The old blog was an express server that looked for .md files in a directory, converted them to HTML, and served it. That worked mostly fine. However, the converter didn't support many Markdown features, and was pretty buggy. Using an express server also meant that the site wasn't static, and had limited options for hosting.

Replit's free tier was pretty unreliable, and Render only allowed one free web service per account, which I was already using to run my faucet. I could've created a new Render account, but they recently started requiring credit card verification, which I didn't really want to do.

Besides, making the blog static instead of relying on an express server wouldn't be that hard. A month before, I had already wrote a better (?) or at least more fully featured Markdown to HTML parser, Makoto, so that was already one of the problems with the old blog solved.

Anyways, once I decided to start completely rewriting the blog, I established some goals that I wanted the new blog to accomplish.

Technical Goals

• Static, so it can be deployed by eg, Github Pages or Cloudflare Pages
• Built from scratch with no third-party dependencies (builtin modules like path and fs are ok of course, makoto is not third party, also doesn't count)
• Use Typescript
• No Javascript running client side - the web pages should be pure HTML and CSS
• Style-wise, should be minimalistic, nothing fancy
• Load quickly and be small in size

Non-Technical Goals

• Make two things I can call "Ryuji" and "Saki" to go along with "Makoto" (those are the three main characters of one of the best manga series ever)
• Move over some of the old blog posts (only the stuff I like), after rewriting them
• Start writing stuff on the blog again, at least semi-regularly

Code

Hedgeblog (oh, that's what I'm calling it by the way) is made of three components: Makoto (Markdown to HTML parser), Ryuji (templating language), and Saki (build system).

You can find the code here.

Makoto

Makoto is the Markdown-to-HTML parser, made with no dependencies. It was made around two months before Ryuji and Saki, and is meant to be more of a standalone thing. This is the sole npm dependency of the project. I npm installed it instead of just copying the file over mostly because I published Makoto to npm and wanted to make sure it worked. Also, it has a different license, documentation and stuff.

All the standard Markdown are supported (headers, bolds, italics, images, links, blockquotes, unordered lists, ordered lists, code, code blocks...), as well as superscripts and tables (although tables are probably buggy). Makoto also does some pretty neat stuff like passing on the language of the code block (if given) as a class in the resulting div: code-<language>, or automatically add ids to headers, so they can have anchor links.

It also has a very cool warnings feature, which isn't used in this project, but can be seen in action if you use the Makoto Web Editor.

Ryuji

Ryuji is a simple templating system that supports if statements, for loops, components, and inserting variables. It isn't quite as fully featured as Jinja/Nunjucks, but on the upside, Ryuji is less than 280 lines of code, and worked very well for my usecase. I think it's pretty cool.

Here's a quick overview of the syntax:

After finishing writing Ryuji, and writing some tests to make sure it all worked, I realized that I needed a few features that were missing.

There was no way to see the current index in a for loop, or the max index (length-1) of whatever variable we were looping over. This was needed for the tag links in the post.

In addition, if statements only checked if the variable was truthy (not false or 0 or ""), but I needed if statements making sure two variables were equal, as well as if statements making sure two variables were not equal. You can see this being used in the post's tags along with the new for loop features, as well as the "Next Post" link at the bottom of the post.

So, I added those features. Let's take a look of these new features being used to show and link post tags.

Formatting the code a little nicer, this is what it looks like:

Ok, in the first line (for:post.tags:tag:index:max), we are looping over the variable post.tags, and assigning each item in post.tags as the variable tag. That's nothing new, what's new is the :index:max portion. index is the index variable, starting at 0 and incrementing every loop, while max is the maximum index (the length of the variable to loop over - 1).

If you look at the Ryuji code, now you can see that it is looping over the tags of the post, and creating a link for each tag. If the tag is not the last tag (index is not equal to max), we will also add a comma (and a space).

Here's the equivalent python code, if it helps:

While Ryuji is meant for HTML, there is no reason it can't be used for other formats.

Take a look at the docs for Ryuji.

Saki

Saki is the build system that uses Ryuji templates to generate all the HTML files, and then outputs everything (including static files, of course) to the build directory. Even more simple than Ryuji, it is just around 70 lines of code.

Here are the very short docs for Saki.

Putting It All Together

When building, the program (index.ts) reads /posts/_metadata.json and passes all the post metadata information to the templates/index.html template, which is the home page! Then, it renders all the posts with the templates/post.html template and outputs to /posts/*. Next, it looks again at the post metadata and gets all the tags used. Once it has all the tags, pages for the tags are generated at /tags/*. That page lists all the posts with that tag. Scroll up and try it! It also outputs the static directory, as well, static files.

