Clarification: scalar coercion rules and default input values

[tonyghita]

I'm hoping to get some clarification around the Float portion of the specification.

Specifically, I'm considering the input coercion section and how this relates to default values.

When expected as an input type, both integer and float input values are accepted. Integer input values are coerced to Float by adding an empty fractional part, for example 1.0 for the integer input value 1.

Say we have the schema

input Example {
  f: Float = 1
 }

and we do an introspection query for the default value of Example.f

{
  __type(name: "Example") {
    inputFields {
      defaultValue
    }
  }
}

Should the defaultValue that comes back on the introspection query response be "1" (as written) or "1.0" (according to the input coercion rules)?

[benjie]

No, the 1.0 is just an example of a number with an empty fractional part; what you do in your programming language of choice is implementation specific (for example in C you'd cast from int to float, in JS ints and floats are both just numbers, etc). The float values 1, 1.0 and 1.00 are all equivalent and GraphQL does not expect (or want) you to distinguish them.

When it comes to output, the spec states:

Response formats that support an appropriate double-precision number type should use that type to represent this scalar.

For example if you were using protobuf as your response format then you'd want to use the double type.

Mostly with GraphQL we use JSON (but that's not required) which doesn't explicitly differentiate between int and float, so representing the float in the simplest format (1) is perfectly fine.

TL;DR: no, just returning the float 1 is fine.

[tonyghita]

@benjie that seems reasonable. What does this sentence of the specification refer to?

Integer input values are coerced to Float by adding an empty fractional part, for example 1.0 for the integer input value 1.

If the coercion of Float is implementation specific, why specify this behavior? Where would this transformation be reflected?

[benjie]

If the coercion of Float is implementation specific, why specify this behavior?

That's a programming-language-agnostic description of how to convert an int to a float.

Where would this transformation be reflected?

During coercion; i.e. if you receive an int where a float was expected, convert it to a float that has the same integer part and an empty fractional part.

[tonyghita]

Thank you for the clarification!