Hacker News Reader

HTML has only become well-specified recently in its entire history, and it's still riddled with archaic complexities from its time as something with very little rigid specification. I don't see how you can possibly point at HTML and call it the opposite of Markdown in that regard.

And let me tell you, HTML's early non-specificity did NOT cause people to follow the spirit of the law, nor did it make many people consider it usable and understandable by ordinary humans.

You’re totally missing the point. HTML is a document language which is intended to be consumed by machines, whereas Markdown is designed to be good for humans to read directly.

If someone writes a Markdown document and it’s littered with complicated directives and tricky bits of syntax which rely on the specific implementation to work, but aren’t readable as plain text, then they’re not really following the spirit of Markdown.

By contrast, since HTML isn’t intended to be read directly, any hacky bit of markup, however awful it makes reading the raw file, is fair game as long as the browser output is pretty.

"HTML is a document language which is intended to be consumed by machines, whereas Markdown is designed to be good for humans to read directly."

AND intended to be consumed by machines.

For example. John Gruber's markdown perl script transforms his Markdown into HTML.

Although HTML was not meant to be read by the end user, many of the compromises it made from its parent, SGML, do seem to have been intended to make it more easily manipulated (as in both read and written by a somewhat technical human user). No unnamed close tags, extreme flexibility on presence or absence of opening and closing tags, case insensitivity, overlapping nesting, etc. All these things make things much harder for a computer to reliably process the document but theoretically much easier for a human to follow them. Of course, in the long run as these things pile up it actually ends up being counterproductive. It's just hard for both.

"According to Gruber, Markdown is the hardest thing he's ever worked on. He didn't spend a week on it, the idea was not obvious"

It's easy to see how Gruber is affected by the Reality Distortion Field... the idea is entirely obvious, has gigabytes of prior art (albeit without, dare I say it, standardization).

Witness decades of email, Usenet, BBS postings that used _, * etc for emphasis, let alone the claim that lists in that format are non obvious... well, frankly laughable.

"The lack of a standard helps keep people honest by making them follow the spirit of the law, not the letter, as you see happening with things like HTML and CSS."

I recognise that line of reasoning from the RSS 2.0 days. And how RSS feeds were okay for years until Dave Winer decided one day that they were funky. Having a deterministic standard is a quality metric.

On Gruber's podcast: he notes that writers should check whether something is valid Markdown by rendering it and seeing the output ( https://overcast.fm/podcasts/episode/344902019595#t=4527 ). paraphrasing his logic: If the output is wrong, that means the input needs fixing. That is an assertion that the code implementation is the specification. And we've been down that road before ( bugwards compatible implementations of Internet Explorer, for example )

Ambiguity is not a good basis for a data structure specification. An undefined behaviour means that all possible behaviours are correct.

It's an uninformed defense of a fundamentally incorrect position; no wonder most people here disagree with it. People here know that a language is syntax and semantics, he seems to be arguing that semantics aren't important, or that ambiguity in semantics is good? It only makes sense if your Markdown never has to be processed in any way (e.g. converted to HTML). Which ok, maybe there are some use cases for that, but they are dominated by the ones where Markdown is transformed into another representation. That requires semantics.

> This gives it the rare characteristic of being intuitive to understand and use, but not necessarily easy (or even unambiguously possible) to implement

Ambiguity does not make anything easy to understand or use. Not possible to implement is not a feature of any language.

> Markdown is the hardest thing he's ever worked on

If that's even remotely true, I hope he never touches any piece of code in his life again.

For god's sake, just look at his implementation: "Hacked together after a night of binge drinking" would be the perfect description of it.

Puff your chest out all you want, call me back when you've made a bigger contribution and no longer have to hide behind 'froaway001'.

I'm quite comfortable with the contributions I have made, thanks.

The hard part is figuring out what the format should look like (a labor of love that he spent months if not years on, iteratively refining), not the implementation (which is a hacky ball of perl).

That format is essentially (sans the link syntax) what plaintext email has been looking like for decades. I don't get the whole praise for Gruber of "inventing" something novel. It's not.