Tip: Serve Without The .html

Say we want to serve a HTML file at /posts/example instead of /posts/example.html. Just create a directory called example and put example.html inside it, renaming it index.html.

Basically, /posts/example.html becomes /posts/example/index.html, and now the HTML file is served at /posts/example. You probably already knew that, but if you didn't now you know ^{[citation needed]}.

Buttons Without Javascript

If you think back to around 950 words ago, you may recall that this site has zero frontend Javascript (or WebAssembly, as cool as it is). If you didn't recall that, I just reminded you. I will be sending an invoice later.

So how does the "Show MD", "Fancy Title", and Dark/Light theme toggle work? The key thing here is that all three of these are checkboxes. Yes, even the Dark/Light theme toggle is a checkbox. That toggle hides its checkbox and takes advantage of the fact you can click on a <label> with an appropriate for attribute to toggle a checkbox, and uses the ::after psuedo-element to change the content between the moon emoji and the sun emoji, depending on the state of the checkbox.

What's so special about checkboxes? The :checked (see MDN) property, that's what! With the :checked property, we can apply styles depending on whether a checkbox is checked. See the following, which scales the checkbox by 3x when the checkbox is checked:

And because of the a ~ b (select selector b after selector a) and a + b (select selector b immediately after selector a) CSS selectors, we can change the properties of other elements. Unfortunately, there doesn't seem to be a pure CSS way to change the style of elements before the checkbox. There are ways around this, like putting the checkbox before the desired element in the code, but using position: relative, position: absolute, and other ways to make the checkbox visually look like it is after the desired element. I didn't really want to do that though, so you can notice that all the checkboxes on this website are before the element whose style they change.

Here's a real example. The following CSS makes the "Show MD" checkbox functional:

Other Style Notes

I used Verdana for headers, Courier New for code, and Times New Roman for everything else. These fonts were chosen mostly because they are websafe, ie included in most OSes. You cannot stop me from using Times New Roman. I like the look.

The colour for visited links is orchid, and the colour for non-visited links is forestgreen.

Since this blog is basically for my own "enjoyment", and doesn't need to look "sleek" or "professional", this site's design philosophy is moreso a rejection of those overdesigned corporate sites with too many popups, as well as certain React (or other framework) sites that don't even load with Javascript disabled, than a specific set of tenets. Combined with the fact that choosing colour schemes makes me angry, don't expect very consistent or aesthetic designs.

But I'll try my best.

Running

After git cloneing the repo, cd into the directory and install the (sole) dependency:

To build:

To build and preview locally at http://localhost:8042:

To run tests for Ryuji:

Todo

In the future, I would love to have those fun box gifs^[0] you used to see on geocities and other websites (like the ones on the bottom of the pensquid website), plus also something similar to Wikipedia Userboxes.

And I'll keep improving the site and fixing bugs, and occasionally write articles for the roughly four readers of this blog.

===

• [0]: This has been done! Plus, there's a RSS feed now too.

It's difficult to express my feelings on this. Simple living, or thinking about it, tends to give a nice feeling. Thinking of yourself as a person amenable to such things gives the same. Looking at the small stacks of books laying around (and the pile of currently unused but very important^tm laptops) pretty much shattered my delusions of that I was living simply. I couldn't deny that I was greatly attached to my books, and to the idea of having many more.

Digital hoarding, on the other hand, feels guilt-free, as long as more drives don't need to be bought. Why not, then, just download all the books as pdfs?

It's just not the same. It is nice to be able to insert tons of little bookmarks and notes, and flip through and read them and review them. I do have books that are too expensive for me to afford downloaded (thank you Anna!), and the reality is that I just don't look at them often. Only very rarely, and only ever a quick ctrl+f search to find something on my mind. All the software I've used have been just too cumbersome. But surely there are better softwares that can do all that stuff in a probably even more ergonomic manner than physical books. Perhaps paper can still be justified with that oft-cited study about having a physical component to your learning improving memory retention (or whatever). I have yet to see the study in question, mostly because I don't care to search. It does, at least, seem completely reasonable, and intuitively expected. In the end though, it doesn't matter and it is probably better to not have a reason to take it seriously.

