Presumably the advantage is speed? Being a compile-time thing rather than a run-time eval?

That's a beneficial side-effect of avoiding string eval. But mostly just a healthy aversion to using it, if it is easier to avoid it.

I don't subscribe to the evil-eval notion--if I need to evaluate a runtime mathematical expression, I will. Far easier to use Perl's expression parser than write my own, or discover the limitations of someone else's--but for the usage outlined, everything is known at compile-time, so no need to invoke the compiler again (multiple times) at runtime.