Pure functions produce the same output for the same input and have no side effects.
TL;DR: Your functions should be replaceable by the computation result.
Breaking referential transparency occurs when the code introduces side effects or relies on a mutable state. This violates the principle that an expression or function can be replaced with its value without changing the program's behavior.
# Global mutable variable
counter = 0
# Function with side effect
def increment_counter():
global counter
counter += 1
return counter
# Function with implicit dependency and non-deterministic
def get_random_number():
import random
return random.randint(1, 100)
# Function with non-deterministic behavior
def get_current_time():
import time
return time.time()
import random
import time
# Function without side effects
def increment_counter(counter):
return counter + 1
# Function without side effects (but not deterministic)
def get_random_number():
return random.randint(1, 100)
# Function without side effects (can also be injected)
def get_current_time(timesource):
return timesource.time()
Many linters warn you when you violate referential transparency
Most AI assistants will avoid violating referential transparency.
Functional programming is known for its ability to enable concise, expressive, and maintainable code, as well as facilitating parallel and concurrent programming due to its emphasis on immutable data and pure functions.
There are many concepts to borrow from it.
Coupling - The one and only software design problem
Disclaimer: Code Smells are my opinion.
Credit(s): Lead photo by Wilhelm Gunkel on Unsplash
Referential transparency is a very desirable property: it implies that functions consistently yield the same results given the same input, irrespective of where and when they are invoked - Edward Garson.