How do you judge HTML5 compliance when there's no such thing as HTML5?

While the world waits for news of HTML5's supremacy, the current state of the art leaves something to be desired -- but it's getting better

Early today Ted Samson reported that Microsoft has definitely, positively, almost, sort of given up on Silverlight, in some situations, maybe. We know for sure that Microsoft will continue to push Silverlight on Windows Phone, and it'll remain the language of Microsoft's choice for locked-down Internet Explorer intranets. But for the wild Web at large? Microsoft's Bob Muglia waffled mightily.

Over the weekend, the wailing of Silver sycophants ripped across the Internet, with Steve Ballmer drawing brickbats for abandoning Microsoft's Last Hope for Internet Domination. How dare he sell out to HTML5?

If I were a Silverlight developer, I wouldn't be too happy about it, either. But Microsoft hasn't cut Silverlight loose -- at least, not yet -- and HTML5 is in no danger of becoming an overnight standard, much less a long-term solution.

HTML5 means different things to different people. Remember the story of the blind men and the elephant? If you ask 10 different industry observers to describe HTML5, you'll get 10 different descriptions -- or 11, maybe.

I decided to see just how far the latest browser betas have progressed toward finding a common HTML5 ground. Part of the problem in judging HTML5 compliance or compatibility is that, to put it bluntly, there's no such thing as HTML5, at least not in a concrete sense. It currently exists primarily as a Public Working Draft. In an interview with Tech Republic two years ago, Ian Hickson "Mr. HTML5," predicted that the standard would reach the Proposed Recommendation stage in 2022 -- yes, 12 years from now.

See what I mean about an elephant?

Playing the part of the blind men, I pushed the three main browsers, currently in beta, in the direction of the HTMeLephant to see how they would react. The results surprised me. Specifically, I used Firefox 4 Beta 6, Google's Chrome 7.0.517.41 beta, and Microsoft's Internet Explorer 9.0.7930.16406 beta (without Platform Preview 6).

First stop: the trunk. Niels Leenheer, from Groningen, the Netherlands, has spent years working on HTML5. He describes himself as a Frontend engineer and graphic designer. His HTML5 testing website is widely held to be the most thorough compatibility cruncher, although (expectedly) many browser companies take umbrage with his approach. Niels' site rates browsers based on their ability to handle capabilities described in the current HTML5 Working Draft, with a possible maximum score of 300, and he assigns bonus points for useful features that aren't currently in the Draft.

For compatibility, Chrome 7 scores 231 + 12. Firefox 4 scores 204 + 9. And Internet Explorer 9 comes in at 96 + 5.

Perhaps Microsoft isn't quite as committed to HTML5 as the headlines would have you think. Or maybe, charitably, it's concentrating on a different part of the elephant. According to W3C compatibility tests, for example, IE9 implements more of the Canvas spec than either Google Chrome 7 or Firefox 4; see Peter Wayner's preview of IE9 Platform Preview 6.

Next: the eyes. Google has a site called HTML5 Rocks devoted to HTML5 -- eye candy, for the most part. I took the three browsers to that site and tried a few of the goodies. The Samples Studio has a demo slider carousel that flips and slides through pictures. Chrome handled it well (you expected any different?); Firefox wouldn't slide but the pictures did appear; IE couldn't even show the pictures. A different application in the Samples Studio lets you drag around a video cube with a picture of a humming Chrome eye. On my test machine, Chrome was sluggish but managed to slog through the test; Firefox sprouted scroll bars in weird places, with a frozen picture that didn't drag; and IE 9 didn't fare any better.

Continuing with the eyes, I moved over to the Internet Explorer Test Drive site and tried the simple rendering of a static picture in HTML5, called A New Day. Internet Explorer displayed the picture beautifully (as you would expect), as did Firefox. But Chrome distorted the picture badly. Worth noting: Even Firefox 3.6 displays it well.

Next I tried the hands. (What, you didn't know that elephants have hands?) There's a very simple wireframe pinball game on the IE Test Drive site, drawn entirely in fairly straightforward HTML5, called Canvas Pinball. All three of the browsers handled it with aplomb.

Finally, the tail. I played pool with all three browsers on Pixel Lab's Agent008ball site. Again, all three worked splendidly.

So what did the blind men think? In compliance tests, there's no question that the current IE beta trails the pack. In specifically constructed (some would say "contrived") demos, each of the manufacturers can make their browser look good and throw a curve at the others. But for general HTML5 rendering, they're all pointing in the same general direction.

For those of us who have had their fill of Flash, that's good news indeed.

This article, "How do you judge HTML5 compliance when there's no such thing as HTML5?," was originally published at Get the first word on what the important tech news really means with the InfoWorld Tech Watch blog.

Copyright © 2010 IDG Communications, Inc.