Why are mutexes needed in Python if GIL already guarantees single-threadedness?

GIL (Global Interpreter Lock) in Python guarantees that only one thread executes Python bytecode at a time, preventing concurrent access to Python objects at the interpreter level. However, mutexes (Locks) are needed to protect shared resources at the application logic level, especially when it comes to:

• Synchronizing access to data that may be modified by multiple threads.
• Coordinating complex operations where avoiding race conditions is important.

For example, if you have a shared list and multiple threads add elements to it, a mutex ensures that add operations do not overlap and do not lead to data corruption.

import threading

lock = threading.Lock()
shared_list = []

def add_item(item):
    with lock:
        shared_list.append(item)

Thus, GIL provides safety at the interpreter level, while mutexes provide safety at the application logic and data level.