imrafaelmerino / imrafaelmerino

Hi there 👋, my name is Rafael Merino

I am a functional programmer 👩‍💻

Skills: JAVA / SCALA / CLOJURE / LISP / ERLANG / Functional Programming / Actors

• 📫 How to reach me: [e-mail]
• 😄 Pronouns: he
• ⚡ Fun fact: I was a chess ♔ player. I was the champion 🏆 of Spain 🇪🇸 several times during my childhood.

List of my personal projects 💪 in chronological order and their making-of, sit back and grab 🍿:

• json-values

This was my very first project as a functional developer 👏. I learned from Rich Hickey how important persistent data structures are. By the time I started developing json-values, I was working with Json all the time. I really missed a persistent Json and a better API to manipulate them 😡. Developing json-values, I put into practice 🏋️‍ many new concepts for me like recursion, tail-call optimization, trampolines, high order functions, functors, monads, etc.

I started testing 🧪 json-values with a Scala library called ScalaCheck, which is used for property-based testing (PBT). This way, I killed two 🐦 with a stone; I learned a new language, Scala, and an inspiring testing philosophy. I discovered property-based testing thanks to John Hughes and QuickCheck from Haskell. PBT led me to create a beautiful Json generator. I challenged the world, claiming it was the best Json generator in the whole galaxy 🌌 by that time (and still it is!) You can benefit from this Json generator to test your code. Prove me wrong if you have a better alternative!

Spec is an incredible Clojure library. I implemented json-spec based on the ideas I learned from it. At the risk of being presumptuous, I claim it's by far the best way of validating a Json in Java and Scala. It's easy to write, read, and maintain. Defining json generators and specs is a piece of 🍰.

I bought the book 📖 Optics by example, from Chris Penner, written in Haskell. Good read, by the way. I studied Monocle in Scala as well. That allowed me to develop some optics in json-values. Since then, I use optics all the time.

Developing json-values, I started taming effects with FP. I experienced how great FP is describing, composing, and executing programs made up of effectful functions λ. I designed an interactive program that, given a spec, creates a Json asking the user for every value on the console 📺. I also wrote a Json future.

I developed json-values in Scala and Dotty as well. Dotty is a prerelease of Scala3.

• Java version
• Scala version

Scala is, in my option, a good alternative to do FP in the JVM. I've learned a lot from the Scala community. I have tremendous respect for the Scala creator, Martin Odersky, and the ZIO creator, John A De Goes. I never miss any of their talks and articles.

On the other hand, Clojure is a great choice as well. It's really challenging since it's a lispy programming language. As a Lisp fan, being able to use it in the JVM is incredible. You must see any talk from Rich Hickey. He made a significant impact 👊 on me, and he changed my life entirely as a programmer. Being honest, it took me almost three years of my life to develop json-values and really understand functional programming. The Scala version took me only four months because I had already grasped the fundamentals.

Java doesn't have persistent data structures. Scala and Clojure does. This was a problem implementing the Java version of json-values. I tested a lot of alternatives and compare them in terms of performance and design. I ended up using the persistent data structures from the library vavr.

• vertx-effect

Suppose I had to describe vertx-effect in three words. In that case, they'd be without a doubt: expressions, composition, and persistent data structures or values. vertx-effect is the place where actors model meets functional programming in Java.

I couldn't resist naming the most essential function in this library as lambda 😎

I had been working with Vertx for a long time. I always had the feeling that I could do it better 🤷‍♂️. Asynchronous programming is hard. It's even harder if you have to deal with the callback hell 🔥 to do your business logic. Imperative programming doesn't help here.

On the other hand, I started learning Erlang and watching videos of Joe Armstrong. I read his book 📖 and thesis. I really understood the actor model and how powerful it is. I fell in love with failures 🤪. It's critical to know that every system will fail, and you'd better be prepared. Erlang uses persistent data structures, and so does vertx-effect thanks to json-values 👏

I learned from John A De Goes how to use FP to deal with effects. I took the course Principles of Reactive Programming in Scala. The result is vertx-effect. There are some expressions implemented that come from Lisp, like Cond, Case, and IfElse.

It was sad to find out that Joe Armstrong had passed away † in 2019, a few months before I met him. Rest in peace, Joe. I strongly recommend you study his work. He was very brilliant and, at the same time, very kind and humble.

• vertx-mongodb-effect

You can model any effect in vertx-effect using lambdas λ. vertx-mondodb-effect gives you some λ to interact with the MongoDB. Going from Json to BSON and vice versa was really cumbersome and tedious. That's why I developed mongo-values, a set of codecs that abstracts you from that.

• mongodb-values

json-values and mongo-values are ideal for working with MongoDB. mongo-values gives you a set of codecs that free the programmer from doing any kind of conversion to BSON.

• JIO

JIO is the proof you can do Functional Programming in Java 🕺. Values, expressions, and functions on top of the CompletableFuture API. The IO monad is extremely powerful and allows you to tame any imaginable effect: console programs, HTTP requests, database calls, etc. Everything is composable and referential transparent in the world of lambdas λ and values, which helps you cope with complexity 😌. I think, nowadays (beginning of 2021), there's nothing compared to this in the Java world 🤷‍️.

I implemented a reactive MongoDB client and a reactive HTTP client. I developed some interesting console programs that show the essence of JIO.

It's the first time I used JFR and JMC. I'm glad that since Java 11, we have these tools. Logging is a key aspect of any system.

I fiddled with the fork/join framework that came out in Java 7. I thought you couldn't use it for blocking operations, but that's surprisingly false. The ManagedBlocker interface opens the door to submitting blocking tasks to the pool, and I took advantage 💪 of this in JIO.

Since Mockito and all that stuff are not my cup of tea, I developed a native Java HTTP server to test my HTTP client. Of course! The server is configurable just with functions.