如何使用 inheritance 进行多处理？

Question

I'm trying to speed up the running time of the script with multiprocessing. When I've tried the same multiprocessing code with more simple definitions like resizing images on different directories the multiprocessing works well but when I tried it with the code seen below, it runs but it does not give any output or raise any errors and I was wondering what could be the reason for this.

I was also wondering how could I use multiprocessing with this code, maybe inheritance is the problem?

class Skeleton:
    def __init__(self, path, **kwargs):

        if type(path) is str:
                self.path = path
                self.inputStack = loadStack(self.path).astype(bool)
        if kwargs != {}:
            aspectRatio = kwargs["aspectRatio"]
            self.inputStack = ndimage.interpolation.zoom(self.inputStack, zoom=aspectRatio, order=2, 
            prefilter=False)

    def setThinningOutput(self, mode="reflect"):
        # Thinning output
        self.skeletonStack = get_thinned(self.inputStack, mode)
    def setNetworkGraph(self, findSkeleton=False):
        # Network graph of the crowded region removed output
        self.skeletonStack = self.inputStack
        self.graph = get_networkx_graph_from_array(self.skeletonStack)

    def setPrunedSkeletonOutput(self):
        # Prune unnecessary segments in crowded regions removed skeleton
        self.setNetworkGraph(findSkeleton=True)
        self.outputStack = pr.getPrunedSkeleton(self.skeletonStack, self.graph)
        saveStack(self.outputStack, self.path + "pruned/")


class Trabeculae (Skeleton):
    pass

def TrabeculaeY (path):
     path_mrb01_square = Trabeculae(path)
     path_mrb01_square.setPrunedSkeletonOutput()

if __name__=='__main__':

    path1 = (r' ')
    path2 = (r' ')
    path3 = (r' ')
    the_list =[]
    the_list.append(path1)
    the_list.append(path2)
    the_list.append(path3)
    for i in range (0,len(the_list)):
        p1 = multiprocessing.Process(target=TrabeculaeY, args=(the_list[i],))
        p1.start()
        p1.join()

Answer 1

Inheritance is not a problem for multiprocessing.

You must not join() the processes inside the loop. It means that the loop waits until p1 finished doing its work, before it continues with the next one.

Instead, start all processes in a loop, then wait for all processes in a second loop like this:

if __name__=='__main__':

    path1 = (r' ')
    path2 = (r' ')
    path3 = (r' ')
    the_list =[]
    the_list.append(path1)
    the_list.append(path2)
    the_list.append(path3)
    started_processes = []
    for i in range (0,len(the_list)):
        p1 = multiprocessing.Process(target=TrabeculaeY, args=(the_list[i],))
        p1.start()
        started_processes.append(p1)
    for p in started_processes:
        p.join()  

Full code I used for testing:

import multiprocessing


class Skeleton:
    def __init__(self, path, **kwargs):
        self.path = path
        pass

    def setThinningOutput(self, mode="reflect"):
        pass

    def setNetworkGraph(self, findSkeleton=False):
        pass

    def setPrunedSkeletonOutput(self):
        print(self.path)


class Trabeculae(Skeleton):
    pass


def TrabeculaeY(path:str):
    path_mrb01_square = Trabeculae(path)
    path_mrb01_square.setPrunedSkeletonOutput()


if __name__ == '__main__':
    the_list = [r'1', r'2', r'3']
    started_processes = []
    for path in the_list:
        process = multiprocessing.Process(target=TrabeculaeY, args=path)
        process.start()
        started_processes.append(process)

    for process in started_processes:
        process.join()