Thoughts on improving optics package

@jbgi When I updated the JavaDoc of the optics package, I had some ideas and questions:

• The composition methods look very repetitive. Could mixin inheritance help here? Of course a class Foo<A> can’t implement __<Foo.µ,A> and a __<SuperFoo.µ,A> at the same time, so we would need to split off interfaces from the current classes.
• It would be cool to have some predefined lenses, e.g. for tuples.
• It seems Lens uses the conventional getter-setter implementation, not the van Laarhoven style lenses, which seem to be quite hip at the moment (and for a good reason, I suppose).

Implementing them is not difficult (I didn’t want to go polymorphic, which is certainly possible):

import org.derive4j.hkt.__;
import org.highj.data.structural.Const;
import org.highj.data.tuple.T1;
import org.highj.function.Functions;
import org.highj.typeclass1.functor.Functor;

import java.util.function.Function;

public interface VanLaarhovenLens<S,A>  {

    <F> Function<S, __<F, S>> apply(Functor<F> functor, Function<A, __<F, A>> targetFunction);

    default Function<S, S> modify(Function<A,A> function) {
        Function<A, __<T1.µ, A>> targetFunction = a -> T1.of(function.apply(a));
        return s -> T1.narrow(apply(T1.monad, targetFunction).apply(s)).get();
    }

    default Function<S, A> view() {
        Function<A, __<__<Const.µ, A>, A>> function = Const::new;
        return s -> Const.narrow(apply(Const.functor(), function).apply(s)).get();
    }

    default Function<S,S> set(A value) {
        return modify(Functions.constant(value));
    }
}

An instance:

public static <A,B> VanLaarhovenLens<T2<A,B>, A> first() {
    return new VanLaarhovenLens<T2<A, B>, A>() {
        @Override
        public <F> Function<T2<A, B>, __<F, T2<A, B>>> apply(Functor<F> functor, Function<A, __<F, A>> targetFunction) {
            return pair -> functor.map(a -> T2.of(a,pair._2()), targetFunction.apply(pair._1()));
        }
    };
}

(code stolen from here: http://blog.jakubarnold.cz/2014/07/14/lens-tutorial-introduction-part-1.html )

[DanielGronau]

I see your point. Maybe we should instead pull the package one level up to org.highj.optic, it is kind of hard to spot inside data. Further, I think lenses deserve a Wiki page with a short introduction. Of course it makes no sense to repeat the monocle documentation, we can reference to it and just point out relevant differences (if there are any).

[jbgi]

@DanielGronau I think optics are so useful in so many use cases, in modern FP, that I think they should be easily accessible in base/prelude. They are currently underused and I fear that a separate optics sub-project will lower adoption rate.