Example of a data race in Python

python_data_race.py

# Does Python have data races?
#
# As we've seen, Java can have data races, according to
# https://docs.oracle.com/javase/specs/jls/se8/html/jls-17.html
# It says an example of a data race is when:
# - there is a write in one thread,
# - a read of the same variable by another thread,
# - and the write and read are not ordered by synchronization.
#
# This program attempts to cause a data race in Python.

from threading import Thread
from time import sleep

counter = 0

def increase():
    global counter
    for i in range(0, 100000):
        counter = counter + 1

threads = []
for i in range(0, 400):
    threads.append(Thread(target=increase))
for thread in threads:
    thread.start()
for thread in threads:
    thread.join()

print(f'Final counter: {counter}')

# A few trials from my computer (Mac, 2.6 GHz 6-Core i7):
# Final counter: 31735072
# Final counter: 32829326
# Final counter: 31496003
#
# We conclude that yes, Python has data races.

This seems like more of a race condition than a data race. You can get similar results in C/C++/Rust even when not hitting data races due to the operation as a whole not being atomic w.r.t concurrent thread execution:

atomic_store_explicit(&x, atomic_load_explicit(&x, memory_order_relaxed) + 1, memory_order_relaxed);

If you're (like me) struggling to recreate this, here's an easy change to show the preemptive multitasking creating race conditions (works every time for me):

def increase():
    global counter
    for i in range(0, 10):
        val = counter
        sleep(0.0001)
        counter = val + 1

You can also play around with sys.setswitchinterval() that changes the interval for thread switching