We will quickly see that it is super easy to convert Observables from the old interface to the new one.

First of all, let's create an old-typed Observable that we can play with, something like this:

rx.Observable.just("One", "Two", "Three")
        .doOnNext(i -> log("doOnNext", i))
        .subscribe(i -> log("subscribe", i));

To convert that to a new Observable of the io.reactivex.Observable type, we will need to use RxJavaInterop.toV2Observabe():

RxJavaInterop.toV2Observable(rx.Observable.just("One", "Two", "Three"))
        .doOnNext(i -> log("doOnNext", i))
        .subscribe(i -> log("subscribe", i));

It worth noting that it is especially useful to use the static import in such cases so that the code can be simplified to the following:

import static hu.akarnokd.rxjava.interop.RxJavaInterop.*;

toV2Observable(rx.Observable.just("One", "Two", "Three"))
 .doOnNext(i -> log("doOnNext", i))
 .subscribe(i -> log("subscribe", i));

Actually, if the toV2Observable call is used very often, it might be worth to write your own wrapper to make it less intrusive:

static <T> Observable<T> v2(rx.Observable<T> source) {
return toV2Observable(source);
}

So, the code will now look this:

v2(rx.Observable.just("One", "Two", "Three"))
        .doOnNext(i -> log("doOnNext", i))
        .subscribe(i -> log("subscribe", i));

However, that depends on a personal taste and the guidelines that your team prefers. It might not be a terribly useful tip, but it's important away from introducing your own abstractions in the code that make more sense in the context of the domain that you are working in.