In the end, the "feel" of a real book is just not replicable digitally. Is the preference just arbitrary? The reasoning ineffable? No, you can certainly describe that feel: of the paper, of the blackness of the ink, of the lighting, of the sound of pages turning, of cutting strips out of sticky notes and placing them on so they stick out of the top. That's a perfectly good justification on its own. We devalue our own experiences in favour of tryng to quantify things using "science". It is much easier to turn it all into numbers and praise the largest one. Or sometimes the smallest one; I'm told it depends. But we just fool ourselves.

Maybe this line of thinking is getting too far ahead of itself. In the first place, does it make sense to have a strong distinction between digital and physical possessions? Or rather, are digital possessions and physical ones any different in terms of "living a simple life"? Of course, digital stuff exists physically^[0], but you know what I mean. Something something duality, abstractness, etc etc. This isn't a philosophy blog, thank the lord. But we'll take a small cautious step in that dangerous direction. Maybe it is about our attachments to the object? In that case, since many of the digital things are easy to replace, maybe they do weigh less? For example, I would be a little miffed if my mp3 collection got wiped. But I could rebuild it, it would just cost me a few hours, and even if there would be some songs I forget and never add back, that wouldn't bother me all that much.

On the other hand, having tons of stuff, just disposable, doesn't really seem like simplicity. It seems worse than having stuff you hold on to a little too tightly. It just seems wasteful and unthoughtful. If you don't own anything, but you buy or get tons of shit and just throw it away, that's not any different. So then you might consider the net "displacement" of things (as opposed to the "distance"); that is, how many things pass through your possession, however brief.

A more nuanced, though not especially novel, take might be that it is a combination of that "displacement", as well as the lifespan, attachment and mental attitude toward that thing. But whoops, are we trying to make numbers again?

Or maybe this entire line of thought is overcomplicating things, maybe living simply is just the bare minimum you need to survive. You know, sleep in a cave or something. That certainly is simple, but it sounds unpleasant. Though the blurry image of "simple living" in my head is probably similarly unappealing to others^[1]. Surely one could become accustomed to an extreme ascetic lifestyle and begin to enjoy it. I am reminded of the romanticised rail vagrants hopping trains with their cartoon-style sack on a stick and some breed of stringed instrument. Difficult of a life as it was, the romanticism wasn't totally unjustified. After all, much of it was created by the vagrants themselves.

All that is to say, I really don't have a real definition for what a simple but good lifestyle would be. What then?

Well, having such a poor understanding of my current and especially future selves, I struggle to think of anything better than just experimenting. Cutting things out, seeing how life is. If it isn't better, then putting it back in. No different than diagnosing any other problem. Just like hunting down bugs in the code or rabbits in the heat. Happy hunting.

===

[0]: Electrons and such, amirite?

[1]: Much like the monastic "middle way" of the Buddha was in relation to extreme ascetics, and so is super mega different from modern living today.

QED, destroyed with facts and... aw, damn.

mi open tawa lon suno open. mi pini tawa lon pimeja open. noka mi li pilin ike. lawa mi li pilin lape.

mi lukin e nena wawa e kulupu kasi wawa e jan wawa e tomo wawa. mi pilin e ni: "pilin ike ali li lili tawa mi. ali li pona lukin a! noka li pilin ike la lawa mi li pilin lape la tawa mi li awen ken." tan ni la, mi awen tawa.

jan li pana e sona tawa mi: "tenpo poka kama la telo sewi li kama." mi wile ala e telo sewi. mi weka e toki ona tan lawa mi.

tenpo poka kama li kama. telo sewi li kama ala. mi toki insa e ni: "sewi o, pona tawa sina a!" mi awen tawa.

tenpo suno mute li pini. tenpo ni la, kon pimeja li lon sewi. mi lukin e ilo mi la, ilo mi li toki e ni: "telo sewi li kama." mi wile ala e telo sewi. mi weka e toki ona tan lawa mi.

tenpo ona la, toki ona li lon. telo mute li tawa anpa tan sewi. mi jo e len pi weka telo. mi pana e len ni lon sijelo mi. len ni li awen e sijelo sewi mi. len ni li awen ala e sijelo anpa mi. poki noka mi kin li ken weka e telo. poki noka mi li lon telo la insa pi poki noka li kama ala telo. ni li pona mute tawa mi. mi toki insa e ni: "telo sewi li pona ala. telo mute li lon nasin. tan ni la, nasin li kama ike. taso, mi jo e len pona e poki pona. sijelo mi li ken ala kama telo. ni li pona." mi awen tawa.

