When I learn cross product, I find myself always using orthogonal basis vectors (e.g. $\hat{i}$, $\hat{j}$ and $\hat{k}$). But I am wondering does cross product depend on the orthogonality of the basis vectors? i.e. can I calculate the cross product of two vectors which are written in a set of non-orthogonal basis vectors? Why or why not?
Also, what if the basis vectors are changing (e.g. in cylindrical polar coordinates)? Why is cross product still defined in such a set of basis vectors?
How about for dot products? Does the dot product depend on the orthogonality or invariation of the basis vectors? Why or why not?