good start for lagrangian and hamiltonian treatments.
I remember many a long night with my buddies trying to work out the damn Marion & Thornton problems. The problem with this textbook is that it doesn't teach you techniques you wouldn't think of otherwise the first time on your own. Sure, after finishing your degree, it's obvious that (if you're a physics student) when doing a small amplitude oscillation approximation, you should do a first order Taylor expansion of the sine or cosine term. But the first time you come across a problem that needs that and you've never seen it before and the book never told you, how are you supposed to know? You can spend hours on it until someone figures it out, you manage to find an old solution from years past, or some upper year student tells you how to do it, and you say...."ahhhh, that's so stupid!"
Other than that...