The I-JSON option - Githubissues

jansupol commented 6 years ago

Having an I-JSON option is great but since serialization of numbers outside of the IEEE-754 double precision universe isn't standardized it doesn't help that much: https://github.com/OAI/OpenAPI-Specification/issues/1519#issuecomment-377964659

BTW, does the I-JSON option really address serialization of long? If it doesn't then it is not really I-JSON compliant.

A list of things affected by the I-JSON option would be much appreciated.

jansupol commented 6 years ago

Issue Imported From: https://github.com/javaee/jsonb-spec/issues/80
Original Issue Raised By:@cyberphone
Original Issue Assigned To: Unassigned

jansupol commented 6 years ago

@cyberphone Commented I took the liberty researching the issue a bit further.

public class Dog {
    public String name;
    public int age;
    public long ijsonlong;   // Added test property
    public boolean bitable;
}

Updated demo program:

    dog.ijsonlong = 9223372036854775807l;  // Largest long there is
    JsonbConfig config = new JsonbConfig().withStrictIJSON(true);
    Jsonb jsonb = JsonbBuilder.create(config);
    String result = jsonb.toJson(dog);
    System.out.println(result);

Result:

{"age":4,"bitable":false,"ijsonlong":9223372036854775807,"name":"Falco"}

This is (as far as I can tell...) not compliant with I-JSON.

jansupol commented 6 years ago

@m0mus Commented Spec defines big double numbers behaviour in 3.16 as:

JSON Binding implementation MUST serialize/deserialize numbers that express greater magnitude 
or precision than an IEEE 754 double precision number as strings.

I see that the same should be done for long numbers and BigInteger when I-JSON support is switched on. These numbers should be serialized as strings and JSON-B implementations must be responsible for proper handling of it.

jansupol commented 6 years ago

@cyberphone Commented The Johnzon folks consider I-JSON a useless option and converted it into a no-op: https://github.com/apache/johnzon/blob/f9a916200233f8777addcbaf73807067aa7559f6/johnzon-jsonb/src/main/java/org/apache/johnzon/jsonb/JohnzonBuilder.java#L264

jansupol commented 6 years ago

@rmannibucau Commented Well johnzon defaults already match MUST of I-JSON so the option is actually useless.

Important side note to avoid spec overlapping: serialization must NOT be handled by JSON-B but JSON-P here so it can be something to port to JSON-P attention.

jansupol commented 6 years ago

@cyberphone Commented @rmannibucau This interpretation may be correct RFC-wise but it doesn't solve the interoperability problem. I see no justification for an option that does nothing...

jansupol commented 6 years ago

@rmannibucau Commented @cyberphone fully agree and I'm all for to deprecate it.

jansupol / jsonbapi

The I-JSON option #80