Symbols not visible to child processes, breaking multiprocessing

[akshayka]

Describe the bug

Attempting to execute a notebook-defined function in another process with multiprocessing fails, because Python's multiprocessing bootstrap can't find the function:

Process SpawnProcess-1:1:
Traceback (most recent call last):
  File "/Users/aagrawal/.local/share/uv/python/cpython-3.14-macos-aarch64-none/lib/python3.14/multiprocessing/process.py", line 320, in _bootstrap
    self.run()
    ~~~~~~~~^^
  File "/Users/aagrawal/.local/share/uv/python/cpython-3.14-macos-aarch64-none/lib/python3.14/multiprocessing/process.py", line 108, in run
    self._target(*self._args, **self._kwargs)
    ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/aagrawal/.local/share/uv/python/cpython-3.14-macos-aarch64-none/lib/python3.14/concurrent/futures/process.py", line 242, in _process_worker
    call_item = call_queue.get(block=True)
  File "/Users/aagrawal/.local/share/uv/python/cpython-3.14-macos-aarch64-none/lib/python3.14/multiprocessing/queues.py", line 120, in get
    return _ForkingPickler.loads(res)
           ~~~~~~~~~~~~~~~~~~~~~^^^^^
AttributeError: module '__mp_main__' has no attribute 'compute_square'

import multiprocessing
import time
import matplotlib.pyplot as plt
from concurrent.futures import ProcessPoolExecutor
import numpy as np

# Embarrassingly parallel workload: calculate square of numbers
def compute_square(n):
    # Simulate some work with a small delay
    time.sleep(0.001)
    return n * n

def run_parallel_workload(num_workers, data_size=100):
    """Run parallel workload with specified number of workers"""
    data = list(range(data_size))
    
    start_time = time.time()
    with ProcessPoolExecutor(max_workers=num_workers) as executor:
        results = list(executor.map(compute_square, data))
    end_time = time.time()
    
    return end_time - start_time

# Test with varying numbers of cores
max_cores = multiprocessing.cpu_count()
core_counts = list(range(1, max_cores + 1))[:4]
times = []

for num_cores in core_counts:
    print("Number of cores")
    elapsed_time = run_parallel_workload(num_cores)
    times.append(elapsed_time)
    print(f"Cores: {num_cores}, Time: {elapsed_time:.2f}s")

# Plot time vs number of cores
plt.figure(figsize=(10, 6))
plt.scatter(core_counts, times, color='blue', s=50)
plt.plot(core_counts, times, color='lightblue', linestyle='-', alpha=0.7)
plt.xlabel('Number of Cores')
plt.ylabel('Execution Time (seconds)')
plt.title('Execution Time vs Number of Cores for Parallel Workload')
plt.grid(True, alpha=0.3)
plt.xticks(core_counts)
plt.gca()

Will you submit a PR?

• Yes

Environment

Details

Replace this line with the output of marimo env. Leave the backticks in place.

Code to reproduce

No response