Not sure this is the right place to mention this, but I'm realizing handling symlink loops will be interesting. It is "interesting" because, when e.g. you are globbing, your glob's results may not include the symlinks' target path; so there are cases where you need to keep the symlink in the results.
(and in the general case you may or may not want to keep some duplicates, depending on what you do with the links)

I am adding rglob() to pathlib and currently ignoring (i.e. crashing :-)) symlink loops. I would be interested in any ideas as to how handle them correctly.