The thing I hate about HTML (2010) | Hacker News Reader

The thing I hate about HTML (2010)

blog.jgc.org

32 pointsjgrahamc63 comments

Saying we should write HTML using the following syntax?:

(div class="red"

  (p

    (Hello, World!)

  )

)

I'm not sure what problem this solves. It eliminates the tiny bit of work that is typing the extra characters in a closing tag, and in doing so you lose all of the other information as to what exactly you're trying to close. The end of most HTML douments would be ))))))) (plus a mess of indenting).

I can see this causing many more problems than it would solve; I'm okay with typing three extra characters to avoid lack of confusion.

The current syntax may not be ideal, but I don't think removing the object type from a closing tab is a good way to make it better.

Hah, but aren't most HTML documents a mess of </div></div></div></div> at some point :)

Not when you know what you're doing and don't use auto code generators.

What is "knowing what you're doing?" I'm honestly curious, because I handwrote most of the HTML on my website and I don't know how to have it cleaner than something like this:

  </div> <!-- footer -->
  </footer>
  </div> <!-- row -->
  </div> <!-- container -->

(If you actually look at the website, it will be pretty obvious that I'm no skilled web dev)

Indenting all of the tags would help, but that's probably about it.

    <div>
       <footer>
       </footer>
    </div>

In theory, maybe.

In practice, where the code is coming from 3 or 4 different levels of templating? Bets are off.

I use Twig for templating a lot, and I have to admit that while I indent per template, i don't even bother keeping track between parent and child.

But then, Twig has its own weird issues sometimes with adding whitespace so I've given up caring how templated html works altogether in that case.

Assuming a mythical "knows what they are doing" user, then parens aren't any different.

Are you really on hackernews bragging about HTML skills?

Bragging? I don't think s/he is bragging, but just saying that html can be done wrong.

Editing structures like this using paredit is quite easy and fun though - once you're over the hump.

The only use for this I can think of would be to make it take less characters.

which is good, because I keep running out. :)

It solves a problem which vexes many Lisp programmers, which is the existence of data structures and languages which are not lispy.

I thought I remember reading this was debated extensively when XML was being ratified. The consensus was that it was better for reading the markup even though it made it harder to write vs SGML's empty end tag.

In any non trivial HTML doc many of the close tags will be far from the start tag, such as <body></body> and looking at a bunch of </> tags and having to go match them up with the correct start tags is more difficult. I run into this all the time with parentheses in your typical programming languages.

The IDE I usually use does auto close tags but also does folding both help with verbose close tags or matching parens so it doesn't matter much what the language does anymore. I generally do think html/xml is more readable because it so verbose including end tags.

The reason this is a problem is not because documents are long, but because HTML does not have a concept of encapsulation. You can't make a subroutine--or rather, subtree--of HTML.

Of all the things to complain about HTML and modern web development, this ranks pretty low on my list.

The fact that the end tags are named means that people are also tempted to write markup like:

    <strong> one <em> two </strong> three </em>

This is technically invalid, but some browsers will "helpfully" parse it and apply strength to "one", strength and emphasis to "two" and emphasis to "three". This would be avoided if the closing tag weren't named.

Invalid markup means you are relying on the browser to read your intent or your mind. They are very good at doing that but no self-respecting web developer would ever do such an insane thing as your example.

EDIT: Interesting about the downvotes. It shows a total lack of understanding the fundamentals of how browsers work but, alas, downvote away.

Don't feel bad. HN has turned into a political, anti-government step up from Reddit. But a step up from Reddit is no compliment.

You're being downvoted because of your argument from intimidation ("no self-respecting web developer...") and your wording ("such an insane thing"), not due to a lack of understanding browsers.

So don't tell the truth? It's interesting that people on Hacker News are intimidated so easily when confronted with such things.

No, tell the truth - just don't be a fucking asshole about it. Everything about the way you're talking makes you sound like an aggressive jerk. Even your username screams "I'm an asshole who thinks he knows it all and always has a snarky reply ready". It's OK. We've all been there. Exercise is a good way to get the negativity out.

"I can't think of one" as a username screams "asshole who thinks he knows it all"? Wow. You should hear my thoughts now.

Jade does this elegantly.

http://jade-lang.com

It's just so refreshing not having to type closing tags(Emmet aside).

Hiccup gives you HTML in Lisp https://github.com/weavejester/hiccup

how about just use whitespace, like python:

  div class="red"
    p
      Hello, World!

(of course I know would not look nice if had to place <b>tags</b> inside of single line of text, and might get a little too indenty)

This is basically jade[0]. Although jade isn't quite as smart, so the expression you wrote would be considered invalid.

[0] jade-lang.com

interesting. Very clean looking.

