Question: Why is FortranModule a Singleton?

[smiet]

I see that the FortranModule class is implemented with __metaclass__=Singleton, such that only one instance can be created.

This creates some issues, if on the python side I want to do things like mymodule1 = mycode.mycode_f90wrapped and mymodule2 = mycode.mycode_f90wrapped.

We are using the FORTRAN module in optimization, and it would be nice to have several instances in memory at the same time, for example to: compare differences between initial and final state; plot the properties of two different objects in one script; run the fortran code with slightly different initial parameters.

The class docstring says

class FortranModule(object):
    """
    Baseclass for Fortran modules

    Metaclass is set to Singleton, so only one instance of each subclass of
    FortranModule can be created.
    """"

Before I go barking at the wrong dead horse, and trying to modify the code to allow for several instances, I wanted to ask what lead to the design choice to implement the module as a Singleton, and if there are insurmountable obstacles to allowing for several top-level modules of f90wrapped modules in python.

[jameskermode]

Fortran modules really are singletons, that is all module data is stored only once in memory. This is why I chose to mirror this in Python. If you want several different instances in memory, you will need to use a Fortran derived type, not a module. There is no problem with having multiple instances of FortranDerivedType since this does not implement the singleton pattern.