Closed russelljjarvis closed 6 years ago
This issue is resolved but not in the most 'maintainable', way. I welcome suggestions on a more transparent fix.
Lines 491 to 499 now read:
##
# init_backend is the most reliable way to purge existing NEURON simulations.
# however we don't want to purge the model attributes, we only want to purge
# the NEURON model code.
# store the model attributes, in a temp buffer such that they persist throughout the model reinitialization.
##
temp_attrs = copy.copy(self.model.attrs)
self.init_backend()
self.set_attrs(**temp_attrs)
Fixed with self.init_backend()
, which purges the information associated with a previous simulation.
unit tests of neuronunit tests fail with serial neuron because the latest backends no longer refreshes neuron models with current injection.
serial rheobase unit test not working with serial neuron as model, /model/backends.py is no longer
refreshing the neuron model between runs as you would expect
adding in the line
self.init_backend()
at 515 of backends.py eg:works.