We defined the oscillation for a bounded function $f: X \to \Bbb R$ on a metric space $X$ as $$\omega(f,x) = \lim_{\delta \searrow 0} \left( \sup f\left(B_\delta(x) \right) - \inf f\left(B_\delta(x) \right)\right)$$
and I showed already that $f$ is continuous at $x \in X$ iff $\omega(f,x)=0$. Now we proved a lemma which says for every $\eta \ge 0$ the set $N_\eta = \left\{ x \in X \mid \omega(f,x) \ge \eta \right\}$ is closed.
My question is: Why is $M := \left\{ x \in X \mid f \text{ is continuous at } x \right\}$ not necessarily closed?