This is Problem 1.17 from Billingsley's Convergence of Probability Measures.
Let $C=C[0,1]$ be the space of continuous functions with the sup norm. Show that every locally compact subset of $C$ is nowhere dense.
Below is the solution to this problem, but I have several things I don't understand. Since we assume the contrary, we have some open ball $B \subset \bar{L}$.
However, how do we arrange that $\bar{B} \cap L$ is compact? And how do we choose points from $B$ so that $\rho(x_m,x_n)\ge \epsilon>0$ for $m\neq n$? Also, we can choose such $y_n$ from $\bar{B}$, but how do we ensure they are also from $L$? Finally, how does this conclude that $\bar{B} \cap L$ is not compact?
I would greatly appreciate some help.

We can arrange that $\overline{B}\cap L$ is compact because $L$ is locally compact. Since $B\cap L$ is an open neighbourhoods of some point $x$ in $\overline{B}\cap L$, there is a relatively compact neighbourhood $U$ of $x$ with $U\subseteq B\cap L$. By making $B$ smaller if necessary, we can ensure that $\overline{B}\cap L\subseteq \overline{U}$, which is compact. (Note that $U$ and $\overline{U}$ are subsets of $L$ here.) We are allowed to consider a smaller ball because we need only show that $\overline{L}$ has empty interior, and we are working with arbitrary balls anyway.
To find the points $x_n$, we must use a fact about the space $C[0,1]$: it is an infinite dimensional Banach space. Here we can invoke the Riesz geometric lemma to get the points we are after.
To find the points from $L$ that are close to the points $x_n$, you need to use the fact that $L$ is dense in $B$.
Finally, once we have chosen our sequence $(y_n)$, we note that this sequence cannot possibly have a convergent subsequence because we have $\rho(y_n, y_m) \geq \varepsilon/3$ for all $m,n\in\Bbb N$ with $m\neq n$. This means the sequence $(y_n)$ does not satisfy the Cauchy condition and so cannot be convergent, and every sequence in a compact metric space must have convergent subsequence.