Removing parts of pyrates from sys.modules in the backend causes troubles in other parts of the pipeline about pyrates HOT 3 CLOSED

This code was introduced at some point in order to prevent exactly those problems "in other tests". When running different tests with different circuit structures that, however, included the exact same module names, sys.modules would not update the modules between tests. Thus, this code was added to ensure that modules from older versions of the same circuit would be forgotten, so that the new ones could be added to sys.modules. Since every CircuitIR instance comes with just one backend, I did not expect this to cause trouble in "productive code". Also, I never ran into strange bugs. However, it is possible that changes to the code generation structure further down the line made this code unnecessary. Thus, I will see whether removing it causes any other problems. If not, we can get rid of it.

from pyrates.

Okay, so the problem is that Python registers modules for the backend that may need to be refreshed, correct? In my case, the backend removed all modules that contained "circuit" in their name, thus also removing PyRates' own modules for circuit templates and IR. That again cause a problem somewhere else, when I wanted to verify instances of template classes. Checking for instance itself is not a good practice in Python, so it might also be worth rewriting those checks...

Either, I'll still leave this open until we have decided, whether the current fix is acceptable or not.

from pyrates.

So I removed these parts of the code and they did not change anything about running the pytests or other test scripts on my end. I pushed the changes and will close this issue now.

from pyrates.