IE and HTML5 testing

In the 18 months I’ve really focussed on HTML5, I’ve seen approximately 238 different HTML5 “testing” sites appear. Most of them wildly pick and mix specs, checking for HTML5, related WHATWG-derived specifications such as Web Workers and then, drunk and giddy with buzzwords, throw in SVG, CORS, CSS Media Queries, and some Apple proprietary CSS extension before hyperventilating and going to bed for a lie down.

(Added 4 June 2010: As a case in point, take Apple’s hilariously disingenous “HTML5″ showcases, of which only the video and audio demos have anything to do with HTML5, and which offer “browser upgrade” messages even to other WebKit browsers (screenshot courtesy of Peter Nelson). And don’t get me started on “Standards aren’t add-ons to the web. They are the web” coupled with browser-sniffing and proprietary vendor extensions.)

As an analogy, imagine that HTML5 is the Starship Enterprise to HTML 4′s pogostick. Imagining it? Good.

237 HTML5 testing sites check for

Therefore, it’s particularly refreshing to see the new Microsoft IE9 HTML5 Testing Centre bringing some sanity to the party. None of the scope-creep for our friends in Microsoft. It’s purely HTML5, and only selected bits so we can’t even speculate about thinking about considering accusing them of scope-creep.

So they avoid any mention of fripperies like canvas (who uses that?) or native multimedia (who’s even heard of that?). Why would any web developer care about Web Forms?

Instead, we look only at The text selection APIs, parsing foreign content and getElementsByClassName to determine that IE9 has the best HTML5 support.

To return to our analogy, the Microsoft test to detect Starship Enterprises is:

Yes to all? Wow! It’s the Starship Enterprise!

Actually, it’s HTML5, Jim, but not as we know it.

(Related: What HTML5 is and isn’t, an IE9 speed comparison.)

Caveat: Yeah, I’m having a chuckle. This does not represent the position of my employer. I wrote it outside company time. On a different computer. With a different personality. So chill out, it’s a sunny day.

25 Responses to “ IE and HTML5 testing ”

Comment by Dominykas

They did make a mistake in their methodology, I must say.

They should’ve picked ONE test in each category that only IE9 passes, and then have a table where IE is 100% and everyone else is 0%

Comment by Pete LePage

Hey Bruce,

As your post doesn’t reflect the views of your employer and was written on personal time – so are my comments. Heck, I just started my vacation and am on my way to Disney World…

The tests the IE test team have published so far are just the start. At MIX this year, Dean said they’re going to do for the rest of the platform what they did for CSS 2.1 in IE8. And in the IE8 time frame they donated over 7200 tests to the W3C, effectively increasing the number of CSS 2.1 tests from less than 100 (I think)!!

It’s takes a decent amount of time to create the test plans, implement the tests so there’s plenty more coming. Oh, and they’ve only published the tests for features they’ve implemented and published. There’s still lots more to come!

Just cause the warp core reactor tests aren’t published yet, don’t assume they won’t be! :)

Comment by Bruce

Thanks Pete. I’m not sure the page that houses the tests explains the situation as clearly as you have.

Enjoy your vacation. And remember, when we’re next in the same town, we need to have that Opera/ IE drink-off. You, sir, will have Layout.

Comment by Daniel Walker

I don’t think you’ll get leadership from Microsoft, on this sort of thing, nowadays.

Not for lack of people with opinions, mind – but fears, in senior ranks, of being accused of “steamrolling the decision”, or “enforcing its will through market dominance”, combined with the atrophying layers of ‘process’, that smothers everything everyone now does at Redmond, by all accounts, seems to mean that Microsoft is content to just “shuffle about”, on these sorts of issues, in the hopes that the ripples will cause the market to build up a momentum in some given direction, eventually.

Comment by Ms2ger

I’d say MS has done a pretty good job with these tests… It’s just rather unfortunate that a lot of these tests are just plain wrong. For example, most the SVG parsing tests require removing whitespace-only text nodes from the DOM to pass, while HTML5 requires the opposite. I’ll be (more) convinced the day the IE team releases test cases that test the specification, rather than their own implementation.

Comment by Steven Nash

I’m not entirely sure who Microsoft are trying to kid with such a rigged set of tests.

Are developers who are using HTML 5 now likely to be so easily hoodwinked? I’d wager not.

As for the browser detection on Apple’s HTML 5 demos, that’s a new low! It’s the antithesis of what standards are about.

Comment by Jérôme

I thought MS were pretty clear on video support in IE9 on their blog: “Another Follow-up on HTML5 Video in IE9″ (H.264 and VP8)

I have seen elsewhere that canvas will be supported too, but then again, that was not an official Microsoft source

Still waiting for web forms, though ;o)

Comment by Peter van der Zee

Pete, the point of this rant was not so much Microsoft showcasing a few HTML5 bits or how they test or whatever. The problem we are having is that it states IE9 is 100% HTML5 compliant while competitors are far from it.

If you’re developing IE9 and are in the works of supporting the other sections of the spec, by all means just say so. But don’t go and present this as being it and IE9 being supporting all of HTML5. It just rubs wrong, you know.

Comment by Netgeds

[...] touted HTML5 at Opera, was better-humored if not any happier, calling Apple’s site “hilariously disingenuous.” Adds an Opera colleague Haarvard Moen, “When the page doesn’t work in Opera or [...]

Comment by JJ

I know I’m several months late on this, but I had to say, “CSS 4D Canvas Gradient Transforms” is just about the most amazing thing I’ve ever heard of. I’m gonna start writing up the spec for it. Unfortunately, it will be hard to get around the problem of the transforms being affected by spacetime curvature.