Hacker News Reader

Comment by mdpm | Hacker News Reader

HTML was for providing declarative markup. CSS was for styling that on a user-agent. JS was for enhancing the capabilities of the user-agent in the context of the markup.

We then tried to make what was basically gopher++ into an app framework. Then we wanted all user-agents to behave the same, wanted pixel perfection. This meant we didn't empower user-agents to make their own presentation decisions.

Glad you refer to SGML, I thought XHTML was amazing at the time too, then HTML5 broke compatibility with everything. JSON has been inferior in every way, but XML and attendant standards had also blossomed into fractal complexity. Ever try XSL-FO?

Everyone can roll a wheel, and imagine a vehicle, but not everyone can build a car, at least not without engaging with some complexity. We need better tools for the middle ground, but need to avoid building new layers of abstraction that pretend the complexity isn't there.

View on HN

> Glad you refer to SGML, I thought XHTML was amazing at the time too, then HTML5 broke compatibility with everything.

HTML5 was explicitly designed for backward-compat with the large body of HTML4 (and earlier) content out there. And HTML5 can still be parsed using SGML, which as a superset of XML is every bit as powerful as XML, yet also has tag inference, attribute shortforms, and other minimization features dropped from the XML subset of SGML that are necessary for HTML, though. So there's no reason to be bitter about XML's demise on the web; I've even written a tutorial for XML workflows involving HTML integration or output using SGML [1].

[1]: http://sgmljs.net/docs/sgml-html-tutorial.html

> XML and attendant standards had also blossomed into fractal complexity.

Fully agree. It appears that once XML was out there, people wanted to apply it to each and everything. When markup is really only useful for semistructured text. In that department, you actually might find SGML much more useful, since not only can it deal with HTML, it can also parse markdown or other custom syntax (even JSON and s-exprs!), has built-in styling and transformation facilities capturing core CSS, and much more powerful templating and transclusion/fragment reuse mechanisms for actual text authoring and processing.

Interesting. Makes me wonder if even YAML could be brought to heel ...

There was a moment in the web's history where we were trying to figure out how to make XML, HTML, RSS, and 'microformats' all play well together. Next to that were a million authors of WordPress themes who thought 'semantics' was academic rubbish, and just wanted to make the web into another word processor. We got OWL, RDF triples, and no-one paid attention, and just carried on farming complexity.

The cries for simplification are more the result of the tooling, build chains, etc., in my mind than the core tech of HTML, CSS, JS. I'm just concerned that we redouble the complexity by inventing new abstractions to wrap the old.

> HTML5 broke compatibility with everything How? It's backward compatible, it's the first spec that tells the browsers how they should handle errors and introduced new semantics. Semantics that gracefully degrade in non html5 browsers, I find this thing simply wonderful. It's for this same reason sites like https://spacejam.com/ still renders fine like they did back then.

HTML5 is no longer XHTML compliant. There were a suite of tools that worked with markup that weren't browsers. This allowed the web to be the input, output, and middle of other chains of tools. It was sort of like removing '|' from the CLI.

I still don't get it, HTML5 lets you write your documents in XHTMLish way. The spec doesn't care, you just decide. And tools should still work as expected

And BWT it was xhtml to broke backward compatibility "XHTML is not compatible with the IE8 and other older versions of the browsers. HTML5 is capable of using XHTML style tags but not vice versa. While writing code in XHTML, there are several restrictions that the developers need to adhere to"

HTML5 documents are not XML; this means that tooling that isn't a browser, or isn't 'HTML aware' became excluded. The web 'special cased' itself at this point, and HTML and common data processing pipelines of the time parted ways, this is my only point.

You can't compare it to the pipe operator, you can still write xhtml valid documents in html5

But most of HTML5 documents aren't written like that, which is the problem.

Reply on news.ycombinator.com