Next.js and React are the problem here. Adding another layer adds complexity. To simplify, you should remove complexity. To comprehensively address the problem you should strike at the root causes, rather than adding yet another layer of duct tape.
It is not difficult to get away from that ecosystem, they simply need to not be included. It is in fact more work to include them and introduced unnecessary complexity.
Several markdown to html generators exist that are well established which simply use the base languages: node, python, go, etc. Instead of using a framework that isn't suited to the task, they focus on structure and features and work on that. By working on the features, structure, and grammar, the tooling will emerge.
2. For turning markdown into html on the client, write a custom component that parses the markdown into DOM nodes allowing the page author to simply do `<dynamic-markdown remote-src=some/path/to/page.md>`. You can even use one of the many existing Vanilla JS libraries to do this within the custom component.
In either of those two options above, what value does React add?
It’s the I-sometimes-want-to-use-React-components issue that is my holdup. They make it easy to do some visual/behavioral stuff and are often directly portable from other sites or oss projects.
since it directly injects the resulting html, you may set "unsafe HTML" option in marked (or markdown.js) and include <title> tags within your markdown document. this will _also_ rewrite the browser title(s)
These look really promising! Can I bring existing React components with me to any of them? That might be the biggest hold up. I’d also worry about maintainability as web standards evolve, since these (mostly) seem like pretty small projects.
I was expecting a sort of MD to HTML conversion since it's hard to think of a simpler way of doing this. I don't see how react or nextjs constitutes this being easy or simple unless you're targeting folks with domain knowledge with those ecosystems.
I was expecting a really simple static-site generator too, but this idea is interesting as well. The heavyweight aspect of using Next/React dampen things a bit.
If it is a client-side solution, I'm imagining a single index.html file (with inline JS/CSS) that you could have a content/ or markdown/ child folder for. The HTTP server would have directory listing on for that folder. The index.html file enumerate/fetch the markdown in the content and folder and render it. You could route using the URL hash.
This is exactly how my (personal) page(s) work! I wrote it pre-covid.
Initially I was using the hash as you suggested, but the redirect/404 was flaky when I use github-pages as the host, hence I reverted to the query-string (? or search) as it creates a complete refresh. Which allows me to set up "virtual" routes like `/blog`.
Topping with worker/service worker, it is quite robust!
You’re getting close. The biggest thing is that I want to be able to use React components sometimes and anywhere on the page. If you just want md -> html, there are of course much better options.
I actually previously used Jekyll! Built this largely because I want full React component functionality sometimes. Also I think Jelyll gave me some issues with routing that I didn’t like.
It has server-rendered mdx, client components, inline footnotes, layout bleeding, server-rendered code syntax highlighting, content-collections, and opengraph image generation. It can be rendered fully static or from a server. If you choose to deploy it as static, the "server"-rendering just happens at build-time.
It's a really decent starting point for someone who wants an efficient, lean, powerful, and flexible blog.
obviously it's deeply unhinged to suggest this is method is easy, or indeed a good idea at all, but if you want actually easy ways to do this, Zola and eleventy are both actually quite easy and will get you OK looking static html output in a few minutes of effort.
True, and I was going to comment the same thing, but I think this project is about explicitly allowing React components - which presumably means some level of interaction higher than eleventy can provide.
Neato indeed, but the color contrast for copy and other elements doesn't meet AA accessibility standards (even my young-ish eyes have a bit of trouble). Still, this is in line with a project I've been meaning to do, so this is pretty great stuff.
I like Quarto¹ which supports authoring websites, blogs,² books, and presentations. It also supports publishing directly from Jupyter Notebooks, if that's the type of work you're doing. My only (minor) complaint is there isn't much of a thriving theming community for it.
The output is just a bundle of files for your web server of your choice, supports good navigation with the sidebar, custom theming if you need it (but otherwise gets out of your way), alongside being a single executable.
Looks great, but it is clearly crying for "yaml front matter", eg: `import ...\n----\n# Content ...`, either that or `<?mdx/php ... ?>` tags... ;-)
Seriously, love this simplification of content, and taking care of all the hairy details around `<CustomComponent>...</...>`. It's what the web should have been all along!
