I have the following problem:
$$(y+u) u_x + y u_y = x - y $$
I only know characteristic curves so I did:
$\frac{dx}{dt} = y + u $
$\frac{dy}{dt} = y$
I began solving the second equation resulting $y = C_1 e^t$. Using that result I solved $\frac{dx}{dt} = C_1 e^t + u $. This is where my doubts began, since $x$ has $u$. I decided to give it a try so I got $x = C_1 e^t + u t $. Here there's a constant that I've been said the method omits. And I've seen some examples where it's omitted. So I continued by using the chain rule, ending up in this equation:
$$\frac{du}{dt} = x - y = (C_1 e^t + u t) - C_1 e^t = u t $$
Solving this gives $u = e^{\frac{t^2}{2}} + K $ where I know $K$ is an arbitrary function. Replacing $t$ from $y = C_1 e^t$ we get $$u = \ln \big(\frac{y}{C_1}\big) + K $$ but clearly this is incomplete. I don't know the form of $K(x,y)$ and $C_1$. I would like to know how to manage the arbitrary functions since I feel it's what's happening, and how to manage them in this case where we have a quasilinear PDE. Also, if there's another way to solve this, I don't know if substracting $\frac{dx}{dt}$ and $\frac{dy}{dt}$ helps in some way since it leaves $u$ or if there's a common change of variables that's useful in these problems. Thanks for the help.