Assume the following code:

```python
from multiprocessing.pool import ThreadPool
class A(object):
    def __init__(self):
        self.pool = ThreadPool()
    def __del__(self):
        self.pool.close()
        self.pool.join()
a = A()
print(a)
```

The code snippet above hangs forever on Python 3.8+ (works ok on Python 3.7 and earlier). An example output where I've added some extra prints on to the thread joins:

```
<__main__.A object at 0x1104d6070>
join thread <Thread(Thread-13, started daemon 123145699672064)> None
done
join thread <Thread(Thread-14, stopped daemon 123145716461568)> None
done
join thread <Thread(Thread-15, stopped daemon 123145733251072)> None
done
join thread <Thread(Thread-13, stopped daemon 123145699672064)> None
done
join thread <Thread(Thread-14, stopped daemon 123145716461568)> None
done
join thread <Thread(Thread-15, stopped daemon 123145733251072)> None
done
join thread <DummyProcess(Thread-1, stopped daemon 123145498198016)> None
done
join thread <DummyProcess(Thread-2, stopped daemon 123145514987520)> None
done
join thread <DummyProcess(Thread-3, stopped daemon 123145531777024)> None
done
join thread <DummyProcess(Thread-4, stopped daemon 123145548566528)> None
done
join thread <DummyProcess(Thread-5, started daemon 123145565356032)> None
```

I've tested on MacOs, but could replicate on Linux too within the CI.