Do I point out the errors?
Certainly... but have you really found some?
I bet that "this doesn't run under strict" won't be considered to be an error by anyone who doesn't use strict in the first place. If you have uncovered real errors somewhere (i.e. the code doesn't do what it is supposed to, or it is vulnerable to an attack, or will fail on bad data), then certainly point them out in a spirit of cooperation and teamwork.
If not, perhaps you would consider building some tests for the existing code instead? This will first make you better aware of what their code is suposed to do, rather than having you working backward and trying to figure out what existing code actually does. Once you understand the specifications, and have tests to exercise the code, then you can say with some authority whether there are true errors. If not, I'd say leave that code alone and move to the next piece (there is always more). It may well be that an ugly little piece of cruft is perfectly functional and even robust and need never be touched again.
I'd like to be able to assign to an luser