Internet Explorer, WebGL and a Return to the Bad Old Days
Most of the response from Microsoft’s competitors regarding “native HTML” is mockery. Some of that is deserved – it’s clearly a marketing buzzword. But where the substantive responses to Microsoft’s performance claims? Demos aren’t evidence of better performance, and the overall value of a browser is about more than a bunch of fish in a virtual aquarium. But how does the performance of Microsoft’s browser hold up? And more importantly: what does Microsoft’s approach mean for open standards and the future of the Web?
What exactly is “native HTML5″? InfoQ has a post looking into it a bit, and concludes that it’s about more than just performance: it’s about deep integration into the OS. I’m not too worried about that at this point though. What about Microsoft’s claims – is its implementation in HMTL5 of in IE9 and 10 legitimately faster than Chrome or Firefox? My informal testing of IE9 and the IE10 Platform Preview are inconclusive. OF course, IE smokes its competitors with its own demos. But there’s not really enough HTML5 in the wild to make a real-world comparison.
So what about standards? If IE can do things that other browsers can’t, what does that mean for developers? If IE’s HTML5 is really better, can other browser makers catch-up?
Here’s something I haven’t seen discussed in relation to the native HTML5 announcement: WebGL, a standard that enables hardware accelerated 3D animation in HTML5 (you can see some examples here). Unlike WebSockets and other unfinished specs, WebGL is now a complete specification with support from every other major browser. Microsoft, however, has no plans to implement WebGL.
Here’s the statement the company gave us this week:
Browser vendors are implementing WebGL as a way to get partial hardware acceleration after developers rewrite their code. With Internet Explorer 9, developers receive GPU-powered hardware acceleration without rewriting a single line of code which we feel is a better outcome. We look at the real world patterns of the things developers are building today and want to build tomorrow, and balance this against the risk with including things that will likely change in the future. Sometimes there seems to be a race going on to try to make headlines “by claiming to be first” to implement specifications. We don’t think that is healthy. Interoperability is not about being first. Instead, we wanted to provide a stable platform for the web. We focused on building test suites with the W3C for specs that were considered “done”, but where there were real interoperability gaps. These are not that the “sexy” problems that get talked about.
With HTML5Labs.com we are focusing on creating a world-class environment for developer to experiment with emerging web specifications. As an example, we have updated our WebSockets implementations 5 times in just three months.
Notice the Websockets mention. Microsoft is obsessed Websockets, it even created one of those text-to-speech animations to developers how horrible Websockets is and how Microsoft is protecting them from unfinished standards. But it’s a classic misdirection. OpenGL is not Websockets.
And judging from the comments on the IE10 announcement on the official IE blog, developers want WebGL.
Internet Explorer General Manager Dean Hachamovitch told CNET last year. “I think it’s different markup. You’re telling developers, ‘Go write something else.’”
In other words: Microsoft thinks it knows what developers want better than they do. I suppose it’s the Steve Jobs approach – maybe it will work for them. Native HTML is arguably “better,” and maybe Microsoft can force other browser makers to support hardware acceleration and lower level OS integration to speed up rendering and animation without the need for WebGL.
That, or Microsoft is ignoring an accepted standard in favor of its own technology – and setting the groundwork for the “bad old days” of Web fragmentation all over again. I wrote earlier about the mixed messages Microsoft is sending – on the one hand saying that developers should be able to write markup once and run it anywhere, and on other saying developers should create alternate versions of their sites for non-IE9+ browsers.
It’s a shame. IE9 and 10 look to be great browsers. But the harder I look at what Microsoft is doing, the harder it is to believe that its browser strategy developer friendly.
Disclosure: Microsoft paid for Klint Finley’s travel and lodging to attend MIX, and MIX is a ReadWriteWeb sponsor