A union declaration inside a lib declares a C union:

lib U
  # In C:
  #  union IntOrFloat {
  #    int some_int;
  #    double some_float;
  #  };
  union IntOrFloat
    some_int : Int32
    some_float : Float64

To create an instance of a union use new:

value = U::IntOrFloat.new

This allocates the union on the stack.

A C union starts with all its fields set to "zero": integers and floats start at zero, pointers start with an address of zero, etc.

To avoid this initialization you can use uninitialized:

value = uninitialized U::IntOrFloat
value.some_int #=> some garbage value

You can set and get its properties:

value = U::IntOrFloat.new
value.some_int = 1
value.some_int #=> 1
value.some_float #=> 4.94066e-324

If the assigned value is not exactly the same as the property's type, to_unsafe will be tried.

A C union is passed by value (as a copy) to functions and methods, and also passed by value when it is returned from a method:

def change_it(value)
  value.some_int = 1

value = U::IntOrFloat.new
change_it value
value.some_int #=> 0

Refer to the type grammar for the notation used in union field types.

results matching ""

    No results matching ""