Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Markdown page – Create a web page with just markdown (github.com/oscarmorrison)
169 points by oscargeorge on Oct 6, 2018 | hide | past | favorite | 84 comments

Props for the simplicity in getting something published. That's really cool.

Along similar lines, I went back to LibreOffice recently and made a few web pages with it, just to try it out. It worked better than I expected.

After adding a simple PHP controller to the mix, injecting some CSS for responsiveness and other tags as needed, I have been impressed with how easy it makes publishing. If a single-pager is what's needed (or multi-pagers perhaps; I just haven't tried it yet), if what matters most is getting links and information public, it's a useful tool.

This is kind of neat but why not just use pandoc?

pandoc -s input.md -t html -o output.html


Pandoc is kind of neat but why not just use markdown?

    markdown input.md > output.html

It is a personal preference but I would rather know one tool they is more versatile (especially for such a task)

Why would you want to constantly recompile documents every time you make an edit to them? Why would you want to juggle two files for every document you produce?

For what it's worth, this is the problem GitLab Pages solves. A simple "git push" to a properly to figured repo will rebuild and redeploy your static site, the actual html output doesnt go into the repo.

This is for people who don't want to setup and manage a deployment pipeline hooked up to a git source code repo.

Sooo a live editor like this? http://www2.hyperfiddle.net/:markdown/ If you like that, here's a little tweetstorm about hyperfiddle markdown: https://twitter.com/dustingetz/status/1048686717975388160

Wow, thank you for this. That's quite incredible.

Scriptable Word to HTML sounds so useful.

Exactly my thoughts. That’s where I landed for my site, and it’s really simple. Edit file, pandoc, scp to server, done.

Thanks everyone for taking the time to look, and give some feedback. Most definitely better options out there, but I was really aiming for simple, and low tech knowledge to get a md page on the internet.

I added a wiki page https://github.com/oscarmorrison/md-page/wiki explaining a couple of use cases, and some more details. Feel free to leave more feedback / open issues. Thanks Oscar!

There has been Strapdownjs[1] around for quite a while now. It's simple, supports some theming as well. Anyway, good luck with this. Always fun to build something from scratch.

[1] - http://strapdownjs.com/

This is more evidence to me that <xmp> needs to not go away, I know browsers are considering it and I really hope they don't go that route.

I use <xmp> for html code commenting - see the source code at https://elementcss.neocities.org/

The one thing these markdown examples are missing is a way to do grids! I bet there's a clever way to make columns and rows work.

I like it a lot, both concept and execution. Little nitpicking:

> Create a webpage by just writing markdown, nothing else!

The „nothing else” bit looks inaccurate, there’s a script tag at the very beginning.

Nitpicking more: raw html (including script tags) is valid markdown! So it technically is markdown and nothing else.

Noted! Then again, super nitpicking, if that script tag in the beginning is part of the markdown document, then it should also be displayed in the final document.

haha. thanks @boffinism

Thanks @posamari for the feedback, I was a bit generous with the title. I have updated HN title and GitHub Readme.

Throwing in to the list of alternatives to this: http://mdwiki.info which actually does not even require you to create a .html file, but just plain .md files and a standard/template version of mdwiki as index.html.

Disclaimer: I am the author of it, and it is nowadays unmaintained.

Why did you archive the repo?

A neat party trick, but what actual use-case is there that isn't better served by server-side Markdown to HTML conversion?

Here are three use-cases:

1. You are editing documents locally on your computer. This lets you double-click to view nice markdown in a web browser with a `file://` url.

2. You are editing a website, but don't have control over the server-side code.

3. You don't want to install a markdown rendering library on the server.

> You are editing a website, but don't have control over the server-side code.

You don't have control over the server-side code, but it will let you inject <script> tags into the content? That seems a bit weird.

GitHub pages is literally like this, no?

You're right, though GHP is a bad example since it includes a Markdown renderer.

Back in 2013 I did a similar thing for a client with a popular code library.

Their documentation was written in markdown and the site was hosted on static hosting, so no server-side conversion.

Although the web page chrome (navigation, etc) was in HTML, they wanted the documentation to be in markdown under the hood after upload. They didn’t want to use any kind of conversion prior to uploading because they needed to make on-the-fly changes to the documentation after upload without writing in HTML.

So we opted to go with client-side markdown to html conversion on pageload. complete with a clickable legend and tabs to switch between language implementations for code examples.

Putting the markdown inside regular page formatting tags wasn’t reliable, as different browsers did different things to the content (for example some browsers converted “&” to “&amp;” and the indentations for code was not respected). So we had to store the markdown in some sort of preformatted tag block.

However <pre> required special escaping of code characters ( <, >, &, etc).

So we opted to go with using a little-known <xmp> tag to store the markdown as search-engine indexable code. Since content inside <xmp> will be read as plaintext instead of being formatted by the browser, it was also great for being the input of a client-side markdown-to-html converter.

