Declare decimal-valued fields such as measurements, rates, or prices.
Schema design
GraphQL · Const
GraphQLFloat
GraphQLFloat is GraphQL’s built-in Float scalar for finite signed double-precision IEEE 754 values. It accepts integer or floating-point inputs, while invalid or non-finite values cause errors.
How it works
GraphQLFloat is GraphQL's built-in Float scalar. It represents finite signed double-precision IEEE 754 values, accepts integer and floating-point inputs, and rejects strings or non-finite values when they cannot be represented correctly.
Syntax
FloatReturns
A finite signed double-precision IEEE 754 value.
Key points
- Float represents finite signed double-precision numbers under IEEE 754.
- Float arguments accept both integer literals and floating-point literals.
- An integer input is converted to its floating-point equivalent.
- Resolvers should normally provide double-precision floating-point values for Float fields.
- A service can convert another numeric internal value when the conversion is reasonable and lossless.
- The transport representation should use its available double-precision numeric type.
Common mistakes
- NaN and Infinity are not valid Float results and must produce a field error.
- Numeric strings are not accepted as Float inputs, even when their contents look like numbers.
Example: Query a Float field
query ProductPrice {
product {
price
}
}A field declared as Float can return a finite double-precision numeric value.
When to use it
Accept numeric filters or calculation inputs that may contain fractional values.
Query arguments
Expose finite double-precision values from application data.
Resolver results
Interview questions
QWhat does GraphQL Float represent?
Float is a built-in scalar representing signed, finite, double-precision IEEE 754 values. A service should expose an appropriate double-precision number representation for it.
Follow-up: What happens when a resolver returns an integer? It may be coerced to Float when reasonable without losing information, such as returning 1.0 for 1.
QWhich input values can a Float argument accept?
Both integer and Float literal inputs are accepted for a Float argument. An integer is converted by adding an empty fractional part, so 1 becomes 1.0.
Follow-up: Are numeric strings equivalent? No. Strings with numeric content must be rejected as an incorrect input type.
QHow are resolver results serialized or coerced for Float?
A field returning Float expects a double-precision floating-point internal value. Non-floating-point values may be coerced when reasonable and lossless; otherwise execution must report a field error.
Follow-up: What special values are invalid? NaN and Infinity cannot be coerced to Float and must cause a field error.
Related
IntStringBooleanIDScalar types