Handling of Pascal and Snake name conventions for attributes is a nightmare

Related to MR !161 (merged).

I find that handling both conventions is very complex and there are still several loopholes in the code (in delAttr, in getAttrType, ...). Plus, it is not straightforward for the users, even less with namespaces... In C++, you would define an attribute like Mem.Data.B and it should become mem.data.b in Python. But worse, you cannot define an attribute mem.data.B in C++ with the current proposed implementation because of these constrains, which is quite unexpected for the user!

Why not choose the same convention for attribute naming for both Python and C++?

added Discussion 💬 label

As the Python convention is snake_case and Python is the prefered language for most users, I would go with snake_case

I agree we going with the snake_case, which is much more common especially for Python users. Plus it would only require to change the C++ code, which has had a much lower exposure than Python binding.

Actually, it would require to change the EnumString of each operator's hpp file as well as removing both snake_case and PascalCase assertions in StaticAttribute.hpp and DynamicAttribute.hpp

Done in !161 (merged) (ready to be merged).

I will change this in MR !161 (merged).

closed