Avoiding cyclic imports, but how...?

Start by reproducing the recursion error from the minimal mylib example using test.py and test.html, then inspect the generated JavaScript around the imports in mylib/a.py, mylib/b.py, mylib/examine.py, and mylib/init.py. Compare the Transcrypt output with CPython's behavior and determine whether cyclic imports can be supported or need a documented limitation and workaround.

Hi, I'm still trying to port a library that initially was written to run with PyJS to Transcrypt, and got into one more issue where I was not able to find any running solution. I am trying to describe the problem in a much simpler test case, which does exactly the same and runs in the same problem, a "too much recursion" message in the browser.

.
├── mylib
│   ├── a.py
│   ├── b.py
│   ├── examine.py
│   └── __init__.py
├── test.html
└── test.py

mylib/a.py:

from mylib import examine

class A(object):
        def examine(self):
                examine.examine()

mylib/b.py:

from mylib.a import A

class B(A):
        pass

mylib/examine.py:

import mylib

def examine():
        for key in dir(mylib):
                print(key)

mylib/__init__.py:

from mylib.a import A
from mylib.b import B

import mylib.examine

The reason for the function examine() in mylib.examine is, that it both should be callable outise of A or in context of A, and it examines the contents of the mylib module, its objects and classes.

CPython compiles and runs this without compromise. The Transcrypt-generated output runs into an recursion error, because mylib.examine imports examine. According to issue #106, cyclic imports are a problem. So, surely, this is more a question how I have to style my code, but I can't find a solution right now... can this problem be solved in any way?