It worked quite well. If you want to see it in action it’s at https://www.genivia.com/dev.html

Sometimes I just want to get some information out there on the internet, and rather than sharing a Evernote page or having to create a webpage I can just write markdown.

Main use-case is when I want to share a markdown doc in a private GitHub repo, with a public url e.g http://oscarmorrison.com/md-page/oimasdoijasdmadeupurl-01231...

Very cool. Thanks for sharing. Btw, I got an SSL certificate error when I went to check out your site. I know there's a few different ways to account for this when serving up a website from github, but I'd welcome guidance on what the best practice is for fixing this.

Hey @jonnydubowsky thanks for catching that. Just fixed it all up now, so should be good. This is what I did to fix https://gist.github.com/cvan/8630f847f579f90e0c014dc5199c337...

I really like this, thank you for writing and sharing it!

I take notes and write personal documentation in Markdown for tickets and projects. Nowadays I usually write in Visual Studio Code or Vim. I can see myself using this to share pages with people. Although I wonder how my syntax highlighting will work - I guess I'll have to write a shell script to copy the .md file to .html and insert the call to the md-page.js at the top.

I think this has already been done by Jr [1].

And for other markup languages, e.g. Dr [2] for AsciiDoc.

[1] https://github.com/Xeoncross/jr [2] https://github.com/jonathonf/Dr

Honestly, I prefer writing directly plain html5 with implicit tags. I cannot see any advantage of markdown over html for this particular usage.

Markdown doesn't require tag closure, and eliminates the <p> tag entirely. This makes for a cleaner and faster composing experience.

I did little but wruite in HTML from the late 1990s to mid 2010s. Markdown is my preferred route now.

Also conversant in LaTeX, formerly; nroff, DocBook, WP4.2, Wordstar (both with 'reveal codes' capabilities), and probably more. Pandoc is a game-changer.

> Markdown doesn't require tag closure, and eliminates the <p> tag entirely. This makes for a cleaner and faster composing experience.

The HTML5 specification also allows omitting most closing tags (except for the obvious ones, that you also need to close in markdown).

You are right about the <p> tag, still. It is the main source of annoyance when writing long documents in html by hand.

Well-formed HTML5 needs closure though, no? Hrm ... ok, looks like specs are far looser than I'd thought.

h[1-6], li, blockquote, pre, can all be specified via an opening-only syntax.

p, ul, ol, and dl can be omitted entirely.

Arguably anchors require closure, but the syntax is still briefer than HTML. Images likewise.

Header, footer, and aside don't exist (they can be supplied in a template). article, div, and span are also ignored.

Italic, bold, strikeout, code, subscript, and superscript require a lightweight closure, and are actually the main cause of grief in my Markup writing.

Tables are hugely more sane.

And footnotes automajickally exist. Swoon!

It's not that HTML is foreign or annoying to me. It's just .... more frictional.

And I have all kinds of nice words to say for HTML5. The base spec is quite good.

It's easier to read? It's easier to write?

it is just as easy to read and write as markdown, and it does not require javascript libraries to be rendered

Is there a WYSIWYG simple editor for creating documents in HTML5? Something that's basically wordpad and produces pages that actually look the same on different browsers? Documentation for my projects is currently in raw ASCII because I couldn't find a satisfactory solution.

If you edit html inside a modern editor it colorizes the output in a useful way. You see the headings, and the basic formatting in a rather clear way. For simple html (as the one you would expect from markdown conversion) this is more than enough.

WYSIWYG editors are evil as they mostly produce crap code, good news are you don't need a WYSIWYG editor with markdown as it's very easy for [almost] everybody to write by hand.

I dunno, I like the WYSIWYG editors built around Draft.js[0]

0: https://draftjs.org/

The flicker when loading the page sucks. I guess displaying Markdown in case of failure is not the worst thing, but it would be nice if there was a possible way to allow getting rid of the flicker at the cost of not having a graceful fallback.

One way you could accomplish this is by supporting script tags for this, like wrapping the page in:

    <script type="text/markdown">...</script>
...as an optional way to use the library. Browsers won't (shouldn't?) try to render the contents of the script tag before you get to it, so it would act like a typical SPA.

Yesss, finally a use-case for the `<noframes>`-hack! This is actually even better than the approach of OP since we're guaranteed to get a plain text representation of the text (instead of the browser parsing it as HTML).


    <script src="mdpage.js"></script><noframes>

    # Header
    Welcome to my simplest site

    - A
    - awesome
    - list

    document.write("<div id=main></div>");

    document.addEventListener("DOMContentLoaded", function() {
      var script = document.createElement('script');
      script.src = "https://unpkg.com/showdown";

      var noframes = document.querySelector('noframes');

      script.onload = function() {
        var converter = new showdown.Converter({
          emoji: true,
          underline: true,
        main.innerHTML = converter.makeHtml(noframes.textContent)

Thanks @judofyr + @matt-tingen. I just updated it to use the 'noscript' opening tag. Good idea, and solved my normal html not working problem! Thanks a bunch!

