Ambiguous ampersands | Hacker News Reader

Ambiguous ampersands

mathiasbynens.be

30 pointsriledhel12 comments

The reason it's confusing is that the spec is attempting to document the confused logic that browsers implement (almost certainly for backwards compatibility with bugs in older browsers). As an HTML author you shouldn't waste the brainspace on learning these rules, but rather just encode ampersands correctly.

It's not even backwards compatibility though. Incorporating common mistakes into the spec is just plain silly (as in the &amp without the semicolon).

I don't understand why HTML maintains its relaxed attitude. Back in the day, it was understandable since humans wrote HTML pages one by one but nowadays it's all written once and then generated through templates. Thus, we can afford to have a strict standard that just breaks when it sees nonconforming markup. If you don't want your customers to see the "page broken" sign, fix your damn website! Losing money and customers is a very good incentive to conform to a standard.

The reason I support stricter markup is because it makes parsing and rendering way simpler. As it is, browsers have "quirks mode" to guess at what the author really ment when they wrote their markup, and no two browser guesses the same. Then there are specialized "html parsers" like beautifulsoup that deal with the bad markup. Away with all that!

You go ahead and make a browser that breaks like that, and I'll go ahead and spread the word to everyone I know how crap it is, and generally do my best to wipe it from the surface of the earth - on the principle that it thinks the page encoding is more important than delivering value to the user.

I wish the people who downvoted me would tell me what software they work on, so I can avoid it...

The kind of personality dysfunction required to deprioritize the authority and autonomy of the user in favour of machine concerns is enough that I don't want them involved in software I use. Such user-hostile malice is common in open source (due to the intrinsic incentives involved); IMO it's one of the primary reasons Linux fails on the desktop. Everybody is always eager to please the machine, forever creating new "consistent" abstractions, but not so happy to compromise even if such compromise lets the user win.

It's a serious problem. These dysfunctional people are far too numerous.

I think the downvotes may have been less about the point you're making and more about its delivery. While I generally agree with you, the rationale behind making stricter markup is to ultimately push web developers to write cleaner code, which makes parsing easier, and the page load faster - so in the end it does come back to what the user wants. In reality not all web developers are so careful.

The medium is the message; if I didn't deliver it in the way I did, much of the vehemence of what I said would be lost.

C'est la vie, though, I don't really mind.

This is the kind of thinking that makes XML RSS feeds break entirely when someone happens to use an ampersand in their blog post.

That was the idea with XHTML and it didn't really work so great. People aren't very good at creating valid XHTML or tools that create valid XHTML. Pity though, it would have made a lot of things easier.

Strict markup is a prisoner's dilemma. It will not work — it's only a matter of time until everyone defects. Here's the first one:

http://my.opera.com/ODIN/blog/2011/09/28/no-more-xml-parsing...

> If you don't want your customers to see the "page broken" sign, fix your damn website!

Or switch to HTML. User is not the right person to show errors to — they can't even report those errors when browser makes "Contact" link inaccessible!

The spec doesn’t make &amp valid though, it just makes it work the way browsers have always treated it. I don’t see the harm in that. “Support existing content” is one of the HTML design principles.

If you want to write valid markup, that’s great. Heck, we should all do it! But there’s a lot of invalid existing content out there that isn’t going to fix itself anytime soon.

Authors of html minifiers should be interested in and potentially take advantage of these rules --- but that's not very many people, and I agree that most html producers should not try to get clever about this.

Stupid handling of ampersands is exactly why I prevent viewtext.org from displaying articles from https://grepular.com/

Viewtext.org rewrites all links to a redirector on their own site, but doesn't decode the entities first. So if I have this link on my site:

  <a href="http://example.com/?1=2&amp;x=y">link</a>

Clicking that takes you directly to:

  http://example.com/?1=2&x=y

However, clicking the munged viewtext version of the link, sends you to:

  http://example.com/?1=2&amp;x=y

That is a different URL! Can everyone please learn how to encode/decode html entities and uris properly.

Reply on news.ycombinator.com