But local also works with lexical variables, doesn't it? IMHO, I didn't think it was necessary for $my_sub_level to be a global, per se. Since perl 5.6 it's been possible to localize nearly any variable; I've localized member variables of class instances, lexicals, arbitrary scalars within complex multi-level structures (hash containing arrayrefs of hashrefs containing hashrefs, etc.).

Granted, localizing such a thing may (temporarily, subject to the enclosing scope) hide the thing being localized with what is effectively a global, but that's an internal implementation detail that seems to have mattered not, at least with the code in which I've used it.

To return to the original question, isn't $depth referred to as a closure? Or (perhaps my semantics are incorrect) is sub recurse the closure? Semantics aside, I'm surprised no one mentioned the word.

dmm