Functional Design in Clojure

Ep 098: Composed Learnings

Each week, we discuss a different topic about Clojure and functional programming. If you have a question or topic you'd like us to discuss, tweet @clojuredesign, send an email to feedback@clojuredesign.club, or join the #clojuredesign-podcast channel on the Clojurians Slack. This week, the topic is: "composition is life". We reflect on keeping the necessary mess at the edges so our core can be composed together with beauty and simplicity. Our discussion includes: Our reflections on the episodes in this series. The best concepts from the episodes in this series. How to separate data from behavior. Why zucchini bread is at the core of functional programming. What baking has to do with functional programming. How to get objectivity for better understanding. Why you should focus on trade offs. How Clojure and immutability steer you toward composition. An inventory of different kinds of functions. Why side effects ruin composition. How Clojure lets you focus on your problem. How composition works in the small and the large. The paradox of constraints providing more freedom. The cognitive load of functional programming vs object-oriented programming. Selected quotes: "This episode we thought we would take it and put it all together." "The last draft." "Functional programming is built on composition." "Object-oriented programming tends to have at least a little bit of behavior and data together, so you can't separate them if you want to make a new thing." "If you just want classes to be one thing, just data or just methods, all of a sudden you end up with something that looks a lot like functional programming." "Constraints breed freedom and clarity. If you can't reach other functionality through data, then data will never hurt you." "It takes a while to discover what is worth separating and what's fine to leave in larger chunks." "You have to use it." "So by moving [a problem] into a separate domain, you're able to see the essentials, the fundamentals, more clearly because you're not stuck in those ruts of thinking that you do without realizing." "I've been known to eat the same thing for breakfast every day, for weeks on end." "Clojure core plus immutability really steers you in the direction of composition." "'Transforms' is a little unfair of a category because that's almost all of functional programming." "Data cake. Delicious!" "I want to do something with this data that affects the world around me besides generating heat." "Side effects ruin composition because side effects are a second dimension that is difficult to compose." "When you write functions in the way that Clojure expects them, this whole world of core opens up, and the only thing you had to implement was your the code that was relevant to your domain." "There's almost nothing to Clojure itself. Most of my time writing in Clojure is thinking about my problem--not thinking about how I'm going to express it or how I'm going to model the object graph or any of that stuff." "I'm just left thinking about my own problem." "The key in all composition is a clear delineation and separation of responsibilities." "Because they're smaller, you can understand them at a glance." "With side effects, there's an infinite number of additional dimensions that are possible. You can't just worry about what the function says it does, you have to worry about all the other things that it could be doing!" "Side effects are like really advanced politics between hostile countries. There are the words that come out of their mouth, and then there's all the stuff that really matters that no one's talking about!" "Clojure is for the peaceful." "If you like to keep all the messiness at the edges so the core of your application can be composed together with beauty and simplicity, making your day wonderful, then Clojure might be for you!" Common kinds of functions: Predicates: provide truth values Reducers: take the current "state" and an input and return a new state Transforms: broad category that covers many kinds of functions Conversions: take one kind of thing and turn it into another kind of thing eg. parse-long Extractors: return part of a data object Mergers: combine data together Decorators/enrichers: takes a data shape and adds more into it while preserving the original data shape Side effectors: affects the world in a clear and specific way Links: Ep 093: Waffle Cakes Ep 094: Concrete Composition Ep 095: Composing Core Ep 096: Gaming Data Ep 097: Application of Composition

