Assuming that a bunch of regexps is equivalent to a parser generator is misguided and naïve, I think we both agree on that point.

But I think you miss my point about cross referencing a Perl program. Cross referencing involves static analysis of source code. For some subset of Perl, this will be fine, but once you start to involve source filters, run-time evaluation of code, closures, ties, autoloading... the problem becomes impossible. Some of the code simply does not exist before the program runs. It is not written, the program writes as it goes along. How can you cross reference that? And the more complex a system becomes, the more likely the above techniques will be employed in order to get a handle on the complexity.

I'd suggest looking at the Devel:: and Test:: hierarchies, for looking at tools to help you instrument a running program and get feedback on its behaviour, and ensuring its correctness (or at least that you haven't broken anything that used to work when you introduce new code). For instance, I have used chromatic's Devel::TraceMethods to understand a bug in an OO hieararchy that relied heavily on autoloading.

Honestly, I didn't even know that there was such a thing as B::Xref. I think Malcolm Beattie wrote it more as a proof of concept, rather than expecting people to really use it. Cross referencing just seems so '80s to me.