If you take an `*.mdx` file and run it through a typical markdown renderer, you'd get something like... well... lemme give the counter-example: if you had the "dashes" delimiting your "frontmatter" from the document, you'd get something like:
<hr>import foo, bar, baz<hr>\n<h1>Markdown</h1><ul><li>...etc...
...the critical part being to relatively unambiguously demarcate "junk" from "content" in the markdown ecosystem. In the far future when this project is but a forgotten memory of time, people can still just straight up render the markdown and "obviously" delete the stuff between the <hr's> up at the top.
The contrarian is that there's ambiguity as to when your `*.mdx` "import junk" stops and when the actual content starts. Do you stop parsing [for imports] after the first blank line? Do you stop parsing when you get a syntax error? Do you only parse lines that begin with [import ...]? What about [from ... import ...]?
...strong recommendation: demarcate "headers" from content (a-la HTTP). Your format effectively becomes `---\n $PROG_LANG_GOES_HERE ---\n $CONTENT_HERE`, and you slide 1000% into "pure HTTP":
GET /foo.mdx
Content-Type: application/mdx
Date: 2020-10-10 01:02:34
---
import SomeComponent
---
<SomeComponent>My Header</SomeComponent>
* Some
* Content
I highly recommend Astro and Starlight for building websites from Markdown. It is very simple to set up and can grow with you as your needs become more complex.
I wanted a dumb & simple way to make a modern website with text files, so I made one. Small, static-first Next.js boilerplate, intentionally minimal, just markdown → website, with optional React components when you want them.
Just updated it; what’s new:
• Next.js 15.5.2 + React 19.1.1
• Better typing for error boundaries (React 19)
• Robust handling for Next 15 async params
• Updated DaisyUI, MDX, and tooling
Use it for docs, personal sites, landing pages, or simple content sites that still want modern Next.js ergonomics.
Well since markdown has been designed to be relatively comfortable to read in its source form, I initially thought this link would be an opinion piece on serving raw markdown pages using the browser ability to show text files as is.
I meant add a `<script src=...>` to the `<head>` section of the page.
Not perfect, to be sure, but it lets you create a custom element `<mark-down src=...>` that lets the page author sprinkle markdown everywhere in their page.
I have a fairly old mental bookmark for this purpose: blot.im. Wondering if it still holds up, or perhaps newer entrants can do the same but more elegantly.
Seems like markdown is back in vogue. The more the merrier as far as I'm concerned.
If you prefer to generate a static site, take a look at https://soupault.app/
It generates a static website from markdown or other formats. Mentioning it here because it deserves more attention than it currently gets.
Do you have suggestions on a better alternative?
Several markdown to html generators exist that are well established which simply use the base languages: node, python, go, etc. Instead of using a framework that isn't suited to the task, they focus on structure and features and work on that. By working on the features, structure, and grammar, the tooling will emerge.
1. I want to edit the pages sometimes, and I want to do so using markdown without extra steps.
2. I still want to use React components *sometimes*.
3. I want it to be really easy to deploy and monitor. Nextjs is really easy for those things.
Solving the first issue and third issues don’t require React, but the second definitely does. So I’m still stuck.
Why? What specifically are you using for "website with markdown" that needs React?
> Do you have suggestions on a better alternative?
Yes.
1. For turning markdown into html on the server (i.e. render the HTML on the server and then deliver it) use pandoc. I use it like this for my blog: https://gist.github.com/lelanthran/2634fc2508c93a437ba5ca511...
2. For turning markdown into html on the client, write a custom component that parses the markdown into DOM nodes allowing the page author to simply do `<dynamic-markdown remote-src=some/path/to/page.md>`. You can even use one of the many existing Vanilla JS libraries to do this within the custom component.
In either of those two options above, what value does React add?
example:
since it directly injects the resulting html, you may set "unsafe HTML" option in marked (or markdown.js) and include <title> tags within your markdown document. this will _also_ rewrite the browser title(s)> edit: formatting
[0]: https://unsuckjs.com/
Just try to make something without React and you light see the light. It's actually not difficult.
If it is a client-side solution, I'm imagining a single index.html file (with inline JS/CSS) that you could have a content/ or markdown/ child folder for. The HTTP server would have directory listing on for that folder. The index.html file enumerate/fetch the markdown in the content and folder and render it. You could route using the URL hash.
