Scroll: Extensible Alternative to Markdown | Hacker News Reader

Scroll: Extensible Alternative to Markdown

scroll.pub

50 pointspodiki26 comments

I'd argue that one of the main benefits of MD is the portability, but it is remarkably stupid how different various platforms subtly tweak MD syntax. I swear ADO, Slack, Reddit, etc. all have slightly different rule sets and it takes me two or three tries when swapping between then to get it right.

It is for that one thing that uh... I can't even use it for that purpose.

The need to use keywords for basic stuff like plain text paragraphs loses sight of the entire point of Markdown in the first place.

I generally agree that this isn't actually a markdown alike, and so the tag line doesn't make much sense. In their defense, though, there isn't really a popular ~semantic lwml to refer back to in their elevator pitch--I can understand how they end up framing it like this.

The newspaper-like multicolumn design is quite fun. For some reason, it's very rarely used online.

I'm not sure it's an improvement over the standard 1 column, but it's also refreshing to see people doing these kinds of experiments.

I'd love an easy to use method of putting markdown into this layout

It's hard to get right, sadly. On this page here, I have to scroll down to see the last quarter of a column and then immediately after scroll up again to read the start of the following column.

I love these kinds of layout, but I dont know how to get them to work properly.

I like how it adjust the number of columns to the width of your browser window instead of hard-coding a width (that too many web sites do).

I'd rather scroll up and down for each new column than left and right for each new line.

Happy coincidence to see scroll here. It's in a short list, along with its ~parent format (Tree Notation), of lwml projects that I discovered while trying to census the domain a few weeks ago and loosely intend to tweet or maybe blog about. <3

I've always been annoyed by markdown for one reason: [Link text](link.url) is backwards.

URLs, particularly Wikipedia links, are so much more likely to have parens in them than they are to have square brackets. And users can choose link text much more easily than they can choose a URL. (Link text)[link.url] would have been much much better.

On the topic of Scroll, why is https://scroll.pub/screenshot.png such a tiny screenshot? 300x201 pixels

I always think of it like a function call with the URL as the parameter

URL syntax you described is exactly one used by Asciidoctor ( https://docs.asciidoctor.org/asciidoc/latest/syntax-quick-re... )

I like the look for the web, I think it'd be really cool if there was a "publish to markdown" option. It could even include CSS at the bottom so it would render the same way it does now, but then the docs in github would render and it would probably help to plug it into more things that accept markdown

How do they intend to convince people to switch to this?

Restructured Text didn't go anywhere despite being technically superior. Now there's even more inertia here

I spent some time working with the Sphinx source code (the main user of rst), and found that reStructuredText is very poorly suited for porting to any other platform because a lot of the "spec" is just whatever the Python implementation is doing. IIRC it's also not context-free, so it's more laborious to write a parser. "Technically superior" it is not. I also think a lot of the syntax is obtuse.

That said, I really dislike that Markdown does not have any affordance for adding custom markup. Because of that, it's poorly suited to technical documentation; there's no official way to say "insert the autogenerated docs for function X here." Every doc system that uses Markdown invents its own special thing.

So, I think there's still space for markup languages that find a balance between prettiness and extensibility. I did my own experiment that let you define custom HTML tags but didn't have a reason to really finish the project.

Markdown lets you drop into inline or block-level HTML doesn't it? And if that isn't enough and you want your own markup vocabulary, render-to-HTML pipeline or markup decorator/macro expander, there's always SGML which can be customized to parse Markdown to HTML, or your own extension or custom syntax.

OK, yes, but at that point you're writing your own parser for a custom language that is not Markdown anymore.

Markdown _does not_ support SGML. Look at the source for any CommonMark parser—there's a whitelist for the HTML tags.

Compare to rst and Sphinx. It's very obvious how you're supposed to add new inline and block elements. There is a standard syntax for adding parameters to directives. Markdown has nothing, so there are a dozen incompatible variations that are only available in some parsers.

On the other hand, reStructuredText also defines a whole system for transforming the document tree, which as far as I know, no other markup language does.

> reStructuredText also defines a whole system for transforming the document tree, which as far as I know, no other markup language does

I guess we're not agreeing what a "markup language" is then, since the things you mention, plus defining your formal vocabulary etc., is very much the entire point of SGML/XML/HTML. I'm aware the term "markup" is used inflationary to mean all things text to all people ever since YAML came up with its original, flawed meaning of that acronym, but "markup" in the original sense really is just any layout instruction and proofreading comment written in a shorthand notation on a preprint copy, and has a very specific meaning developed from there in digital text, quite different from "any convention to encode text formatting using ASCII" (I know since my boss, a typesetter by profession, way back used that notation for reviewing/dissing my prints, and ruining my weekend ;)

Markdown indeed is only a set of Wiki syntax conventions for HTML specifically, and as such only allows HTML rather than arbitrary inline/block elements. But it's quite natural to implement Markdown (or rST) as a SHORTREF syntax within the greater SGML framework to give you site building, validation, transformations, text macros, etc. etc.

It does, but you can't use markdown within an html tag. That is, once you go down to html, you can't mix abd match, which would have been incredibly helpful (think html tables)

Was it just me that couldn't find a single example of what this Scrolldown actually looks like? I found a link to it's grammar file buried at the bottom, but that was it.

So, a more verbose (or less terse) syntax for troff?

Do you have a strong defense of this dismissal? It doesn't seem apt to me. It even cites an example of embedding syntaxes like markdown and bbcode within it.

Sorry for offtopicness - could you please email hn@ycombinator.com? I want to send you a repost invite for a different submission.

Noooooooooooooooooooooooo. I'm not even going to bother trying to remember the xkcd number about 14 competing standards.

https://xkcd.com/927/ Always topical

It's not a competing standard, it's another tool in your toolbox. The more tools you have available, the better.

Reply on news.ycombinator.com