When backends NEURON lives in a new file, new model objects can't be minted.

[russelljjarvis]

reduced.py needs Backend NEURON to reside in the same file. One strategy might involve just making code climate ignore backends.py file. As shortening files may not always make code more maintainable.

~/neuronunit/neuronunit/optimization/nsga_parallel.py in dtc_to_rheo(dtc)
     39     dtc.model_path = get_neab.LEMS_MODEL_PATH
     40     dtc.LEMS_MODEL_PATH = get_neab.LEMS_MODEL_PATH
---> 41     model = ReducedModel(dtc.LEMS_MODEL_PATH,name=str('vanilla'),backend='NEURON')
     42     model.set_attrs(dtc.attrs)
     43     model.rheobase = None
~/neuronunit/neuronunit/models/reduced.py in __init__(self, LEMS_file_path, name, backend, attrs)
     22         """
     23         super(ReducedModel,self).__init__(LEMS_file_path,name=name,
---> 24                                           backend=backend,attrs=attrs)
     25         self.run_number = 0
     26         self.tstop = None

~/neuronunit/neuronunit/models/__init__.py in __init__(self, LEMS_file_path, name, backend, attrs)
     44         self.rerun = True # Needs to be rerun since it hasn't been run yet!
     45         self.unpicklable = []
---> 46         self.set_backend(backend)
     47 
     48     def get_backend(self):

~/neuronunit/neuronunit/models/__init__.py in set_backend(self, backend)
     81         else:
     82             raise Exception("Backend %s not found in backends.py" \
---> 83                             % name)
     84         self._backend.model = self
     85         self._backend.init_backend(*args, **kwargs)

Exception: Backend NEURON not found in backends.py

At the bottom of neuronunit/models/backends.py I have:

from neuronunit.models import section_extension
from neuronunit.models import backendsNEURON

possibly not the standard pattern for expanding out a module.

[rgerkin]

set_backend does a check for which models live inside backends.py. We can fix this by having it instead search inside backends/__init__.py, where we will have all the backends in the various other files (including backends/neuron.py) imported.

[russelljjarvis]

How complicated would these expressions for searching look? If they are simple expressions than I agree.

[rgerkin]

Simple. I'll take care of it on Monday.

[russelljjarvis]

Yes, actually come to think of it we need it for PyNN and Brian, you want backends to grow with increasing neuronunit support. We can't make the file ~1,000s of line long.

[rgerkin]

See implementation in 294c0db9b8adb008ec48bcbe6c9737130302d970

[russelljjarvis]

Fixed.