I'd argue Clojure is setting some new standard here with its powerful literal notation for basic data structures (list, vector, set, and maps) and its practice of integrating/composing first on data. Most of the time that's enough to get some libraries work together. If not enough, integrating/composing on functions is the next stage. Only after having covered data and functions should you consider macros, which should be easier to write if you have done a good job at the data and function levels.

If such powerful and ubiquitous data structures work so well for programs written in clojure(script), why would that not be true in the larger picture for integrating systems? JSON worked much better than XML because it actually represents simple data structures, not objects talking to each other. Redis, Avro, and similar tech are continuing that story while Kafka adds a fantastic transport and storage mechanism. Sounds like we're closer to having better foundations for integration in the small and in the large.