Is it possible to abort execution on FP exception or NaN voltage values?

[pramodk]

This is question form one of the BBP users. Copying text from internal ticket:

When running some Gap Junction debugging I noticed that in some simulation,
[due to an error which I am trying to understand ( I will open a different
ticket on the error itself)], the voltage of some neurons goes to inf and then
to  NaNs. The simulation continues to run normally and finishes normally. 
This is a major problem, as in many simulations we do not record the voltage,
or even when we do record it we do not analyze the voltage of all neurons to
check that there are no NaNs. We could be analyzing the results of simulations
in which we just assume that neurons just stopped firing (but actually their
voltage is NaN).
I do see some error warning in the stderr, but maybe it would be better to also
get something in the stdout? or maybe the simulation should just break? as
there is no use in NaN values

Is there any API already provided in NEURON?

[nrnhines]

In nrn/src/hoc.c #define NRN_FLOAT_EXCEPTION 1 if linux and HAVE_FENV_H.
Were you runnng on mac. I don't know how to activate that functionaliy in that circumstance. It causes hoc_execerror in the
circumstances #define FEEXCEPT (FE_DIVBYZERO | FE_INVALID | FE_OVERFLOW )

[pramodk]

Ok, thanks Michael.
I was wondering if there is python/hoc API to enable this at runtime. Not strictly required but I was wondering if other users have asked something similar in the past.

[nrnhines]

That would be a very nice thing to have and interact very nicely with #785