taso, mi kama sona e ni: len anpa mi li weka ala e telo a! sijelo anpa mi li kama telo. telo li tawa anpa tan sijelo anpa mi, li tawa insa pi poki noka mi. noka mi li kama telo mute. len noka mi li kama telo mute. mi tawa la len noka mi li kalama: "mu mu mu." noka mi li pilin telo li pilin nasa li pilin ike mute.

mi pini tawa lon tenpo lili lon tomo. mi weka e poki noka e len noka. len noka li lon luka mi. luka mi li weka e telo tan len noka. telo mute li weka. mi pana sin e len noka e poki noka tawa noka mi. len noka mi li awen pilin telo. mi awen tawa.

tenpo lili la, len noka li kama telo mute sin. mi tawa la, len noka li kalama: "mu mu mu." mi pini tawa sin. mi weka e poki noka e len noka. luka mi li weka e telo tan len. mi awen tawa.

mi tawa lili. mi kama telo. mi pini tawa. mi weka sin e len. mi weka sin e telo tan len.

pimeja li kama. mi tawa tomo lape. tenpo suno ni li ike mute tawa mi. mi pilin ike mute. mi pilin e ni: "ken la, mi wile ala tawa lon tenpo suno pi telo sewi. jan li wile tawa la tenpo pi telo sewi li tenpo ike. sewi o, sina o weka e telo sewi."

tenpo suno sin li open. mi open tawa. telo sewi li tawa anpa. len noka mi li kama telo. len noka li kalama: "mu mu mu." taso, mi pini ala tawa. mi pilin nasa. mi awen tawa.

mi tawa la, pilin telo li weka tan mi. mi lukin e ma e kasi e tomo e sewi. ali li sewi lukin. kute mute li lon. kon li pona. mi pilin e ni: "ni li ike ala a. telo sewi mute li lon. len mi li telo mute. taso, mi pilin ike ala."

nasin li awen ike tan telo sewi. tenpo la, mi toki insa e ni: "mi lukin pona ala li tawa pona ala la mi ken moli lon nasin ni." ni li tan telo sewi. taso, telo sewi li ike ala. telo sewi li telo sewi. telo sewi li pona e ijo mute li ike e ijo mute. ijo ali li sama ni. mi awen tawa.

toki wawa li kama lon lawa mi:

"mi wile ala pilin ike. telo sewi li lon la, ijo tu li ken:

nanpa wan la, telo mute li lon sijelo mi, mi pilin ike mute.

nanpa tu la, telo mute li lon sijelo mi, mi pilin pi ike ala.

mi tawa la mi pilin telo. ni li lon. taso, pilin mi li ken pona li ken ike. mi lawa e pilin mi."

tenpo suno tu la telo sewi li awen tawa anpa. tenpo suno tu li pini la, telo sewi li pini li kama ala.

The big new thing is support for forms with text inputs that send a GET request on submit. For me, this primarily means Wikipedia's search box is now supported. Yay!

The interface and UX has been overhauled to be more similar to Malvim, the Vim-like text editor bundled with ming-wm. Some bugs and annoying things were fixed. The usual. Here are some fun images, the same ones in the README.md:

I plan to mostly use it to browse Wikipedia and HN, so I'm not too worried about the browser becoming useless^[0]. It's quite unfortunate that so many sites cannot be simply rendered, or even don't work without Javascript. I really miss the many sites that basically only used the <p> tag, and maybe some <b> and <img> if the economy was good.

You can still find a decent amount of these simpler sites, especially on university subdomains, where older professors (rightfully) simply don't see the point in going beyond basic readability, but there's obviously more of these sites going down every year than coming up. Though, things like neocities seem to be "cool" again, in certain circles. Most neocities sites wouldn't really be readable on Koxinga, but another recently popular site, bearblog.dev is perfectly usable with Koxinga.

What is interesting is that unlike the other obstacles that anyone trying to use a simple browser (eg, the Cloudflare challenges, Javascript loading the content instead of it being in the static HTML, etc), there's no one to really blame. Modern web development is awful, but having only text-based sites (like in the Gemini protocol) is too restricting.

So I don't think Koxinga is the ideal browser in an ideal world. I would have loved for it to support more complex formatting, I just don't really want to work that hard on it, and I'm not really sure how I would do all this complicated rendering anyways. It does what I want.

...By the way, the ideal browser, in my opinion, would be Lynx.

===

[0]: It is a great POC app for ming-wm, anyhow. Plus, the development process for Koxinga led me to add several things to ming-wm itself that I realised were kinda important, like font data caching for faster text rendering.

===

Update 19/08/2026

POST forms and cookies mostly work! The commit has been pushed and it'll be in the next release, whenever that will be.