I'm studying p. 160 in Ryder's book of QFT and there is an example where the standard path integral equation is not valid
$$\langle q_ft_f|q_it_i\rangle = N \int Dq \exp \left( \frac{i}{\hbar}S \right) \tag{5.15} $$
meaning when we have a position-dependent kinetic energy like in the Lagrangian:
$$L=\frac{\dot q^2}{2} f(q). \tag{5.15a} $$
Using the "explicit" expression for $Dq$ I have
$$\langle q_ft_f|q_it_i\rangle = \text{Const} \cdot \lim_{n \rightarrow \infty} \prod_j dq_j \exp \left( \frac{i}{\hbar} \sum_j (q_{j+1}-q_j)^2 \frac{f(q_j)}{\tau} \right) \tag{1} $$
where I separated the path in equal time intervals of $\tau$. The result in the book is
$$\langle q_ft_f|q_it_i\rangle = N \int Dq \exp \left( \frac{i}{\hbar} \int dt (L - \frac{i}{2} \delta(0) \ln f(q)) \right) \tag{5.15d} $$
I don't understand the meaning of $\delta(0)$ and I don't get where the second terms come from (the one with the $\log$), I don't seem to get it from eq. (1).