An option for the best of both worlds would be to wrap the markdown in a <noscript> tag.

fixed the flicker, was a Cloudflare rocket script issue

I wonder what effect having a page solely in markdown has on SEO? Without markup for headers, links, etc, how will search engines prioritize content? Will they use the raw content or the parsed HTML?

I'd imagine anyone particularly concerned with SEO would be likely to prefer a more fully-featured method of content management anyway.

One of the problems of the modern web.

Sure we can add JavaScript and make the web work the way we want, like this example, but many more... but what about SEO?

For example, I had a page for my app, it was very minimalistic and pretty, I targeted the 400 msec cold loading time, instead of a screenshot, I had a very well encoded h.265 screen recording (it's okay, people who can run the app, can certainly see the video), that video didn't cost more than a well optimized JPEG.

But I had to drop the video, and only load it after the JPEG loaded because of SEO reasons. Google isn't smart enough to recognize the video unless it's a YouTube video.

I looked for something exactly like this a while ago. Thank you very much for posting this.

Awesome, nice work! I like the simple usage of just adding a script tag to the top. I also experimented with this idea[1].

[1] - https://github.com/hachibu/md_spa

http://oscarmorrison.com/md-page/ shows only a blank page on firefox.

I even disabled NoScript for the site. Thanks, I'd rather have plain HTML.

Just tested in on FF 62. Seems fine. But yeah I have really only tested it on chrome

Works fine for me on Firefox.

Sorry to sound cynical but this 10-15 minutes of work?

time well spent

So how does this work exactly?

It looks like you wrote a script.js file, which calls showdown.js library (converting markdown to html). A uglified version called `md-page` is made pulling both resources

Basically, the script does following

1) wait for page to load

2) add css-styles via javascript

3) shove all the markdown in a variable "markdown", passing it as one giant string

4) Add settings for showdown.js

5) Rewrite entire markdown on page to HTML via showdown.js

6) Append to body and render

I could fork it and add breakpoints / run debugger but this is what I got from looking at source code

yep. you're 100% on the money. no wizardry here. Couple of extra features like ability to hash link, and external url open in new window. but that it


Line 43 to 48, I am not entirely sure what that hashlink does. What is the setTimeout function for? Could you expand on that?

It's to make linking to headers work. Eg oscarmorridon.com/md-page#code

As the element doesn't exist at time of first load this is a work around to focus on id once content is rendered in html

Interesting. Out of curiosity, does anyone know if is there a webserver that natively converts Markdown to HTML?

Caddy server has this feature baked in: https://caddyserver.com/docs/markdown

I wrote a web server that converts markdown to HTML dynamically. Not sure if it's what you're looking for but it works decently well. I host my personal site on it.


I wrote one in Go called Lanyon, available at: https://github.com/mkaz/lanyon

It works pretty well, but fairly bare bones and haven't updated it in a bit.

Here's one http://harpjs.com/

Jekyll and Hugo do this statically.

I don't know about dynamic webservers.

[Markdeep](https://casual-effects.com/markdeep/features.md.html) has waaay more features than this.

So does HTML/CSS+JS. A project aiming at simplicity probably isn't one to criticize for lack of features.

the mechanism for using both the projects is the same -- include a JS file. If the effort to use both is the same, why would I not use the one that has more features?

Maybe he doesn't need the features, or maybe it is slower

I always chuckle when I see people trying to use Markdown in HN.

Here [1] is the list of "features" HN supports in the comment section.

[1] https://news.ycombinator.com/formatdoc

First I thought your "waaay" is a bit to much, but man, that's I call maaaany features!

That will be usefull. Nice work.

Why can't web browsers render basic markdown natively (no plugins)? Wouldn't that be a much simpler and more secure parser than HTML?

A Markdown Web would look like the early Web.

Markdown isn't part of the HTML spec any more than BBCode, and is therefore not "native" as far as browsers are concerned.

So they could but... why should they?

> Wouldn't that be a much simpler and more secure parser than HTML?

No, because Markdown allows for raw HTML, so a compliant Markdown parser must also include a complete HTML parser.

Thank you!

I've thought before that it would be neat to have the ability to specify an alternative markup inside any element. A kind of:

<div markup="markdown">Browser now _interprets_ content as markdown until end of div.</div>

It would make user-submitted content so much better since you could just directly embed bbcode, markdown, etc. without needing to parse it to html or validate that their submission is valid html/only using approved tags, etc.

I really wish they would. With just a small number of elements added to the Markdown standard (e.g. for plots, videos and audios and, perhaps, a standard way include referenced files in the markdown file itself) this would make the web (and e-mail!) I want.

Not standard, but it shouldn't be very hard to write an extension to handle this use case. I would like to see that happen, actually.

It's definitely not a can't.

Why don't they? Coordination is costly.

static site frameworks like jekyll, hugo and gatsbyjs support this.

gatsbyjs supports md, yaml and external graphql

Is it me or you folks just invented HTML?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact