numbers v4.0.0 Numbers.Numeric behaviour
Any module that wants to be a Numeric type, and to be able to be called by the functions in Number, should make sure that this behaviour is followed.
Your callbacks will only ever be called with two versions of your Numeric struct.
If one of the functions in Number is called with one of the arguments being an integer or float,
then it is first converted to your Numeric struct by calling YourStructModule.new(the_int_or_float) on it.
Summary
Types
numeric_struct should be a struct that follows the Numeric behaviour
To be used in your typespecs at any place where a Numeric type can be used
Callbacks
The absolute value of a number
Adds two numbers together
Divides the rhs by the lhs
Unary minus. Should return the negation of the number
Multiplies the two numbers together
Creates a new numeric_struct from the given built-in integer or float
Power function, x^n
Subtracts the rhs number from the lhs number
Convert the custom Numeric struct to the built-in float datatype
Types
numeric_struct should be a struct that follows the Numeric behaviour.
To be used in your typespecs at any place where a Numeric type can be used.
Callbacks
The absolute value of a number.
Adds two numbers together.
coerce(t, numeric_struct) :: {numeric_struct, numeric_struct}
coerce(numeric_struct, t) :: {numeric_struct, numeric_struct}
Divides the rhs by the lhs.
To be clear, this division operation is supposed to keep precision.
Unary minus. Should return the negation of the number.
Multiplies the two numbers together.
Creates a new numeric_struct from the given built-in integer or float.
In the case of reading a float, it is okay to lose precision.
This callback is optional, because there are data types for which this conversion is impossible or ambiguous.
If more control is needed over the creation of a datatype from a built-in type,
or coercion between two custom data types, implement coerce/2 instead.
Power function, x^n.
When this optional function is not provided, Number will use the ‘Exponentiation by Squaring’
algorithm to calculate the result (which uses log(n) repeated multiplications).
Add it to your data type if it is possible to compute a power using a faster algorithm.
Subtracts the rhs number from the lhs number.
Convert the custom Numeric struct to the built-in float datatype.
It is okay to lose precision during this conversion.
This function is optional, because there are many numeric types that cannot be (unambiguously) converted into a floating-point number.