In the introduction to algorithms proof of Dijkstra, I don't understand why the statement "both y and u were in V-S when u was chosen". We add x before y, and so we relax d[y] with the the edge $$\omega(<x,y>) + \delta(s,x)$$, d[u] can't be smaller than delta[u], so shouldn't y be in S when we add x to S? thanks.
$u$ is supposed to be the first node to be insert in $V$ such that $d(s, u) \neq \delta(s, u)$. Before the insertion, $u$ is in $S-V$, it's clear. In addition, $y$ is supposed to be the first node in a choosen optimal path $s -> u$ to belong to $S - V$ when picking $u$ from $S - V$. So before we insert $u$ in $V$, $u$ and $y$ are both in $S - V$. Note that $y$ can be equal to $u$.