No particular challenge, really. In the end I went for the Python solution because (1) the problem is too trivial to require a serious language, (2) the result is arguably more accessible, (3) was a bit faster to write because the file monitoring libraries in Nim aren't quite as mature. I just wanted to have it done quickly to use it for other Nim projects.