As you might have guessed, I think I’ve finished the refactoring, or at least the first draft; if I ever have to go back to it, it’ll be much easier to work with.

I don’t have much more to add that hasn’t already been covered in my previous notes! I made a few visual tweaks and did some minor clean-up here and there.

I’m sure there’s still plenty left to do and clean up! Maybe I could consider a dark theme, for example? Or an improvement of the responsive css? Only time will tell.

We’re nearing the end of the second refactoring of the site! I’ve mostly focused on reducing the JavaScript, cleaning up old code, and making a few visual changes that you might notice! The site still has a few flaws I need to fix, and I may have occasionally made the wrong choice between client-side JavaScript and the server-side HTMX paradigm. The result, however, is a thousand times cleaner than what I had before, and it’s much more enjoyable to work with, whether in terms of the pure development experience or simply writing articles.

I’ve chosen to use Web Components on a number of occasions, and one of their biggest advantages is that they free me from having to worry about HTMX restoring history, or listeners that linger and cause memory leaks without ever being destroyed, etc. It’s actually become an evolution of the classes I’d started implementing. Anyway, the CSS still needs a good polish, and I should see if I can reduce the JS even further!

Oh, and I also took the opportunity to add two new tracks to the miniplayer, enjoy!

I’m still in the process of refactoring the site! However, I’ve decided not to use WASM; in short, it quickly became clear that loading the WASM file itself, as well as its glue code, was particularly inefficient relative to its function, namely, manipulating the DOM. The most intensive computations are mainly those on the music page using the WebAudio API, but they remain relatively trivial and don’t justify using all those layers to perform the computations in another language, here in Rust. I also considered several JavaScript solutions such as Surreal, Alpine.js, and Hyperscript, but each had its own flaws that would have made them dead weight for this site.

For example, take Hyperscript; the concept is really cool, and I plan to use it for more down-to-earth website projects that don’t require a lot of extra bells and whistles like this one, but it would have quickly become too limited for my needs. You could say more or less the same thing about Alpine.js, which would have required creating a looooot of data tags, making the HTML unreadable. As for Surreal, the library is really promising and lightweight, but it didn’t really offer a huge reduction in code compared to what I already had (my JS files are still relatively small, after all); I might as well just use vanilla JS. That’s exactly what I’m going to do! I think I can still clean up and simplify several parts of it, for example, by creating a class for the miniplayer instead of writing pre-ES6 JS.

Plus, my project was starting to get out of hand; one crate for the website, one for WASM, one for the elements shared between those two crates, a Makefile to build and watch them without too much hassle, and so on.

It seems my hatred of JS blinded me to the point of making some very inefficient decisions…

I’m refactoring the site again! Although this isn’t as massive a technology overhaul as the last one, I’ve chosen a new templating engine and a much simpler way to manage blog posts, whether on this page or on the dedicated blog page; all I have to do now is create a new Markdown file in the appropriate folder, fill in one or two pieces of information in its header depending on the page it’s intended for, and that’s it! It’s much more convenient than nesting Markdown within JSON strings.

I also considered using frameworks like Leptos, but after looking into it, it unfortunately seems like a terrible JavaScript framework with a hat and a mustache. Rendering is client-side by default (CSR, like React, you can guess what I think of that), and although you can use SSR with it, the interactive parts are handled using “islands,” which use that same CSR-style JavaScript wrapping to render the elements. It took years for web developers to realize, “This is funny, we’re wrapping the entire site in JavaScript that the client downloads and renders on the client side. That’s awful, it makes users download 2,000 kilotons of JavaScript, and all for the sake of most pages that use it for no valid reason, which have nothing but a navigation bar and maybe an interactive form; and what if, instead of wrapping everything, we only wrapped the interactive components?” It’s absolutely fascinating, maybe soon, when all that brain-dead JavaScript has collapsed, people will start building proper websites again.

I’m also taking this opportunity to reduce the awful JS glue code that was there before, as well as the unnecessary complexity caused both by the way I implemented partials with Askama and by everything involved in using partials. As for the inevitable JavaScript, I’m not sure what to do yet, but I’m thinking of trying to use web-sys and related tools to compile a WASM file that I’ll then use! We’ll see what the rest of the refactoring has in store for us.

The new site is now live! I have to say I’m really happy with the result; not only does it perform much better thanks to the rewrite, which went from React and related technologies for HTMX and Rust, but I also find it much more to my liking. I was wondering how to fix the dead spots, and now I’ve got just what I needed! It’ll also be more convenient for me to write articles for the main page and my blog since I’m using JSON. Feel free to make yourself at home and explore my little world!

We’re nearing the end of the migration from the old site to this one! I think there are a few things I could do better, particularly the use of JavaScript ; there might be equivalents in HTMX, but I don’t know enough about it yet to know what’s possible. All that’s left is to create the music page!

I almost got swept up in yet another JavaScript framework. However, the more I work on this project, the more I realize that HTML and CSS, with a little help from HTMX, can be just as powerful, without becoming a terrible unoptimized mess. I sometimes regret that web developers get caught up in this sort of craze for the next shiny framework; CSS alone has made enormous strides and is much more powerful and pleasant to write and read, as this excellent article by Lyra Rebane has shown me. While some communities associated with certain languages seem to be moving in the right direction, the web development community is still wondering whether the latest trendy framework is easy to vibe-code; that’s also why I appreciate initiatives like Neocities, which have inspired me immensely !

It’s been a long time ! I’m writing this as I am completely overhauling the website, but I won’t tell much about it until I’m done with this project ! What I can say though is that this section is far more extensible now thanks to the handy web-articles json file !

Just added this section! For now its implementation is very unextensible, but since it’s my project anyway, I’ll be the only one who may complain about it. We finally have been able to move out from Vercel !! I would’ve said thank you for your service if the CEO wasn’t such a scummy human being. Every step towards self-hosting is a step in the right direction!