Given a set of disjoint line segments in the plane, prove (or disprove) that you can always join the line segments to make a near-triangulation where the vertices are the endpoints of the segments, the outer face is a cycle and every segment is an edge of the triangulation.
This task interested me, at least in non-degenerate case. I've tried to google it and suddenly found a beautiful approach described here (starts with "Next, the constraints are added to the Delaunay triangulation...", but we don't care about Delaunay here).
Name the set of segments $S$. First, forget about segments and build any triangulation $T$ of all endpoints of $S$. Then repeat the following process for all segments one-by-one:
If $e \in T$, we are done with this step. Otherwise, remove all edges of $T$ having an interior intersection point with $e$. One can see that it will unite some triangles of $T$ into one polygon. Add $e$ to $T$; it will break this polygon into two smaller polygons. Triangulate them both and add needed edges to $T$.
The key point is that if $e \in S$ was added to $T$, it will never be removed, because segments in $S$ don't have interior intersection points. So, after this process $S \subset T$.