Loading import3d before loading modfile create an execption (only in Neruon 7.6)

[orena1]

It seems that only in Neuron 7.6 loading import3d.hoc before loading the modfiles will create an exception:
Code for running the example

from neuron import h, gui

h.load_file("import3d.hoc")  ## THIS LINE
h.nrn_load_dll("x86_64_temp/.libs/libnrnmech.so")


MODEL_FILE = 'cNAC187_L23_LBC_3_C230300D1_new_new_fit.hoc'
model_obj_name = 'cNAC187_L23_LBC_3_C230300D1_new_new_fit'

h.load_file(MODEL_FILE)

cell = h.cNAC187_L23_LBC_3_C230300D1_new_new_fit(1,'C230300D1.asc')

Running the code above will create this exception in Neuron 7.6

bad stack access: expecting (double); really (Object *)
NEURON: interpreter stack type error
 near line 0
 {forall all.append}
                    ^
        Import3d_GUI[0].instantiate(Cell[0])
      Cell[0].AddMorph("C230300D1.asc")
    Cell[0].init(1, "C230300D1.asc")
  cNAC187_L23_LBC_3_C230300D1_new_new_fit[0].init(1, "C230300D1.asc")
oc_restore_code tobj_count=2 should be 1
Traceback (most recent call last):
  File "/_____/nrn/lib/python/neuron/__init__.py", line 494, in _declare_contour
    center_vec = secobj.contourcenter(secobj.raw.getrow(0), secobj.raw.getrow(1), secobj.raw.getrow(2))
RuntimeError: hoc error

If you suspect this is an IPython bug, please report it at:
    https://github.com/ipython/ipython/issues
or send an email to the mailing list at [email protected]

You can print a more detailed traceback right now with "%tb", or use "%debug"
to interactively debug it.

Extra-detailed tracebacks for bug-reporting purposes can be enabled via:
    %config Application.verbose_crash=True

NEURON: PyObject method call failed: _declare_contour
 near line 0
 ^
        Import3d_GUI[0].instantiate(Cell[0])
      Cell[0].AddMorph("C230300D1.asc")
    Cell[0].init(1, "C230300D1.asc")
  cNAC187_L23_LBC_3_C230300D1_new_new_fit[0].init(1, "C230300D1.asc")
oc_restore_code tobj_count=2 should be 0
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
/_____/one_file_test.py in <module>()
     10 h.load_file(MODEL_FILE)
     11 
---> 12 cell = h.cNAC187_L23_LBC_3_C230300D1_new_new_fit(1,'C230300D1.asc')
     13 
     14 

RuntimeError: hoc error

but it works well in Neuron 7.4. The main problem is the line marked with a comment (EDIT: which is sometimes needed)

Might also be related to this issue: CINPLA/NeuroCNN#2

In order to test the attached code, you need to compile the mod files and then rename x86_64 to x86_64_temp

[nrnhines]

I'd like to work on this problem but don't have C230300D1.asc or cNAC187_L23_LBC_3_C230300D1_new_new_fit.hoc or probably anything those depend on.

[orena1]

All the files are in the attached zip, in the first line of the issue. Also here:
Code for running the example

[nrnhines]

This bug is fixed by changeset f837b5c and was introduced by changeset 31493b9.
The effect of the bug was that if new POINT_PROCESSES were registered after user defined
hoc templates were created, then the use of the latter could in some circumstances be thought to have
returned a bad return type code when in fact, user methods know nothing about return type codes (for python boolean, int, and float).