A type constraint restricts a generic type parameter to a specified set of valid types. It replaces duplicated type-specific functions and avoids the weak guardrails of interface{} or any. Union constraints such as Cat | Dog permit only listed types, while a constraint can also describe numeric types accepted by a generic function. section 3.4 section 3.4.1
Use constraints to make generic functions both reusable and explicit:
- Start broadly with any when the function operates only on a collection structure, such as filtering a slice through a callback. section 3.4.1
- Remember that listed types are exact by default: a named type whose underlying type is int8 does not satisfy a constraint containing only int8. section 3.4.2
- Prefix a type with ~ to admit named types sharing that underlying type, or use a supplied constraint such as constraints.Integer. section 3.4.2
A numeric constraint with a named type
Define a reusable operation with a constrained parameter: `type Numeric interface { ~int8 | int16 | int32 | int64 | float32 | float64 }`, then implement `func doubler[T Numeric](value T) T { return value * 2 }`. A named `Smallint int8` can pass because `~int8` includes types with int8 as their underlying type. For broader integer coverage, `constraints.Integer` reduces the declaration and accepts `Smallint`. section 3.4.2
