A Clojurescript (jvm-hosted) repl can be seen as two repl processes. One in the jvm and one in the browser. In the browser we have identity for read, us/eval for eval and prn for print. On the jvm we have regular read, then an eval that compiles to JS, write on a stream, read on another, and println as print. The pair of streams is are the input and output of the in-browser repl. What’s the point? Well it could help to have upgradable cljs repls.