How to share a numpy array between 2 processes on Windows?

[qstdy]

I want to make 2 processes that share a numpy array (one of which writes the array and the other reads it).

It works fine when I make 2 processes with 2 functions like this:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

from multiprocessing import Process, Semaphore, shared_memory import numpy as np import time   def reader(id, a, shm):     exst_shm = shared_memory.SharedMemory(name=shm)     b = np.ndarray(a.shape, dtype=a.dtype, buffer=exst_shm.buf)               time.sleep(2)        print('FUNCTION VERSION: ', b[0])           def worker(id, a, shm):     exst_shm = shared_memory.SharedMemory(name=shm)     b = np.ndarray(a.shape, dtype=a.dtype, buffer=exst_shm.buf)     b[0] += 10     if __name__ == "__main__":     a = np.array([0])     shm = shared_memory.SharedMemory(create=True, size=a.nbytes)     c = np.ndarray(a.shape, dtype=a.dtype, buffer=shm.buf)           th1 = Process(target=reader, args=(1, a, shm.name))     th2 = Process(target=worker, args=(2, a, shm.name))           th1.start()     th2.start()     th1.join()     th2.join()

The code above gives me 10, but when using subclasses it gives me 0 (Windows 10, Python 3.8.7 64bit).

Why do they print out different results? Have I made a mistake or is this a bug?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

from multiprocessing import Process, Semaphore, shared_memory import numpy as np import time   class Reader(Process):     def __init__(self, id, a, shm):         Process.__init__(self)                   exst_shm = shared_memory.SharedMemory(name=shm)         b = np.ndarray(a.shape, dtype=a.dtype, buffer=exst_shm.buf)                   time.sleep(2)            print('SUBCLASS VERSION: ', b[0])            class Worker(Process):     def __init__(self, id, a, shm):         Process.__init__(self)                       exst_shm = shared_memory.SharedMemory(name=shm)         b = np.ndarray(a.shape, dtype=a.dtype, buffer=exst_shm.buf)         b[0] += 10           if __name__ == "__main__":     a = np.array([0])     shm = shared_memory.SharedMemory(create=True, size=a.nbytes)     c = np.ndarray(a.shape, dtype=a.dtype, buffer=shm.buf)           th1 = Reader(1, a, shm.name)     th2 = Worker(2, a, shm.name)       th1.start()     th2.start()     th1.join()     th2.join()