I think the best way is to have a good taste when choosing how to solve the problem. Going purely functional (Haskell) has its own benefits but also a lot of d

I think the best way is to have a good taste when choosing how to solve the problem. Going purely functional (Haskell) has its own benefits but also a lot of drawbacks. If you don't go purely functional, it doesn't make sense to avoid imperative or OO features where they make sense. In that sense, I think languages which fuse and unify functional and OO approaches are a good strategy. There is a bit of negativity towards them, because they manage to piss off both OO evangelists as well as functional preachers, but that doesn't detract from the technical benefits of such an approach.