No, we want to replace an incomprehensibly complex module with a simple one that does only what we want. It has one main function (a SQL wrapper), but a lot of tricky side effects (modifies global variables, etc) that may or may not affect how a given program gets run.

We don't want to break anything in production loads this module when we replace it.

The first step is to find out everything that's using the old module, so that we can do tests to prove to the production staff that nothing will break when we do finally get rid of it.