I am investigating Paulin's method of certain limits of actions on hyperbolic spaces being (in some sense) actions on $\mathbb{R}$-trees.
Let $G$ be a finitely generated group. Part of the proof is saying that given a sequence of non-elementary isometric actions $\rho_i: G \to \textrm{Isom} X$ on a proper, cocompact hyperbolic space $X$ such that there are points $x_i$ which generators of $G$ move by at most uniformly bounded distance, we can find a subsequence of the actions, which after conjugation by an isometry converge to an action $\rho: G \to \textrm{Isom} X$ (let's say they converge pointwise).
My question is as follows. Isometries of $X$ extend naturally to the natural compactification $X \cup \partial X$ ($X$ is a hyperbolic metric space). Therefore the actions (evaluated at a given $g \in G$, say) seem to be calling for some topological version of Arzelà-Ascoli. Does anyone know a version which would be useful in the setting of hyperbolic spaces with the natural boundary?