ThreadPoolExecutor marks interpreter as shutdown before non daemon threads exit

[aaix]

Bug report

Bug description:

When the main thread exits, concurrent.futures.ThreadPoolExecutor mistakenly marks the interpreter as shutdown, regardless of if there are non daemon threads still executing, this causes any work submitted to the executor to fail with RuntimeError: cannot schedule new futures after interpreter shutdown.

PoC:

import time
import threading

from concurrent.futures import ThreadPoolExecutor

def work():
    print("Starting work")
    time.sleep(1)
    print("Work complete")

def worker_thread():
    # sleep a little to give the main thread time to exit
    time.sleep(1)

    executor = ThreadPoolExecutor()
    
    future = executor.submit(work)
    

def main():
    thread = threading.Thread(target=worker_thread)
    thread.start()

    # main thread exits now

if __name__ == "__main__":
    main()

Result

Exception in thread Thread-1 (worker_thread):
Traceback (most recent call last):
  File "/usr/lib/python3.14/threading.py", line 1081, in _bootstrap_inner
    self._context.run(self.run)
    ~~~~~~~~~~~~~~~~~^^^^^^^^^^
  File "/usr/lib/python3.14/threading.py", line 1023, in run
    self._target(*self._args, **self._kwargs)
    ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/*****/*****/implementation/./bug.py", line 18, in worker_thread
    future = executor.submit(work)
  File "/usr/lib/python3.14/concurrent/futures/thread.py", line 207, in submit
    raise RuntimeError('cannot schedule new futures after '
                       'interpreter shutdown')
RuntimeError: cannot schedule new futures after interpreter shutdown

Interpreter is marked as shutdown here:

cpython/Lib/concurrent/futures/thread.py

Lines 23 to 37 in f46a17b

	def _python_exit():
	global _shutdown
	with _global_shutdown_lock:
	_shutdown = True
	items = list(_threads_queues.items())
	for t, q in items:
	q.put(None)
	for t, q in items:
	t.join()
	# Register for `_python_exit()` to be called just before joining all
	# non-daemon threads. This is used instead of `atexit.register()` for
	# compatibility with subinterpreters, which no longer support daemon threads.
	# See bpo-39812 for context.
	threading._register_atexit(_python_exit)

Expected Result:
All non daemon threads should be able to use ThreadPoolExecutor until they exit.

example was tested on 3.14.0

cc @ZeroIntensity

CPython versions tested on:

3.14

Operating systems tested on:

Linux

[ZeroIntensity]

In 3.15, we join non-daemon threads in a loop, so we should be able to safely create new futures even after threading._shutdown has been called for the first time. I'll be working on this.