troehling

Right now the example is more or less the opposite of minimal but I am trying to create a minimal example right now. To give a very brief overview: I...

I have observed that `multiprocessing.set_start_method('loky')` does not seem to have any impact as `multiprocessing.get_start_method()` still returns `fork`. Do I have to somehow initialise loky before? The main application is based...

I did something very similar: ``` import multiprocessing from PyQt5.QtWidgets import * import sklearn # Library that imports joblib and loky in my case def main(): print(multiprocessing.get_start_method()) app = QApplication([])...

I do not exactly know what changed but when i call the program directly I can now see a different error message in my programm: ``` Traceback (most recent call...

The issue seems to be definitely on my side. There is a library that I use (https://github.com/CFEL-CMI/pymepix) that seems to be the cause for the issues I have. No matter...

This minimal reproducible example is working on my system: ``` import multiprocessing import sklearn _proc_log_queue = multiprocessing.Queue() def main(): print(multiprocessing.get_start_method()) if __name__ == "__main__": multiprocessing.set_start_method('loky') main() ``` Then I do...

I tried and sadly changing the order of the imports does not change anything. The error is still the same.

Ohh yes at least in my example I can do it this way. Still the error I have already had with `force=True` exists if I shift the `set_start_method`-call above the...

I was finally able to create a minimal example for this issue: ```python import multiprocessing import loky multiprocessing.set_start_method('loky') import ctypes import time from multiprocessing.sharedctypes import Value # _proc_log_queue = multiprocessing.Queue()...