[BUG] AssertionError: daemonic processes are not allowed to have children

[lmBored]

Describe the bug

When using benchmarl and setting parallel_collection to True, I get this error

To Reproduce

Set parallel_collection to True when creating experiment with benchmarl, environment created with vizdoom

Process _ProcessNoWarn-11:
Traceback (most recent call last):
  File "/lib/python3.12/multiprocessing/process.py", line 314, in _bootstrap
    self.run()
  File ".venv/lib/python3.12/site-packages/torchrl/_utils.py", line 789, in run
    return mp.Process.run(self, *args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/python3.12/multiprocessing/process.py", line 108, in run
    self._target(*self._args, **self._kwargs)
  File ".venv/lib/python3.12/site-packages/torchrl/envs/batched_envs.py", line 2373, in _run_worker_pipe_shared_mem
    env = env_fun(**env_fun_kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "ViZDoom/.venv/lib/python3.12/site-packages/torchrl/envs/env_creator.py", line 201, in __call__
    env = self.create_env_fn(**kwargs)
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "ViZDoom/.venv/lib/python3.12/site-packages/torchrl/data/utils.py", line 260, in __call__
    return self.fn(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "ViZDoom/examples/python/train_multi_agent_extensive.py", line 188, in _make
    pz_env = make(
             ^^^^^
  File "ViZDoom/pettingzoo_wrapper/__init__.py", line 44, in make
    env = VizdoomParallelEnv(
          ^^^^^^^^^^^^^^^^^^^
  File "ViZDoom/pettingzoo_wrapper/base_pettingzoo_env.py", line 341, in __init__
    p.start()
  File "/lib/python3.12/multiprocessing/process.py", line 118, in start
    assert not _current_process._config.get('daemon'), \
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: daemonic processes are not allowed to have children

Expected behavior

Returns no errors

Fix proposal

Set process.daemon = False in torchrl/envs/batched_envs.py

Screenshots

The log output above is also what I want to attach in the screenshot

System info

Describe the characteristic of your environment:

• Installed with uv
• Python version: 3.12.3
• Torchrl: 0.8.1
• Benchmarl: Latest version from github
• Torch: 2.8.0
• Vizdoom: 1.2.4
• numpy: 2.2.6
• Platform: [Clang 17.0.6 ] darwin

Additional context

I was able to fix it myself, but report as an issue to verify if this is case-specific or torchrl issue.

Reason and Possible fixes

So basically in my MARL setup, Benchmarl worker is trying to create child processes, but Python enforces only non-daemonic processes can do that, thus causing the issue.

Checklist

• [ x ] I have checked that there is no similar issue in the repo (required)
• [ x ] I have read the documentation (required)
• [ x ] I have provided a minimal working example to reproduce the bug (required)