Html 5.1 and Living Standards

[mattur]

(Split off from other thread to focus on web standards)

... [HTML5] is still officially a Candidate Recommendation, with some elements at risk of being dropped "due to lack of implementation" which apparently means "We've been selling it for ten years but nobody's buying".

The W3C HTML5.0 Spec is a bit irrelevant to implementors and web developers now. It's a dated snapshot, and its target audience is the patent lawyers who work for W3C Members like Google, Microsoft, Intel, Apple, Netflix, IBM etc. Loads of bugs in the spec have been fixed since then, and new functionality has been added, so it's already well out of date.

The W3C HTML Working Group* is currently working on a new snapshot called HTML5.1 [w3.org]. This snapshot will also become out of date as the Living Spec (or in W3C parlance Nightly version) moves on. These numbered snapshots are for patent lawyers, not web developers.

Web developers should always refer to the latest version of the HTML Spec which is maintained here, under an open licence that allows copying and reproduction:

[whatwg.org...]

The W3C version is basically the same, apart from a few bike-sheddy differences, and a more restrictive licence. The W3C Nightly HTML spec and the WHATWG HTML Living Spec are kept broadly in sync so you can use either.

[w3.org...]

Website authors, designers and developers will probably find this a (far) better reference to use:

[caniuse.com...]

CSS3

See also A Word About CSS4 [xanthir.com]

*I'm sort-of a member of the HTMLWG

[lucy24]

Website authors, designers and developers will probably find this a (far) better reference to use:

Matter of fact, that's on the very short list of sites I don't have bookmarked... because I've got the name memorized ;)

But the question remains: If there is no standard, what does a "conforming user-agent" conform to? There's a world of difference between This browser does everything that CSS 2.1 says a browser is supposed to do and This browser does everything that was posted in such-and-such location on such-and-such date, but we make no promises about rules from the day before or the day after.

It would be fine if CSS and HTML only moved in one direction, so today's standard includes yesterday's. You can go a hell of a long way on Euclid and Newton; if a browser can handle sites built today, it should have no trouble with sites built in 1997.

But things are falling off the far end too. One day you wake up and discover that <tt> and <u> are gone. And next week your browser says They're no longer in the standard, so I don't have to recognize them.

[mattur]

If there is no standard, what does a "conforming user-agent" conform to?

There is a standard for HTML, but it's a "Living Standard" [wiki.whatwg.org] so it's updated regularly instead of once every 15 years or so [w3.org]. :)

There's a world of difference between This browser does everything that CSS 2.1 says a browser is supposed to do and This browser does everything that was posted in such-and-such location on such-and-such date, but we make no promises about rules from the day before or the day after.

Implementers use the most up-to-date versions of standards i.e. the specs with the fewest bugs. Using old, superseded specs, would mean baking-in old bugs that have been fixed in newer versions of the specs.

For example, imagine there was some heinous security bug in CSS2.1: no implementer would deliberately release a browser with a security hole just to conform with CSS2.1.

See also: Does that mean the specification can change at any time? [wiki.whatwg.org]

It helps to think of Web "standards" as more of an ongoing, collaborative process for implementers, developers and anoraks to develop patent-free web tech, rather than set-in-stone Standards, like how long a standard unit of length is.

But things are falling off the far end too. One day you wake up and discover that <tt> and <u> are gone. And next week your browser says They're no longer in the standard, so I don't have to recognize them.

The HTML spec defines two things: actual technical stuff that needs to be implemented by browsers to browse the web (e.g. <font> tags, to correctly display old pages), and conformance/validation/ideology stuff for authors (e.g. use CSS instead of obsolete <font> tags). It's quite rare for browsers to actually remove stuff.

So just because a feature becomes deprecated/obsolete doesn't mean that browsers will remove it. e.g. <font> will show as an error in the validator, but is still supported in browsers.

IF everyone stopped using the <font> tag, and an army of volunteers went back and re-wrote the bazillion unmaintained pages currently using it, THEN browsers (and specs) could remove it.

[lucy24]

IF everyone stopped using the <font> tag, and an army of volunteers went back and re-wrote the bazillion unmaintained pages currently using it, THEN browsers (and specs) could remove it.

Ooh, I like that idea :)