Last time I looked into it, ABCL was less performant than more recent JVM hosted languages. I am far from a JVM expert, but I was told that this is due to most

Last time I looked into it, ABCL was less performant than more recent JVM hosted languages. I am far from a JVM expert, but I was told that this is due to most development work on ABCL predating JVM enhancements for dynamism. There is also FOIL, which runs JVM (or CLR) in a separate process and provides a bridge for calling into JVM from lisp. It was written by Rich Hickey (the future author of Clojure), who observed that the reflection overhead was similar in magnitude to the serialization overhead!