CODEWITHSUNDEEP
pythonmultithreading
WoJ
WoJ

Reputation: 29957

Why does multithreading have a Lock object?

My understanding is that threading allows in reality to have only one thread active at a time, continously switching between threads. This is useful when having IO-bound operations where the worload is effectively offloaded somewhere else (an PI, a database, ...).

If so, why is there a need for a Lock() object? There is no risk that a variable is accessed by two threads simultaneously (as it can be the case in multiprocessing) so I fail to see a real usage for locks in this context.

Upvotes: 0

Views: 52

Answers (2)

mouviciel
mouviciel

Reputation: 67831

There is no risk that a variable is accessed by two threads simultaneously

It depends on the scheduler used to implement multithreading. Context switches may occur on any interrupt no matter what the current thread does. Therefore a thread accessing a variable may be interrupted on a clock interrupt and another thread accessing the same variable may be activated.

Upvotes: 3

Daniel
Daniel

Reputation: 42748

First of all, locks secure whole areas, think of updating a file:

with lock:
    with open("some_file", "r+") as f:
        do_something(f)

Even single operations like

a['b'] += 1

might lead to multiple operations (read value of a['b'], increment, write to a['b']), and need to be secured by a lock:

with lock:
    a['b'] += 1

Upvotes: 2

Related Questions