As is the backends file is still probably calling jNeuroML unnecessarily. about neuronunit HOT 3 CLOSED

@russelljjarvis
This and some of the code in the few dozen lines above it are still model specific (referring to LEMS_2007One and such). Can this section be made model-independent by having it determine the name of that directory from some other attribute of the model (like the orig_lems_file_path, for example).

from neuronunit.

The problem in this specific case is that the path self.orig_lems_file_path looks like:

/neuronunit/tests/NeuroML2/LEMS_2007One.xml/x86_64

but we actually just want:

/neuronunit/tests/NeuroML2/x86_64

Maybe instead of using the commands stated below, I can find a command that simply takes an absolute file path and returns just the directory that contains that file, then /x86_64 should be appended to this path.

This is generalizable for different NEURON models generated from NeuroML, xml file paths, but checking for x86_64 obviously won't be helpful for pyNN, NEST, or GENESIS etc.

   sort_file_path, _ = self.orig_lems_file_path.split("/LEMS_2007One")
   sort_file_path, _ = os.path.splitext(sort_file_path)

Alternatively we could check for the x86_64 subdirectory by recursively searching, but I think the above option is better. I

/neuronunit/tests/NeuroML2/LEMS_2007One.xml/x86_64

from neuronunit.

Just ignore this thread. I think some of my confusion comes from a git merge I completed. There is old code that you wrote that used to find the directory that contains the file, but it got buried somehow. I will check older commits.

from neuronunit.