What can be said about the closed definable subsets of an omega-categorical linear order? This question arouse in the investigation of NIP omega-categorical structures and might also be relevant to understanding Ramsey expansions of homogeneous structures. I will discuss a conjectural answer and partial results towards it.