You absolutely have to split jade documents and use includes/blocks though, any decent sized page, even if it's static becomes unmanageable very quickly.

Because whitespace matters in HTML.

good point, but could you elaborate?

One quick example.

Certain elements are, or can be made to be, "inline" which gets them treated the same as characters on a line. Thus those elements, such as an image, will have white space between them as a word does in a paragraph. If you butt those element's tags right next to each other, with no white space between them, then there will be no white space on the displayed page.

This is good for hand-written html but annoying for machine-written html.

machine would just have to count '/t' tabs at beginning of each line to parse. Is that any slower than scanning every character for '<'?

http://en.m.wikipedia.org/wiki/Rich_Text_Format from 1987 ish worked this way. Or as another commenter mentioned SGML in it's more compact forms from the same time period and so therefore early HTMLs. Having a closing symbol at all is a pain and I have painful memories from trying to manually find a missing closing bracket in a book length Rtf document. HAML uses significant indention to accomplish the nesting which I find much much easier to read nowadays.

Browsers are too kind to bad developers (who accidentally omit close tags) but I don't see that changing those to parenthesis or whitespace will make those developers any less rubbish.

The simple fact is: if you're still regularly typing your close tags in 2015, you either need a better editor or you need to learn how to use your current one better.

SGML has null end tags, which do exactly what he wants. For example this is valid:

   <p /Hello World/

10 years ago I was a heavy TeX user and really hate HTML. I made it a little fun with my stupid flex program https://github.com/icy/ktml :D

This is such an odd rant from a very smart guy. Closing tags make plenty of sense and from our web experience definitively are not a problem. The proposed improvement seems far worse.

> PS Yes, I know there are some tags that don't need end tags (e.g. <img>)

e.g. <p>, <li>, <tr>, <td>.

> But that's just messed up.

Nah, it's pretty straightforward once you wrap your head around it: https://html.spec.whatwg.org/multipage/syntax.html#optional-...

> </>

This reminds me of @annevk's XML5 concept: https://annevankesteren.nl/2007/10/xml5

"A p element's end tag may be omitted if the p element is immediately followed by an address, article, aside, blockquote, details, div, dl, fieldset, figcaption, figure, footer, form, h1, h2, h3, h4, h5, h6, header, hgroup, hr, main, menu, nav, ol, p, pre, section, table, or ul, element, or if there is no more content in the parent element and the parent element is an HTML element that is not an a, audio, del, ins, map, noscript, or video element."

Quite straightforward indeed.

> "A p element's end tag may be omitted...

practically always except when it obviously makes no sense. If you mean <p>Hello</p><span>World</span>, you obviously can't omit </p>, since the inline <span> would become part of <p>. But why would you write that anyway?

When you omit </p> where you shouldn't, the validator will usually tell you. You do validate your HTML, right?

Deleted comment

You are confused. HTML is used to describe the structure of the document, not its visuals. CSS has nothing to do with the HTML content so your complaint about the property/value thing is misplaced and inlining CSS, as you show, is frowned upon.

Then you show the tag/tag thing as if that is a HTML pattern but what you show as your example is just gobbldey-gook nonsense and you claim it's typical.

I just don't know what to say to you.

See: https://github.com/dominictarr/hyperscript

After years of working with C++ GUI (Borland VCL's version) I'm loving HTML5 and how the controls layout so nicely.

That is the rather depressing thing about this. The sorry state of HTML5 is an indictment of the much more sorry state of every other cross-platform GUI toolkit. When you start to break down all of the issues of development, training, deployment, and execution, HTML5 is the better tool. It's by no means a great tool, but what else do we have available?

i was actually thinking about this recently. one appeal of json over xml is xml is so wasteful with all those tags. then i thought, why in the world are we still doing web pages in html and not some other json or other syntax? =|

Your <b>opinion <i>is</b> incorrect</i>

Everything should be S-expressions, not just HTML.

Not a fan of how this blog takes over the swipe from screen edge gesture so I can't go back in Safari easily.

(2010)

This is such a trivial complaint and breaks readability/usability for no real advantages.

Such a bummer there's nothing out there that solves this. Oh wait, slim, haml, PHamlP, shpaml...

Two easy solutions:

1. HAML

2. IDE

l2haml

The thing I hate about bikeshedding

Complaints such as his, and most of the comments so far below, show a total lack of knowledge of the science and underpinnings behind the whole thing and the only supported proof throughout is silly, invalid examples of markup no one would ever use.

More laughable are the attempts at making HTML look like a programming language, which it is not, and I'd bet that comes as news to many.

Is HTML5+CSS3 not fully turing complete though?

Quit reading stupid articles written by out-of-work people with nothing better to do.

Reply on news.ycombinator.com