14 December 2020
I created this website about a year ago and it has evolved quite a bit since its inception. From pure HTML to Markdown and a custom static site generator, my focus on a high signal to noise ratio and fast loading times were one of the only constants.
Having an online presence as a researcher doesn't hurt. When I started working on this website, I was applying for funding for my PhD project. When you Googled my name, you would have seen, among others, my GitHub profile and my Master's thesis. I wanted a landing page grouping my research and contact information. I had already made a blog using Jekyll back in 2017, with the purpose of writing rather often. Over the course of time, my definition of often changed quite a bit. After starting my PhD, writing grant applications and papers took precedence over writing for my blog.
As such, when I first created this website, I did not include my blog. It was a very simple website which I wrote entirely in HTML, from scratch. When recently I decided to revive my blog, I wanted to move everything to a single place. I wanted to keep some of my old posts - especially the one I'm most proud of - and didn't want to write future posts in HTML. The need for a static site generator and markdown parsing returned.
Jekyll served me well for my original blog, but after having OS updates break my Ruby install, I wanted something different. Hugo looked interesting, but the extensive documentation discouraged me. I wanted to re-use my old HTML pages and keep my design, without having to learn a whole new system from scratch. Over the past few years, Rust has had a special place in my heart, and the template engine Tera caught my eye. A personal website should be something one can tinker with, and what better way to tinker than to reinvent the wheel?
When people talk about creating their own programming language or game engine, the response is typically discouraging.
Don't do it! You'll never make something better than C/Unity3D! Why are you even trying to compete with a team of 100 engineers?
These responses are correct, in a sense, but miss the point that creating a programming language doesn't have to be about competing with
C. Likewise, a game engine can be tailored to a specific game, and doesn't have to be as general as Unity3D. One can tinker with these
to learn, or because it's fun. The same thing holds for the static site generator I made. It's littered with unwraps (Rust's way of saying, "If this returns
an error just exit") and probably just as many bugs. But I've had fun writing it, it works for me, and if I discover a bug I can easily fix it. And because it's only
for me, and only for this website, it can be as specialized as it needs to be. No edge cases to support, no multiple source languages. Just a solution tailored to my needs.
Sometimes you don't need an entire toolbox, just a hammer and a wrench.
The main benefit about writing the static site generator myself, though, is that I could easily re-use my old HTML code as templates. The blog posts (in Markdown) are first converted to HTML using pulldown-cmark and then inserted into a generic "post" template. The website builds in less than 200 milliseconds. Pages are often only a few kB big, only the blog post pages have Javascript enabled for client-side code highlighting, and the style sheets are rather small too. The only external resources: two fonts, which take the longest to load.
Writing the static site generator only took me part of a weekend. It was a fun learning experience and maybe in the future I can clean up the code and make it available. But for now, it has served its purpose well, as a tool and as a side project.