> I can reproduce it on latest master running on Linux.

Interesting. Perhaps my actual change was to cause it to raise an error 
earlier (outside of a handler)? Which would mean that the Path object 
was always failing before, just silently.

Arguably we should filter them out instead, then, rather than 
introducing a new supported type (as has been argued with similar 
changes elsewhere) in a security branch. Though I'm sure the OP would 
have noticed if their call was never returning any modules, so 
presumably it must have worked somehow.

> I'm happy to open an PR, just let me know or if OP is not willing to do so either.

Feel free. I assume they'd have arrived with a PR if they were keen to 
write it themselves.