I'm working with a large legacy system with an install script that pulls in close to 200 perl modules. Now at least 25% and probably 50% of those are never used.

I can start by greping for all the "use" lines, then diff to see what I can trim. But is there a way to do better? To determine which pm's are in are "use'ed without actually getting used?

The goal here is to convert as many CPAN modules into supported Debian packages, by the way. And sometimes it's awkward to start with a use statement and end up with the proper Debian name.