Frontend log: embracing minimalism

The meme of the land is “new JavaScript framework shows up every week”. As someone who only touches the frontend web development when I have to, I don’t have a counterpoint to that statement.

I did, however, spend 2021-2022 looking into frontend development more than any year prior. Since this is the first installment, let’s take a look at the highlight moments.

• 2015: In college and my brain short-circuited when I first saw Babel. Being the wee lad that I was, I painted with too wide of a brush that “frontend web development is not for me”.
• 2018: Day job was primarily on a Django application with HTML CSS, though it was mid-migration to Rails microservices with React (don’t ask). Somehow landed myself 1 JavaScript ticket for fixing a bug in Internet Explorer (sigh yep) and 1 React ticket for fixing a bug that really should have been tackled in CSS.
• 2019: Went hard to the Kubernetes-land. Frontend meant JSON, YAML, or CLI binary there.
• 2020: Panini Pandemic lockdown got me hooked in Nix and NixOS, straying further from frontend web development.
• 2021: Had heard things about Svelte. SvelteKit was close to v1 and it was pleasant! I made several apps with it, one of them was Noms (I found out today it’s half broken, oh well). Oh, I also got hooked on TailwindCSS and made Cloneponents as an exercise.
• 2022: Did several projects of web development with Rails. Some are Go, though building a whole app with Go’s html/template emotionally drained me. Played around with Deno and it’s neat. By end of year, I thought day job had a chance to be Rails, but ended up Django again by accident (with React).

Given it’s early 2023, where am I at now? Well, roughly…

• Material UI looks nice on the docs, though writing it is challenging. Plain CSS flexboxes just makes sense to me and suddenly I have to work on some abstraction that breaks CSS semantics?
• React and NextJS gave me all the flashbacks of my Babel experience. I just got better at pretending to not hyperfocus on the confusing parts.
• I don’t know how any of you managed to do this prior to TypeScript. Surprising that Microsoft became the voice of reason, but not so strange knowing they acquired GitHub.
• MDN Web Docs maintainers deserve all the raises and promotions they can have.
• Plain HTML + CSS still rocks. Ultimately that’s what all web browsers understand!
• I’d like to see Deno take off, mostly because I look forward to never seeing node_modules/ ever again.
• AlpineJS is quite neat, actually. If the application is mostly 1 action per page, I think AlpineJS makes a lot more sense to be used as full blown server-side frameworks would not be worth the complexity¹, IMHO.
• TailwindCSS has a CLI binary distributable and it’s pleasant to use when the application does not need anything else from NodeJS world.

Okay, the last two is kind of redundant as I just wrote about them, but they both really change the game — Rails used to be an easy default decision for me to pick, but now that the dreadful part of writing server in Go can be solved by writing plain HTML with AlpineJS + TailwindCSS, Go and friends might become the new default².

You may notice that I often take the alternate paths on my personal projects, compared to my day job stack. The justifications for both sides are actually similar: “increasing collaboration productivity”, where one is many different people at the same time while the other is same person at many different times. Simple is subjective and contextual.