Make thread-safe

[mattwthompson]

Most of the setup happens in temporary directories, but something about parsing the output does not¹. As a result, multiple calls on separate threads can easily choke. This example shows two ways:

• Trying to load an output file from a different packing (appropriately!) triggers a WrongShapeError
• Sometimes, the output file packmol_output.pdb does not exist when it's trying to be loaded

from openff.packmol import pack_box
from openff.toolkit import Molecule, Quantity

import threading

water = Molecule.from_smiles("O")

threads = []
for i in range(3):
    t = threading.Thread(
        target=pack_box,
        args=(
            [water],
            [100 + i],
            None,
            Quantity(0.2, "nanometer"),
            None,
            Quantity(0.5, "g/mL"),
        ),
    )
    threads.append(t)
    t.start()

for t in threads:
    t.join()

Out:

Exception in thread Thread-1 (pack_box):
Exception in thread Thread-3 (pack_box):
Traceback (most recent call last):
Traceback (most recent call last):
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1075, in _bootstrap_inner
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1075, in _bootstrap_inner
Exception in thread Thread-2 (pack_box):
Traceback (most recent call last):
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/packmol/_packmol.py", line 853, in _load_positions
    for line in open(output_file_path).readlines()
                ^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: 'packmol_output.pdb'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1075, in _bootstrap_inner
    self.run()
    self.run()
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1012, in run
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1012, in run
    self.run()
    self._target(*self._args, **self._kwargs)
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/threading.py", line 1012, in run
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/utilities/utilities.py", line 79, in wrapper
    self._target(*self._args, **self._kwargs)
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/utilities/utilities.py", line 79, in wrapper
    self._target(*self._args, **self._kwargs)
    return function(*args, **kwargs)
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/utilities/utilities.py", line 79, in wrapper
           ^^^^^^^^^^^^^^^^^^^^^^^^^
    return function(*args, **kwargs)
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/packmol/_packmol.py", line 810, in pack_box
           ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/packmol/_packmol.py", line 810, in pack_box
    topology.set_positions(Quantity(positions[n_solute_atoms:], "angstrom"))
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/toolkit/topology/topology.py", line 2109, in set_positions
    topology.set_positions(Quantity(positions[n_solute_atoms:], "angstrom"))
    return function(*args, **kwargs)
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/toolkit/topology/topology.py", line 2109, in set_positions
           ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/packmol/_packmol.py", line 795, in pack_box
    positions = _load_positions(output_file_path)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/mattthompson/software/dimsim/.pixi/envs/dev/lib/python3.12/site-packages/openff/packmol/_packmol.py", line 859, in _load_positions
    raise PACKMOLRuntimeError(
openff.packmol.exceptions.PACKMOLRuntimeError: PACKMOL output could not be parsed by a native coordinate parser, please raise an issue with code reproducing this error.
    raise WrongShapeError(f"Array has shape {array.shape} but should have shape {self.n_atoms, 3}")
    raise WrongShapeError(f"Array has shape {array.shape} but should have shape {self.n_atoms, 3}")
openff.toolkit.utils.exceptions.WrongShapeError: Array has shape (102, 3) but should have shape (300, 3)
openff.toolkit.utils.exceptions.WrongShapeError: Array has shape (102, 3) but should have shape (306, 3)

Footnotes

1. Or each call ends up in the same temporary directory, which I think is unlikely ↩

[mattwthompson]

I realized recently that this problem boils down to making each function call unique to itself and UUIDs should be a simple fix. Creating UUIDs is at least thread-safe:

In [20]: import threading

In [21]: import uuid

In [22]: def print_uuid():
    ...:     print(uuid.uuid4())
    ...:

In [23]: threads = []

In [24]: for i in range(5):
    ...:     t = threading.Thread(
    ...:         target=print_uuid,
    ...:     )
    ...:     threads.append(t)
    ...:     t.start()
    ...: for t in threads:
    ...:     t.join()
    ...:
b705f6b0-4b2c-4c2f-8106-1f6e56346c61
9233cb5d-d0ae-4380-a7ae-7b1dbe90c222
0b95339b-9776-488c-ad60-71da04f2c411
eefb83ac-4a70-45aa-a498-d10ca2801545
153da39d-01c8-434f-9732-dce4e90ffdae

so I can just throw this or something like it a path(s) to make each call unique enough. Not sure how I or Claude didn't think of this before?