[e-mail]>]

On 2019-05-05 07:39, Nico Williams wrote:
> On Sat, May 04, 2019 at 06:47:35AM +0200, Anders Rundgren wrote:
>> On 2019-05-04 01:59, Ulysse Carion wrote:
>>> Hello all,
>>
>> Hello Ulysse,
>>
>> I'm a former user of XML Schema.   I was therefore looking for similarities
>> like data types and facets but didn't find them.
>>
>> Example: although 10.0 is a valid JSON Number, in system where you expect
>> an integer, this should be flagged as a syntax error.
> 
> As a maintainer of one popular implementation, this idea that having a
> zero fractional part makes a number non-integer annoys me a great deal.

As I described, this is incompatible with other popular implementations like Jackson for C#.

A schema language should provide the flexibility you want but users should be aware of that certain combinations may not work so well in practice.

A schema may for example specify an BigInteger "huge" and suggest serializing this as JSON Number which will give surprises when processed by ECMAScript's JSON.parse().

Anders

> 
> I would insist that any parser that distinguishes real and integer
> representations should parse numeric values with a zero fractional part
> (and non-negative exponent) as integers.
> 
> 10.0 is an integer, even if a parser only supports, e.g., IEEE754
> representations of numeric values.
> 
> Nico
>