sounds like you would have to initially differeniate the curve(circular wire) and find the gradient of it....hence the straight lines. and then working out the magnetic field for each of these infinitely small lines and then intergrating it back to the function. close? :p