The below 'blog' is really just an incoherent ramble from when I was making my app, if you want to read my actual post about this, go here. If you want to play with the app I made, the links are here. Also there's a playlist of various videos here, including a demo of the app.

I plan to do a 'part 2' where I move my app to krell (or perhaps make a new one). I did give it a quick go but found too many issues for my liking (js require doesn't work, can't import js files without making them into node modules, repl needs restarting when requires change). David Nolan has said before that Krell is made specifically for his company (vouch) to make their app, and his aim isn't for people like me to have the best experience making toy apps. Though I will try again in a month or so and see where it's at.

In the meantime I am working on 'site', a Clojure platform built on Crux for making webapps without writing backend code (instead you write an openAPI spec with datalog queries/rules). Me and Malcolm are posting videos of the progress here, though they are just raw streams so nothing polished.

Intro

I started working for JUXT around 4 years ago now, and on my first project we made a mobile app using ClojureScript. At the time, I was told this was a groundbreaking technological achievement, though to be honest, it didn't really feel like it. Doing simple things could sometimes take days due to bugs in the underlying rats nest of tooling, and after 6 months we couldn't even publish any updates because the version of expo we were using was no longer supported, and the new ones used a new version of react native which had too many breaking changes to justify updating everything. 🤮

We abandoned the project after that and never touched mobile apps again, but after reading about Krell, written by the core maintainer of ClojureScript, I became interested again and decided to give it another go.

State of tooling

There are a few ways to make a mobile app with Clojure. This approach uses Graalvm and Sci, and this one uses ClojureScript to build views for an otherwise native app, but currently React Native is the only realistic option if you want something 'stable'. It still won't be as stable as a native app, and nowhere near as stable as a web app, but that is just something that comes with the territory in the native app world.

React Native compiles JS into native code for either iOS or Android targets, so to use ClojureScript, we just need something to turn Clojure into JS that the React Native tools can understand.

<aside> 💡 You can build RN apps for the web too, but don't go thinking that you can easily copy and paste code between 'standard' React and React Native, because React Native uses native mobile components rather than web components. There's no concept of a div, things like buttons are imported from React Native and generally, everything in your views will look pretty different if you're coming from web apps.

</aside>

Here are the current options for tooling then:

• Krell - Very simple, very few external dependencies. Designed to get the CLJS into React Native as 'natively' as possible with no added frills. You will need a Mac and Xcode to test your apps on iOS.
• Shadow-Cljs - Many CLJS developers will already be using this for web projects, but it can build code for a React Native target too. You can either use it with raw React Native or Expo.
  • Expo - A layer on top of React Native that aims to solve many of the more frustrating parts with cross-platform mobile development. With Expo, you don't need to set up Xcode or Android studio. You can test apps on your actual phone just by scanning a QR code, and with Shadow, you will still get a REPL and hot code reloading! You can push code updates instantly without needing Apple or Google to approve them, and you can even develop and deploy an iOS app to the App Store without owning a Mac. However all of this magic comes at a cost, complexity. The main focus of this article will be evaluating the Shadow + Expo combination to see how many issues I run into.
• Re-Natal - uses figwheel to compile the CLJS and adds some extra features for a more 'Clojure like' experience with React Native. However, the cost of these abstractions is that maintaining such a project is time-consuming, and as such, this project is now very outdated (in mobile terms at least).

What's the app?

I wanted the app to be simple enough that I could build the logic for it in a day or two, but interesting enough that I wouldn't lose motivation. I've been getting into meditation recently and thought I would make yet another meditation app, though it does have a somewhat unique twist.

To give an example, there is a popular set of meditation/breathing exercises known as 'The Wim Hof Method'. The main exercise being the following:

Now this can be pretty daunting and difficult to remember for some, and others may find it difficult to stick to without a reminder, so Wim Hof has an app that will play audio clips of each step and keep track of things like how many cycles you have done.

This is also the case for many other activities such as yoga. An app will be made with a series of 'activities' such as 'tree position' with maybe an image showing the position and a voice explaining it. After some duration, the next step will start.