Ep 097: Application of Composition

Each week, we discuss a different topic about Clojure and functional programming. If you have a question or topic you'd like us to discuss, tweet @clojuredesign, send an email to feedback@clojuredesign.club, or join the #clojuredesign-podcast channel on the Clojurians Slack. This week, the topic is: "composing your application". We get a handle on bringing I/O resources together in an application. Our discussion includes: What is a "handle"? Why is it called that? More puns! How do you handle I/O handles? How do you keep the pure and side-effecting parts away from each other? Managing and organizing application components. What is a framework vs a library? "Injection style" vs "singleton style". How far down the call stack should handles travel? Figuring out a start-up order and a shut-down order. How obtaining a resource can negatively affect composition. Separating the knowledge of finding a resource vs using a resource. The concept of an "application index". Why all this matters for long-term maintenance. Using feature flags. The two developers on every software project. What is the "band-aid test"? Selected quotes: "If a program does a bunch of computation, but it doesn't have any I/O, did it do anything?" "Handles inherently are going to hurt you. You don't even have a hold on the whole thing. You have a little bit that you can use to to access that external system." "I visualized an axe or a knife. You have a handle, but you have to watch out for what the handle is connected to!" "In Clojure, I think of frameworks as an integration between your code and the purpose of that framework as opposed to the foundational layer that glues your whole world together." "Even in OO, it's a smell if I/O handles travel too far and wide!" "Lots of assuming!" "Sometimes ad hoc is confused with flexibility. You didn't have to think about it, so it felt flexible, but it just happened." "Composition is about being able to use things in multiple ways." "If you are accessing a singleton state from a function, you are no longer side effect free." "Why would I ever save anything anywhere other than the place the application has configured as its database?" "Have you ever had a physical "dead tree" book?" "Where are all the components?" "With the index style, you have a clear delineation of all the parts of the software and how they're connected together." "Whenever you start using a library or framework, you are hopefully solving a problem, but you're also inviting complexity in. You have to be wary of what it allows you to do, and what it allows future developers to do." "If you pull that band-aid off, how many little hairs are going to come off with it? How badly is it going to hurt?" "One of the tests of composability is the surface area of contact. More is not better!" Links: Injection style: Component + component.repl, Integrant Singleton style: redelay, mount Built on Component: system Others: donut.system, Components in Biff Ep 063: Web of Complexity Ep 067: Handling Handler's Handles Ep 093: Waffle Cakes Ep 094: Concrete Composition Ep 095: Composing Core Ep 096: Gaming Data

Ep 096: Gaming Data

Each week, we discuss a different topic about Clojure and functional programming. If you have a question or topic you'd like us to discuss, tweet @clojuredesign, send an email to feedback@clojuredesign.club, or join the #clojuredesign-podcast channel on the Clojurians Slack. This week, the topic is: "playing games with data." We go back to start and play through a composition strategy to see where we land. Our discussion includes: The return of Tic Tac Toe. It's back! The biggest difference between object-oriented and functional composition. Can a game board play its own pieces? Isn't that the stuff of horror films? What does it mean to be "complected"? What's more important: data or behavior? Why separate data from behavior? The simple domain of batch processing. The dark magic of hidden side effects. What lasts longer: data or your application? Representation vs operations in an information domain. Common kinds of functions. How do schemas fit in? Where does I/O fit in? Rationale for organizing your namespaces. The horror of code organization in Java. The stability of pure information models. What is language imposed complexity? Selected quotes: "It's been a few years for us, but it could have been a few days if you're really on a binge on this podcast!" "The biggest and most important difference between object-oriented programming and functional programming is the relationship between data and behavior. Functional programming keeps those separate, whereas object-oriented programming likes to mix them together into classes." "When I want to play a piece, does the board know how to play pieces onto itself? Or, does a piece know how to place itself on the board?" "You can't just have the functionality sit on its own. It has to be attached." "I arrived at a point where I thought, 'What am I doing?!'" "All kinds of different scripts that operate on the same data file don't need to know about each other at all." "I'm just going to get the name attribute from the Person object, right? You don't know. Am I getting a field, or am I getting some magical overloaded getter thingy that does a bunch of side effects?" "Data is simple, and data will almost certainly outlast your application. Often, the data that your application deals with came from somewhere else in the first place." "This reminds me of the fun abomination known as object-relational mapping where you take the data that was there before your program and you pour it somehow into your objects, then you do something with it, and then you pour it back out into the database. Why is this such a hard problem? Because it's a hard thing to do! It's something that doesn't make sense." "In OO, there are way more dimensions of potential organization, so therefore, more things for a software development team to fight over." "The only place you can put functionality in object-oriented language is attaching it to classes." "In Clojure, you don't have a lot to fight over in your bike shedding sessions." "When you're talking about which object to put it in, you're not actually talking about your problem. It's a problem that is introduced by the tool choice, not by the domain of the problem." "Because they're separated, you get reuse. You get data composition and behavioral composition as separate kinds of composition--because they don't often happen the same way." "I think the lack of reusability comes in object-oriented languages, not functional languages. Because the problem with object-oriented languages is they’ve got all this implicit environment that they carry around with them. You wanted a banana but what you got was a gorilla holding the banana and the entire jungle. If you have referentially transparent code, if you have pure functions all the data comes in its input arguments and everything goes out and leave no state behind it’s incredibly reusable." ~ Joe Armstrong in "Coders at Work" Links: "Execution in the Kingdom of Nouns" by Steve Yegge Ep 002: Tic-Tac-Toe, State in a Row Ep 093: Waffle Cakes Ep 094: Concrete Composition Ep 095: Composing Core

