I know that if $f:R^2\to R$ has continuous second derivatives, then $f_{xy}=f_{yx}$. But is it possible that $f_{xy}$ is continuous but $f_{yx}$ is not, or that $f_{xy}$ exists but $f_{yx}$ not?
If so, when I calculate $f_{xy}$, I don't know whether $f_{yz}$ exists and continuous, and I cannot use the mixed derivatives theorem, then what is the meaning of this theorem?
Straight from Baby Rudin (theorem 9.41, page 235-6), the theorem is thus: