You’ve heard it’s coming in 2012. Or maybe 2022. It’s certainly not ready yet, but some parts are already in browsers now so for the standards-savvy developers, the future is worth investigating today. Ian “Hixie” Hickson, editor of the HTML 5 specification, hopes that the spec will go to Last Call Working Draft in October this year.
Accessibility Task Force member, Bruce Lawson, interviews Hixie on how the specification for the next generation of the Web’s markup language is shaping up. Disclosure of affiliations: both work for browser vendors—Bruce for Opera, Hixie for Google (and previously, Opera and Netscape).
The spec now known as HTML 5 began with a "guerilla" group called WHATWG. How and why did the WHATWG begin?
The short answer is the W3C told us to.
The long answer: Back in 2003, when XForms was going through its final
stages (the "Proposed Recommendation" vote stage), the browser vendors
were concerned that it wouldn’t take off on the Web without being made a
part of HTML, and out of that big discussion (which unfortunately is
mostly hidden behind the W3C‘s confidentiality walls) came a proof of
concept showing that it was possible to take some of XForms’ ideas and put
then into HTML 4. We originally called it "XForms Basic", and later renamed
it "WebForms 2.0". This formed the basis of what is now HTML 5.
In 2004, the W3C had a workshop, the "The W3C Workshop on Web Applications
and Compound Documents", where we (the browser vendors) argued that it was
imperative that HTML be extended in a backwards-compatible way. It was a
turning point in the W3C‘s history—you could tell because at one point
RedHat, Sun, and Microsoft, arch-rivals all, actually agreed on something,
and that never happens.
The outcome of that workshop was that the W3C concluded that HTML was
still dead, as had been decided in a workshop in 1998, and that if we
wanted to do something like HTML 5, we should go elsewhere. So we announced
a mailing list, and did it there.
At the time I was working for Opera Software, but "we" in this case was
Opera and Mozilla acting together (with Apple cheering us from the
How did you become editor?
I was at the right place at the right time and everyone else was too busy.
How do you personally go about editing the spec and incorporating
feedback? What are your processes?
This has varied over the years, as we’ve gone from a nascent organisation
with a few dozen people to a well-established project with a mailing list
with 900+ subscribers. Mostly it’s all down to managing e-mail. When
someone writes feedback on the spec, whether by sending an e-mail to one
of the mailing lists I’m on, or by blogging somewhere, or twittering, I
log their feedback in a folder on my IMAP server. Feedback gets
categorised into either feedback I can work on right away, or feedback
that I can’t deal with yet for whatever reason. An example of the latter
would be requests relating to mutation events, because I’m waiting for
DOM3 Events to update how mutation events work.
Then, I just go through all the feedback I have, e-mail by e-mail, more or
less in the order that I received them, sending replies and fixing the
spec to address the issues that were raised.
This has some disadvantages, for example there’s a big delay in between
when someone spots an error and when I fix it. It also has some really
important advantages. If I respond to feedback on something I wrote
straight after writing it, I sometimes find that I have an attachment to
that section, so if someone suggests a total replacement, I tend to not
like their idea. But if I have a delay, I find my attachment has gone
away, and I’m eager to replace my old stupid idea with their better one.
(Assuming it’s better, anyway!)
What’s the hardest thing to do?
There are a few things that are hard. One is saying "no" to people who have
clearly spent the time to come up with a good idea. The sad truth is that
I reject almost everything that I and anyone else thinks of, because if I
didn’t, the spec would be a thousand times more bloated than it is now. We
get proposals for all kinds of things, and we have to have a very high bar
for what goes in. There’s also the danger that if we add too many things
to the spec too quickly, the browser vendors will each implement their own
bit and it’ll be a big mess that won’t help Web authors.
So I have to make judgements about what is worth adding and what isn’t,
and that’s hard. I’ve upset a lot of people by rejecting their ideas,
because they take it personally. On the other hand, some of the most
productive members of the community now are people who’ve had many of
their ideas rejected, but they stuck around long enough to see a few of
their ideas make it in. The best way to get an idea into the spec is to
find something in the spec that’s just clearly wrong, which is something
that a lot of the most active people do a lot, too!
Something else that’s hard is making up new features. The bulk of HTML 5 is
actually just defining how browsers already do things, which, although
complicated and sometimes unbelievably arcane, is, at the end of the day,
pretty easy to spec: you test the browsers, and you write what they do.
Rinse, repeat, until the spec covers every possible case.
Making up new features, though, means actually thinking about what should
happen, what is the most understandable solution, figuring out how things
should fit together, and so on. It’s often tempting to make something that
is theoretically neat, but which doesn’t fit in with the rest of the
language, too. After all, that’s where all this came from—we don’t want
to create a new XForms, a really well-designed technology that doesn’t fit
into the way people write pages.
What’s in the spec?
You’ve said that HTML 5 is in "direct competition with other technologies intended for applications deployed over the Web, in particular Flash and Silverlight". Why is it so important to do so, and isn’t it a lost cause given that those techologies are already out there while HTML 5 is not yet complete?
HTML 4 is also in direct competition with proprietary technologies, and
it’s winning, hands-down. HTML5 is just continuing the battle, because if
we don’t keep up, then the proprietary technologies will gain ground.
What are the main philosophies of HTML 5?
Backwards-compatibility, incremental baby steps, defining error handling.
Those are the main philosophies.
What else did WHATWG try to achieve with this new iteration of HTML?
We started from trying to put features from XForms into HTML 4, and we
quickly also took the opportunity to fix some of the things in HTML 4 that
were either too vague or disagreed with reality (that is, where the
browsers all did one thing but the spec said another). It turns out that HTML 4 is so vague that this is a pretty big task—it even involved
defining the whole HTML parsing model, including error handling, which is
a huge job (it took me the better part of a month to write the first
draft, and we were tweaking it for about a year before it become more or
Something else we’ve tried to do is make things simpler. We’ve simplified
the syntax (e.g. the rules about what can be quoted, what strings are
ids, etc, are much simpler now). We’ve made things which people
autofocus=""to focus a form field when the page loads, instead of using
control.focus(), which allows the browser to do clever things like not
actually focus the control if the user is already typing elsewhere.
Does HTML 5 legitimise tag soup? Does "paving the cowpaths" perpetuate
No, HTML 5 actually makes the rules for markup even stricter than HTML 4 in
many ways, both for authors (the rules are simpler, but stricter, than HTML 4′s) and for implementers (gone are the days where they can just do
whatever they want when handling parse errors, now every browser has to
act the same).
Hopefully, we’ve managed to make the rules on what is valid syntax more
understandable, which should help with getting more good markup. We’ve
also made it possible to write clearer validators, so I have high hopes.
message about separating behaviour and structure?
I didn’t know about a message about separating behaviour and structure, I
must have missed that memo! HTML 5 takes a pretty hard line on separating
style and presentation from structure and semantics; there are no more
fonttags. Separating the logic and behaviour from the structure and
semantics of an HTML document isn’t as important, generally, as far as I
The main advantage of defining the HTML DOM APIs and the HTML elements in
the same specification is that we don’t let stuff fall through the cracks.
In practice, browsers implement the HTML elements as DOM nodes, there’s no
difference. When we separate the two in the specs, therefore, we introduce
a conceptual gap where there isn’t one in reality. The DOM2 HTML spec, for
instance, doesn’t say what happens when you change the
checkboxon the fly, and the HTML 4
spec doesn’t mention that changing attributes on the fly is possible, so
in the HTML 4 / DOM2 HTML era, there’s a big hole there. In HTML 5, this is
all defined together, so we can tighten this up and make sure there are no
Why no native support for microformats/ RDFa in HTML 5?
Microformats is natively supported in HTML5, just like it was in HTML 4,
because Microformats use the built-in extension mechanisms of HTML.
We considered RDFa long and hard (in fact this is an issue that’s a hot
topic right now), but at the end of the day, while some people really like
it, I don’t think it strikes the right balance between power and ease of
authoring. For example, it uses namespaces and prefixes, which by and
large confuse authors to no end. Just recently though I proposed something
of a compromise which takes some of RDFa‘s better ideas and puts them into HTML 5, so hopefully that will take care of the main needs that caused
people to invent RDFa. We’ll see.
Do the browser makers have too much influence on the spec?
The reality is that the browser vendors have the ultimate veto on
everything in the spec, since if they don’t implement it, the spec is
nothing but a work of fiction. So they have a lot of influence—I don’t
want to be writing fiction, I want to be writing a spec that documents the
actual behaviour of browsers.
Whether that’s too much, I don’t know. Does gravity have too much
influence on objects on earth? It’s just the way it is.
One of the chairs of the W3C working group is a Microsoft employee. Is that giving too much power to one browser vendor, or a good thing,
given that Microsoft’s browsers still dominate and their buy-in on any spec is
Personally I would like Microsoft to get more involved with HTML 5. They’ve
sent very little feedback over the years, far less than the other browser
vendors. Even when asking them about their opinion on features they are
implementing I rarely get any feedback. It’s very sad. If I e-mail them a
question about how I can best help them, I usually get no reply; at best
I’ll get a promise that they’ll get back to me, but that’s it.
There has been a lot of spirited debate (ahem) about accessibility in
the development of HTML 5. How does the spec deal with the requirements
of people with disabilities?
Universal access—the requirement that anyone be able to use information
on the Web—is a fundamental cornerstone of HTML‘s design, just like
security, privacy, and so on. In general, we try to design features so
that they Just Work for everyone, regardless of how you are accessing the
Web. For example, in HTML 5 we’ve added new input controls like calendars.
These will Just Work with screen readers once browsers support them,
authors don’t have to do anything special.
Does your personal support of humanitarian eugenics affect your opinion of giving
extra "help" for people with disabilities?
You’ve been reading too much of our pet troll’s blog! ;-)
[Bruce's note: this refers to Mr Last Week, mysterious author of the blog Last Week in HTML 5, which lampoons the HTML 5 Working Group in very funny, frequently foul-mouthed manner.]
People with disabilities are just as important to me in my work on HTML 5 as is anyone else.
You wrote to ask screenreader vendors to participate in
the specification process. Did they ever reply?
A couple did, but only to say they had little time for the standards
process, which was quite disappointing. Since then, though, Apple has
ramped up their efforts on their built-in Mac OS X screen reader software,
and we do get a lot of feedback from Apple. So at least one screen
reader vendor is actively involved.
When there’s a built-in way to do something, using that is the simplest
and most reliable solution. So for example, if you want to have a
checkbox, using the
inputelement with its
typeattribute set to
checkboxis the simplest solution—it’ll work for everyone, with or
useful when HTML doesn’t let you do what you want and you find yourself
hacking around with many nested
divs, scripting your own controls and so
Can we expect ARIA-specific constructs which have no equivalent in HTML 5, such as live regions, to be allowed under the rules of HTML 5 so it will all validate?
Yes, the plan is to make sure ARIA and HTML5 work well together. Right now
I’m waiting for ARIA to be complete (there are a number of last call
comments that they haven’t yet replied to), and for the ARIA
implementation rules to be clearer (it’s not yet obvious as I understand
it what should happen when ARIA says a checkbox is a radio button, for
instance). Once that is cleared up, I expect HTML 5 will give a list of
conformance criteria saying where ARIA attributes can be used and saying
how they should be implemented in browsers.
Why, when, how, who?
Why would we content authors want to move to HTML 5? What’s in it
Today is probably too early to start using HTML 5.
Long term, content authors will find a variety of new features in HTML 5.
We have a bunch of new structural elements like
footer, and so on. We have new elements for embedded media, like
audio. We have new input controls, like the calendars I mentioned,
but also fields for URLs, e-mail addresses, telephone numbers, and for
color selection. We have control over autocomplete values in text fields,
as well as field validation so that you can say which fields are required.
We have context menus,
pushState()so you can update the URL in Ajax
applications, and offline application cache manifests so that your users
can take your applications offline. The list goes on.
There’s also the benefits that come from using an HTML 5 validator. HTML 5
is much more precise about many things than HTML 4, so the validators will
be more useful in catching real errors. The
embedelement is no longer
Are there advantages for end-users, too?
A more powerful HTML means more powerful Web applications. Just like
XMLHttpRequestresulted in more interactive apps, HTML 5 will result in
a richer and more consistently reliable experience. I hope!
What’s the the timeline? When can we start using HTML 5?
The plan is to have the spec mostly finished by October 2009. A lot
depends on the browser vendors, though. I don’t know when things will be
implemented widely enough that authors can use them reliably everywhere.
Some features, like
video, are getting implemented in most
browsers as we speak. Others will take longer.
What can standards-savvy WaSP readers do to get involved with the specification process?
There are a number of ways of taking part. What we need most of all these
days is technical review of the specification text, calling out places
where I screwed up, where the spec defines something that’s not easy to
use for Web authors, where the spec contradicts itself, typos, spelling
mistakes, grammar errors, errors in examples, you name it.
I posted a blog entry recently detailing how people can send feedback. You can join the W3C HTML Working Group or the WHATWG. There are also lots of other things people can do—write demos, write tutorials, edit other related specs, write articles introducing parts of
the spec on the blog, write test cases… Anyone who wants to help out but doesn’t know where to start should drop
me an e-mail at firstname.lastname@example.org.
Will there ever be an HTML 6, or is it a convenient fiction to park
I’m sure there will be an HTML 6, and 7, and 8, and probably many more,
until someone comes up with something so radically better that we stop
evolving the Web as we know it.
I expect work on HTML 6 will start even before HTML 5 is completely done, in
fact. Putting the finishing touches on HTML 5 will be a long and tedious
job involving writing a massive test suite. HTML 4 never had a serious test
suite created (it was too vague as a specification to really be properly
tested), so we have to start from scratch with HTML 5. The HTML 6 team will
at least be able to build on what we’ve done with HTML 5, I’m jealous!
Actually if it was up to me, after HTML 5 I would probably transition HTML to an incremental model. Once we have a basic spec that is well-defined
and has been proven, instead of releasing a frozen snapshot every few
years, I’d prefer a model where we can slowly evolve the language, call it
"HTML Current" or something, without having to worry about versioning it.
To some extent that’s what we’re doing with HTML 5, but I think formalising
it would really help.
Having versions of specs doesn’t make sense when you have multiple
implementations that are all evolving as well. No browser is ever going to
be exactly HTML 5, they’ll all be subsets or supersets. So why bother with
versioning the spec?
It’s a very unusual idea in the standards world, so I don’t expect us to
do this. But I do think it’d be the best way forward.
Would you like to be the HTML 6 editor?
Too early to tell! It’s been a lot of fun working on HTML 5, it’s quite
challenging and you have to deal with all kinds of issues from the deeply
technical to the highly political. I might want a change of pace when
we’re done with HTML 5, though.
What’s your fave feature that didn’t get into HTML 5 that you’d put
into HTML 6?
In-window modal dialogs or dialog box—the kind of prompt you get when the
computer asks you a question and won’t let you do anything else until you
answer the question. For instance, the window that comes up when you say
"Save As…" is usually a modal dialog.
Right now people fake it with
complicated styles and script. It would be neat to just be able to say
"make this section a modal dialog". Like
showModalDialog(), but within
the page instead of opening a new window with a new page.
I’d add it to HTML 5, but there are so many new features already that we
need to wait for the browsers to catch up.
Finally, is it true that you and Mr Last Week are the same person, like Edward
Norton and Brad Pitt in "Fight Club"?
Oh, no. Our pet troll is a phenomenon all to himself.
Thanks for your time.
- #1 On May 13th, 2009 8:42 am