Importing from other tasks

[jonahpearl]

Hi all — I'm interested in using a little bit of pythonic abstraction / inheritance in my task files. Specifically, I have written a task file for initial behavioral shaping (working towards a 2AFC task) that implements the core skeleton of the task, and has lots of little details that help the mice perform the task well (i.e. accounting for their reaction times before counting certain things as errors). To change the task I have into the next task I want to use, all I really need to do is 1) import everything from the working task, then 2) add a few actions in a few states by using dummy functions in the first task and re-defining them in the new one. I tried to do 1) by something like from my_first_task import * at the top of the new task. The issue is that when I try to upload the task file to the pyBoard, I get the error

Error: Unable to setup state machine.

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "task_file.py", line 2, in <module>
ImportError: no module named 'my_first_task'

Error setting up state machine.

I see that when I do something like import pyControl.utility as pc at the top of a task, it's coming from the pyControl "framework" that is uploaded onto the pyBoard once (the code that lives in source/pyControl). And obviously that's not where my task file is. Is there a way to get the uploader / compiler to look for imports in the local files before uploading the task file to the pyBoard?

Thanks!

PS: I also tried from .my_first_task import * and got an import error, "cant perform relative import", I assume for the same reason.

[ThomasAkam]

Hi @jonahpearl,

pyControl does not currently support splitting task code across multiple files. When you upload a task to the pyboard, the selected task file is copied across to the pyboard and renamed task_file.py on the pyboard filesystem (the code that implements this is Pycboard.setup_state_machine). Space on the pyboard filesystem is limited so it is not possible to duplicate the entire tasks folder on there, and allowing task code to be split across multiple files would substantially complicate the logic for both setting up tasks and saving versioned task code to the data folder for reproducibily.

One possible workaround for reusing code across multiple tasks is the functionallity for automatically copying device driver files needed for a task from the devices folder on the computer to the pyboard. This functionallity, implemented in Pycboard.transfer_device_files, scans the task file to identify any classes defined in the modules in the devices folder that are used in the task, then copies the required modules to the pyboard when the task is uploaded. You could create a new module in the devices folder on the computer which defines one or more classes that you want to use across multiple tasks, then use it in your task file with from devices import My_device_class. The relevent file will be automatically copied to the devices folder on the pyboard when you upload the task. This would allow you to resuse classes across different task files. This approach does have some disadvantages though: i) It would make it harder to version control you task code as it would be split across two seperate folders. ii) You would lose the reproducibility benefit of pycontrol automatically saving versioned task code to the data folder, as task logic would now also be in the custom device class that would not be saved with the data.