pure

Pure functions are side effects free and do not use external information.

The opposite of pure functions are input output io and random functions.

Operators are usually pure meaning that they are side effects free and do not use global information.

Purity can be declared or inferred

square x = x*x
square.pure? true

purely square x = x*x  # ok declared purity confirmed inferred purity by compiler
square.pure? true

time := now!
time.input? true # reads time from system
time.random? true # time changes and is somewhat random
time.output? false

purely calculate fibonacci number n := fib_cache[n]=fib(n)  # error: function declared as pure uses externals
purely calculate fibonacci number n := return fib_cache[n]  # error: function declared as pure has side effects 
purely calculate fibonacci number n := fibonacci(n-1) + fibonacci(n-2) # ok, declared purity confirmed by compiler

Pure functions have the benefit that they can be inlined implicitly or explicitly on demand. Pure functions allow for a better code analysis, optimization and correctness proving. Pure functions simplify or enable type inference. Pure functions are generally a sign of good coding style.

⚠️ construction and data creation is considered pure, because these are internal operations with the only side effect being that some memory is used. Exception: user defined constructors which are explicitly not pure.

Operators are usually pure