Saving Intermediate results

[jbuencuerpo]

There is some how to save intermediate results? It is not posible to pickle the Structure class, which could be beneficial for long simulations. I have tried for example after the MJ_solar_cell_tutorial.py :

import pickle
with open('solcore_test.p', 'wb') as ffile:
    pickle.dump(my_solar_cell, ffile, protocol=2)

---

AttributeError Traceback (most recent call last)
in ()
1 with open('solcore_test.p', 'wb') as ffile:
2 #pickle()
----> 3 pickle.dump(my_solar_cell, ffile)
AttributeError: Can't pickle local object 'MaterialSystem.sopra_material..SpecificMaterial'

Another option is using dill, which typically is smarter, but I am afraid not in this case, fails with the same class. There is another way to solve this?
Thanks.

[dalonsoa]

Unfortunately, the material system cannot be pickled. This is a long lost battle that I’m afraid can only be solved by refactoring the material system altogether. As it is now, it is useful but pretty obscure, with classes inside classes and funciona made available outside, as well as information retrieve from disk on the fly, as needed. It works and it has been useful for many years, but certainly not that helpful when it comes to expand it, maintain it or simply, as you want, pickle the objects.

What is exactly the part that is preventing a material to be saved, I’m not sure. Here is a discussion where the creator of Dill gives some tips of what might be going wrong.

https://stackoverflow.com/questions/29922373/when-can-a-python-object-be-pickled

Improving the material system might be tagged as a very useful enhancement, to be honest.

Cheers,
Diego

_{Sent with GitHawk}