No build process. Just a single file + your markdown content. Maybe I just like this kind of stuff (https://github.com/pseudosavant/player.html, https://github.com/pseudosavant/folder.api)? So tempted to vibe-code an example...
Initially I was using the hash as you suggested, but the redirect/404 was flaky when I use github-pages as the host, hence I reverted to the query-string (? or search) as it creates a complete refresh. Which allows me to set up "virtual" routes like `/blog`.
Topping with worker/service worker, it is quite robust!
Link: https://mert.akeng.in/
But isn't Hugo or Jekyll the easy way to turn markdown into a website?
Jekyll was literally made by github to have a simple way to turn a markdown page into a github pages site. Wasn't it?
Just put an index.md in the root of a empty repo and flip the right flags in Github and have a static site.
https://github.com/obaqueiro/ezmdpage
And inspired by it made a similar thing for mermaid.
https://github.com/obaqueiro/mermaid-js-auto-renderer
Great to add /docs into code repos.
You can see it here: https://rsc-mdx-blog.saewitz.com
It has server-rendered mdx, client components, inline footnotes, layout bleeding, server-rendered code syntax highlighting, content-collections, and opengraph image generation. It can be rendered fully static or from a server. If you choose to deploy it as static, the "server"-rendering just happens at build-time.
It's a really decent starting point for someone who wants an efficient, lean, powerful, and flexible blog.
It is open sourced here: https://github.com/switz/rsc-mdx-blog-starter
[0] https://saewitz.com
1. https://idratherbewriting.com/learnapidoc/pubapis_static_sit...
¹ https://quarto.org/
² https://quarto.org/docs/websites/website-blog.html
The output is just a bundle of files for your web server of your choice, supports good navigation with the sidebar, custom theming if you need it (but otherwise gets out of your way), alongside being a single executable.
Seriously, love this simplification of content, and taking care of all the hairy details around `<CustomComponent>...</...>`. It's what the web should have been all along!
Precisely! Thank you.
Can you share more about what you have in mind for frontmatter? I don’t really understand the need/use cases yet.
Basically:
eg: https://docs.github.com/en/contributing/writing-for-github-d... ... https://jekyllrb.com/docs/front-matter/ ... and google search for "rst yaml frontmatter"If you take an `*.mdx` file and run it through a typical markdown renderer, you'd get something like... well... lemme give the counter-example: if you had the "dashes" delimiting your "frontmatter" from the document, you'd get something like:
...the critical part being to relatively unambiguously demarcate "junk" from "content" in the markdown ecosystem. In the far future when this project is but a forgotten memory of time, people can still just straight up render the markdown and "obviously" delete the stuff between the <hr's> up at the top.The contrarian is that there's ambiguity as to when your `*.mdx` "import junk" stops and when the actual content starts. Do you stop parsing [for imports] after the first blank line? Do you stop parsing when you get a syntax error? Do you only parse lines that begin with [import ...]? What about [from ... import ...]?
More google searches: `http header demarcation rfc`, and eg: https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/...
...strong recommendation: demarcate "headers" from content (a-la HTTP). Your format effectively becomes `---\n $PROG_LANG_GOES_HERE ---\n $CONTENT_HERE`, and you slide 1000% into "pure HTTP":
I've been considering Spinal[0] (no association) to bridge that gap, any others I should be looking at?
[0] https://spinalcms.com
https://starlight.astro.build/
I consider my own static site generator [1] much simpler than this. Around 250 SLOC with 4 dependencies (markdown2, pyyaml, jinja2, Pygments)
[1]: https://github.com/oxalorg/genox
Push updates to the md file and it rebuilds itself.
https://github.com/automationwise/awise.github.io
Just updated it; what’s new:
• Next.js 15.5.2 + React 19.1.1 • Better typing for error boundaries (React 19) • Robust handling for Next 15 async params • Updated DaisyUI, MDX, and tooling
Use it for docs, personal sites, landing pages, or simple content sites that still want modern Next.js ergonomics.
That is not hard to do, if you're okay with adding a script tag in your header.
Not perfect, to be sure, but it lets you create a custom element `<mark-down src=...>` that lets the page author sprinkle markdown everywhere in their page.
You are welcome.
Hmmmmmm no
[0]: https://imgur.com/a/ysgJbCp
Seems like markdown is back in vogue. The more the merrier as far as I'm concerned.