[Python] add federated send class test

[housengw]

Currently in the Python target, when a federate sets its output to a class, the other federate somehow would try to import the sender federates’s module, causing a moduleNotFoundError.

This PR adds federated tests that sets the output port to a class to help debug the issue. Those tests are expected to fail.

This PR is a work in progress.

[lhstrh]

If it's WIP, then it's better to mark it as draft.

[housengw]

The tests fail locally on my machine when I do runSingleTest but somehow pass in GitHub actions.

[housengw]

given that the send class tests pass on the GitHub action as well as a few other machines I tried, seems like the problem is an edge case local to my machine. My machine runs on MacOS Monterey. I'll include the output running test/Python/src/federated/DistributedSendClass.lf on my machine for future reference:

Terminal running the RTI:

oysteryio-2:Python wonghouseng$ rti -i 1 -n 2
RTI: Federation ID: 1
RTI: Number of federates: 2
Starting RTI for 2 federates in federation ID 1
RTI using TCP port 15045 for federation 1.
RTI: Listening for federates.
RTI: All expected federates have connected. Starting execution.
RTI: Waiting for thread handling federate 0.
Federate 0 has resigned.
RTI: Federate 0 thread exited.
RTI: Waiting for thread handling federate 1.
Federate 1 has resigned.
RTI: Federate 1 thread exited.
RTI is exiting.

Terminal running federate a, which receives a class on an input port from federate b:

oysteryio-2:Python wonghouseng$ python3 src-gen/federated/DistributedSendClass/a/DistributedSendClass_a.py -i 1
Federation ID for executable src-gen/federated/DistributedSendClass/a/DistributedSendClass_a.py: 1
Federate 0: Connected to RTI at localhost:15045.
---- Start execution at time Fri Jan 14 12:13:41 2022
---- plus 341069000 nanoseconds.
Federate 0: Starting timestamp is: 1642191240179447000.
ModuleNotFoundError: No module named 'DistributedSendClass_b'
Federate 0: FATAL ERROR: Could not deserialize deserialized_message.
Federate 0: Connection to the RTI closed with an EOF.
---- Elapsed logical time (in nsec): 0
---- Elapsed physical time (in nsec): 9,731,000

Terminal running federate b, which sends a class on an output port to federate a:

oysteryio-2:Python wonghouseng$ python3 src-gen/federated/DistributedSendClass/b/DistributedSendClass_b.py -i 1
Federation ID for executable src-gen/federated/DistributedSendClass/b/DistributedSendClass_b.py: 1
Federate 1: Connected to RTI at localhost:15045.
---- Start execution at time Fri Jan 14 12:13:59 2022
---- plus 179424000 nanoseconds.
Federate 1: Starting timestamp is: 1642191240179447000.
^CFederate 1: Connection to the RTI closed with an EOF.
---- Elapsed logical time (in nsec): 0
---- Elapsed physical time (in nsec): 6,438,575,000

[housengw]

Ok, the beefy desktop started to get the same ModuleNotFound errors as my macbook air when I ran the send class tests yesterday. I'm guessing it has to do with a python-related package that I installed? I can't pin-point which one though

[Soroosh129]

Thanks for creating these tests.

As it turns out, there is a solution to this issue that only requires a minor change. To oversimplify, previously, user-written code was being loaded by treating the .py file (that contains the user-code) as an external module. However, the user code can instead be loaded from the __main__ module (which is the current top-level module that is already loaded). With this change, if the user-written class is defined or imported in the top-level preamble, it would be serializable as well.

Classes defined or imported within federate preambles, however, will not be serializable. I think this should be an expected behavior by people familiar with Python though.

[lhstrh]

Looks OK to me, but I'm a little out of my depth here. Can you add newlines to the ends of the files and modify RELEASES.md with a summary of the bug that it fixes?