issues with "New python bindings using pybind11"

[hrani]

moose.le('/') works but

1. moose.le('/r') which is not defined or created it gives
   moose.element: /r does not exist!
   Segmentation fault (core dumped)

2. isinstance of super class gives False
   n = moose.Neutral('/n')
   c = moose.CubeMesh('/n/cube')
   isinstance(c,moose.ChemCompt)
   returns False should be True as CubeMesh is inherited from ChemCompt which was the True before new python bindings

[dilawar]

The PR #424 fixes the seg-fault issue with moose.le.

isinstance does not work as before in new binding. Have a look at #410 (comment) which summaries the changes.

Also see #404 (comment) and https://labnotes.ncbs.res.in/comment/2322#comment-2322 and follow-up discussion on issue #404 on why isA and not isinstance is not a good way to check the base class.

The current recommendation is to use isA as you have also mentioned in #404 (comment) that you are aware of https://labnotes.ncbs.res.in/comment/2322#comment-2322 (though not sure though what you meant here).

Nonetheless, to summarise, following is the behavious of new bindings:

n = moose.Neutral('/n')
c = moose.CubeMesh('/n/cube')
print(isinstance(c, moose.CubeMesh))        # True
print(isinstance(c, moose.ChemCompt))     # False 
print(c.isA['ChemCompt'])                       # True

The line isinstance(c, moose.ChemCompt) is no longer True in new bindings. You must use c.isA['ChemCompt'] (notice the square bracket, though () also works).

[hrani]

Yes my mistake, I had changed this in the moose-gui, will do for SBML reader and writer also.