Closed visrra closed 3 years ago
There are several differences, but underneath they are the same. pathos.pools.ProcessPool
is built on top of pathos.pools._ProcessPool
(same object as in pathos.helpers
). The latter exactly replicates the multiprocessing
interface, while the former conforms to an interface that is uniform across several backends. The ProcessPool
can take several *args
, is globally cached, and can be reopened once closed. The_ProcessPool
gives you whatever multiprocess.Pool
provides. Both cases can handle lambda functions, and should provide exactly the same level of serialization.
Thanks @mmckerns
Are there any advantages of ProcessPool compared to pathos.helpers.mp.Pool? The latter seems to be able to handle lambda functions, and it can also display logging outputs of tasks.