I actually took a University course in M by one of its maintainers, the late Dick Walters. While I never found any personal utility in the language, it was a curious 3 months of experimentation, and Walters was a great professor and fairly kind and helpful, probably because he realized how eccentric and borderline archaic M was.
I would have loved to follow such a course. Whatever I know and remember of it was learned 'the hard way' by reading code and digging into whatever problems it exhibited, not exactly a fun experience. But I have to admit: some of it was extremely clever and if you let go of the Unix file based mindset for a bit it actually was an interesting and refreshing experience. But I didn't think MUMPS was future proof enough to warrant investing more time into it and the only reason it still hangs on is for the same reason COBOL still does: large codebases that businesses rely on even if there isn't anybody there that still fully understands them.
