[Bug] Simulation tools do not save results to consistent locations

[jonrkarr]

Describe the bug
run_lems_with_jneuroml and run_lems_with_jneuroml_neuron do not save results to consistent locations. The jNeuroML function saves results to exec_in_dir directory. The NEURON function seems to ignore this argument. Navigating this divergence requires extra unexpected work by users.

To Reproduce
Run run_lems_with_jneuroml and run_lems_with_jneuroml_neuron with the same arguments.

Expected behavior
Running run_lems_with_jneuroml and run_lems_with_jneuroml_neuron with the same arguments should produce results in the same location in the same format.

Observed behavior
The functions save results to different directories. This requires the user to be aware of this, use the arguments differently for each tool, and use if statements to manage the use of these methods.

System information:

• OS: [e.g. iOS]: Ubuntu 20
• Python version (python --version): 3.9.5
• Version of tools in use (pynml --help): pyNeuroML 0.5.18

Additional context
Add any other context about the problem here.

[Ragzz1995]

Hi @jonrkarr ,
I ran both the methods with exec_in_dir="./temp", in both cases as long as the files required for execution was placed inside the ./temp directory, they both produced the same results.

pynml -version : 0.5.20
python 3.6.9

Maybe I can check with specific versions you have mentioned next

[jonrkarr]

I can try again with 0.5.20 and copy more specific instructions to reproduce the error here

[sanjayankur31]

I can reproduce this. The issue here is that irrespective of what the exec_in_dir folder is set to, the NEURON scripts are always generated by jnml in the same folder as the LEMS script, and then run there. So, the data etc. is also stored relative to the directory where the LEMS script is run. Making jnml put the generated scripts elsewhere can be done, but it'll require some changes to the logic in jNeuroML.