Well, but if there is an existing (and commonly recognised) way of expressing Perl function prototypes and data structures in UML, I should rather use it.
It will be nice if we all follow the same standard as we can easily understand each other. This is the motivation behind 'Unified' in 'Unified Modeling Language' and behind all the conventions that Perl programmers tend to follow without being enforced to.

So I would rather find a standard/convetion of expressing Perl in UML and get used to it than invent my very own "standard". Or at least I would ask for "common wisdom" on which way to choose.