The Essence of Evolutionary Algorithms

I've been thinking about the essence of evolutionary algorithms, trying to get the right structure for my framework- general enough to use in any project I want to do, but easy to set up and apply to common problems.
The structures involved seem to be- a population of individuals that are strings of symbols that gets exposed to an environment. The individuals learn from their exposure, a set of terminating conditions are checked, and if none are met, start from step 1.
This would looks really nice as haskell code (or really any functional language, I just happen to like haskell right now) , but I'm not familiar enough with it to write a framework that I will be using for a project that starts pretty soon.
Therefore, I will just hope to redo it all in haskell one day, and for now consider building the functional abstractions that I want to use into Java. This would be an interesting learning experience, and may end up improving the architecture of the program a lot.
On the other hand, I will almost certainly end up with only some of the cool stuff built into a functional language, the program will be verbose and incomplete, and may end up getting thrown out. I'm okay with that, cause if it works well enough for now, maybe I can get it right later, and if nothing else it will motivate me to deepen my understanding of certain aspects of functional languages. I will basically be making a little language in Java, and nothing teaches you a concept like having to implement it, and getting it to work correctly.