In Silverman's Book on elliptic curves, in example 4.6, he says:
Let $C:y^2=(x-e_1)(x-e_2)(x-e_3)$ then $div(dx) = div(d(x-e_i)) = div(-x^2d(\frac{1}{x}))$. He concludes that $div(dx) = (P_1) + (P_2) + (P_3) - 3(P_{\infty})$ where $P_i$ corresponds to the point $(e_i,0)$
Now here's what I don't understand. $x-e_i$ should be a uniformizer for $P_i$. Hence, $ord_{P_i}(dx) = ord_{P_i}(\frac{dx}{d(x-e_i)}) = ord_{P_i}(\frac{dx}{dx})$ This feels like it should definetly be $0$. Further, everything I've been led to beleive up to this point would tell me that $ord_\infty(dx) = -2$. Even in his previous example over the projective line this was the case.
Now, logically - since $e_1, e_2, e_3$ appear once each as a factor of the RHS of our curve, I'm inclined to believe Silverman. Also, since the genus is 1, its to be expected that the divisor is 0.
What am I missing?