Ep 095: Composing Core

Each week, we discuss a different topic about Clojure and functional programming. If you have a question or topic you'd like us to discuss, tweet @clojuredesign, send an email to feedback@clojuredesign.club, or join the #clojuredesign-podcast channel on the Clojurians Slack. This week, the topic is: "core and composition." We venture toward the core of a solution. Our discussion includes: Why do functional programmers like to talk about composition? What composition means in object-oriented programming vs functional programming. Must simple be simplistic? What is idiomatic composition? Common kinds of functions. Why functional programming is more like Legos than taxonomies. What Clojure already knows and what you have to teach it. Why Clojure allows you to put things together quickly. Why side effects can ruin everything. How Clojure encourages composition. Selected quotes: "I don't think you want me to read code to you. That sounds like an excellent way to fall asleep if you have insomnia!" "Maybe we need a white noise track of keyboard noise." "Composition in a functional language is much simpler because we just have functions." "Why have we spent so much time talking about composition when it's just functions calling other functions?" "These names are a little on the nose, but programmers aren't always the best at originality!" "Functional programming is essentially based on transformation. Everything is a transform at some point in time." "The only way to get anything done is to return a new thing." "Core is a backbone, a spine, and you're connecting a progression of things together." "Your job, as a functional developer, is to teach Clojure core about your domain." "They're all functions that remix very well." "Everything must be pure because side effects mess up this whole world." "Make functions small because functional programming makes it very easy to combine smaller functions into bigger functions. Build up your functionality from those pieces." Links: Ep 075: Merge With Fun Ep 079: Compose Thyself Ep 093: Waffle Cakes Ep 094: Concrete Composition

Ep 094: Concrete Composition

Each week, we discuss a different topic about Clojure and functional programming. If you have a question or topic you'd like us to discuss, tweet @clojuredesign, send an email to feedback@clojuredesign.club, or join the #clojuredesign-podcast channel on the Clojurians Slack. This week, the topic is: "decomposition." We help our code through a breakup so it can find its true colors. Our discussion includes: How to process text and add color. When is it time to decompose? Return of the box mix! How to separate transformation from I/O. How to use reducing functions to repeat computation. What makes code orthogonal? The woes of packing useful data into bits. Signs of complexity in the Java underlayer. What is the difference between isolation and decomposition? What are some natural points of separation? Selected quotes: "We are composing ourselves!" "I'm a terminal guy, so this is all in the terminal." "You can go look it up on the Internet. I'm not going to try to speak ANSI right now." "The box mix makes it really easy to make a cake, but it also constricts your ability to do more things with it." "By definition, any side effect is an orthogonal concern." "It's screaming for decomposition!" "Separating these out makes it so that it's easier to understand each part by itself." "We have our 'actual code' badge again as a programming podcast!" Links: Ep 093: Waffle Cakes Ep 031: Eager Abstraction Ep 033: Cake or Ice Cream? Yes!