threading : A Thread exit before it finishes the job

Question

I am trying to understand the "daemon" flag of python thread. I know that

A thread can be flagged as a "daemon thread". The significance of this flag is that the entire Python program exits when only daemon threads are left. The initial value is inherited from the creating thread.

But in my case, python program exits before daemon threads are left and the threads does not finish its work.

def ThreadDecorator(f):
    def wrap(*args,**kargs):
        t=Thread(target=f,args=args,kwargs=kargs)
        t.daemon=True
        t.start()
return wrap

@ThreadDecorator
def runSomething(*args,**kwargs):
    i = 0
    attente = 0
    fileName=kwargs['file']
    f=open(fileName,'wt')
    while i < 50000:
        f.write(str(i)+"\n")
        # attente += 0.2
        # attente += random.randint(1, 60) / 100
        # time.sleep(attente)
        i += 1
    f.close()
    return "Finished"

And main program

runSomething("5",file='test1.txt')
runSomething("6",file='test2.txt')

The first thread write only 5000 first integer while the second does not write any number

The quote you added explains the behaviour. When then main thread ends (the one calling runSomething) the daemon threads are killed/ended. — Holloway
If you want them to keep running until completion, yes, although they are False by default if they are spawned from a non daemon thread. — Holloway
See the paraphrasing of what's in the documentation about daemon threads in this answer. — martineau

ForeverLearner ForeverLearner · Accepted Answer · 2016-11-02T09:44:14

I feel the main block should contain a pool. Once all the threads are complete, you can call the pool.close() which ensures that all the threads have completed execution.

from multiprocessing import Pool
from datetime import datetime

pool = Pool(processes = 2)
num_workers = 2
curr_worker = 1
val = 5
if (curr_worker < num_workers ):
    file_name = 'test'+str(curr_worker)+'.txt'
    pool.apply_async(run_Something,args=(val,file_name,)))
    curr_worker += 1
    val += 1
else:
    pool.close()
    pool.join()
    curr_worker = 1
    val = 5
print ("Workers returned result", datetime.now().strftime('%Y-%m-%d %H:%M:%S'))

The main exits after initiating the daemon threads. Since the daemon threads stop on the termination of the non-daemon main, we see that only test1.txt has 5000 records.

tasks for main is to initiate the below, post which it will exit:

runSomething("5",file='test1.txt')
runSomething("6",file='test2.txt')

threading : A Thread exit before it finishes the job

1 Answers