The notion of a “browser war” sounds so last-century. It’s over, Microsoft won, we’ve moved on to bigger and better things — like service-oriented architectures. Unfortunately, browser peace is short on satisfaction, as I was forcibly reminded this week while preparing to give a presentation. I flirted with PowerPoint 2003, but unlike other Office 2003 apps, it doesn’t emit XML I could transform into the SlideML format the conference planners had requested. I also flirted with the OpenOffice PowerPoint-alike, Impress, which does write XML. But neither application primarily targets the Web. In this century, that’s a nonstarter — at least for me. So I dusted off an old HTML/CSS/JavaScript technique from 1998 and took another run at it.
The good news is that everything still works the way it used to. Better news: the browser environment isn’t the monoculture that it was. Mozilla has emerged from its long nuclear winter to become a pillar of the Linux desktop. Alpha geeks everywhere (including Sun and Microsoft) are running Safari on their PowerBooks. But here’s the reality check you knew was coming: cross-browser and cross-OS compatibility remains nearly as elusive as ever. I won’t bore you with the details. Let’s just say that testing CSS and JavaScript effects on the three major OS platforms, in six different browsers, isn’t a good use of anybody’s time.
There’s no single culprit. I hammered out a compromise solution for my upcoming talk because neither browser was a clear winner. Each is ennobled by unique strengths and plagued with infuriating bugs. In nirvana, we’d have all of the former and none of the latter. Why can’t that happen in the real world? Oh, I remember; it’s because platform vendors compete on their differentiation. That’s why Apple’s TCP/IP is faster than Microsoft’s but has a TCP sequencing bug, and Linux’s TCP/IP only works with newer versions of Windows.
I’m kidding, of course. There is innovation in IP networking but it is not a platform. It’s commodity infrastructure.
The browser is commodity infrastructure too, but in a remarkably broken way. When I shut my eyes and pretend that doesn’t matter any more, I can still read the ugly flashbacks inside my eyelids. Recently I spoke with the vendor of a Web services management product. In 21st-century style, the product is a collection of services, thus affording the maximum leverage for integration. Of course, the management console for this impressive piece of work is, you guessed it, the browser. And while other service-oriented products can in theory be integrated into its console using advanced XML techniques, that’s not what happens. “The truth,” said the vendor, “is that URL redirection is still the best integration strategy.” In other words, link from their $US50,000 HTML/CSS/JavaScript console, which they did (or didn’t) test on three OSs in six browsers, to somebody else’s $50,000 HTML/CSS/JavaScript console which was (or wasn’t) subjected to the same battery of tests.
Call me a dreamer, but I think we can do better than this. Hand-waving about service-oriented architectures doesn’t make the problems go away. It won’t be fun, and it won’t be sexy, but we’re going to have to grow up and win the browser peace.