Goal:
Pass the ExperimentConfig class object to a PsychoPy subprocess
Purpose:
Providing a Psychopy subprocess with access to the state of the parent process’ ExperimentConfig class allowing for access to experimental parameters for data output, etc.
Current Progress:
Psychopy QProcess
from PyQt6.QtCore import QProcess
from typing import TYPE_CHECKING
if TYPE_CHECKING:
from mesofield.config import ExperimentConfig
def launch(config: 'ExperimentConfig', parent=None):
"""Launches a PsychoPy experiment as a subprocess with the given ExperimentConfig parameters."""
# Build the command arguments
args = [
"C:\\Program Files\\PsychoPy\\python.exe",
f'{config.psychopy_path}',
f'{config.subject}',
f'{config.session}',
f'{config.save_dir}',
f'{config.num_trials}',
f'{config.psychopy_save_path}'
]
# Create and start the QProcess
psychopy_process = QProcess(parent)
psychopy_process.finished.connect(parent._handle_process_finished)
psychopy_process.start(args[0], args[1:])
return psychopy_processExperimentConfig Head
class ExperimentConfig:
def __init__(self, path: str):
self._parameters: dict = {}
self._json_file_path = ''
self._output_path = ''
self._save_dir = ''
self.hardware = HardwareManager(path)
self.notes: list = []Notes
Using Python’s pickle versus dill for serializing the object
dillhandles more complex objects, such as the ExperimentConfig
Some attributes cannot be pickled, such as pymmcore-plus CMMCore objects.
- Pickling uses the object’s
__getstate__and__setstate__methods; you can override them and ignore the fields you want. source - https://docs.python.org/3/library/pickle.html#handling-stateful-objects
Potential Solutions:
https://oegedijk.github.io/blog/pickle/dill/python/2020/11/10/serializing-dill-references.html
Attempts:
def launch(config: 'ExperimentConfig', parent=None):
"""Launches a PsychoPy experiment as a subprocess with the given ExperimentConfig parameters."""
serialized_config = dill.dumps(config, byref=True)
#dill handles complex objects, pickle does not
b64_serialized_config = base64.b64encode(serialized_config).decode('ascii')
# Qprocesses require strings
# Create and start the QProcess
psychopy_process = QProcess(parent)
psychopy_process.readyReadStandardOutput.connect(lambda: print(psychopy_process.readAllStandardOutput().data().decode()))
psychopy_process.readyReadStandardError.connect(lambda: print(psychopy_process.readAllStandardError().data().decode()))
psychopy_process.start(r"C:/Program Files/PsychoPy/python.exe", [r"D:\Experiment Types\Checkerboard Experiment\CheckerBar_vis_build-v0.8.py", b64_serialized_config])