On vendor prefixes

There’s been a lot of noise lately about vendor prefixes. The sort of noise normally produced when web developers discover that something they thought was merely annoying may in fact be fundamentally broken.

It all started earlier this week during a CSS Working Group meeting, when representatives from some non-WebKit browsers began considering support for the -webkit- prefix. In essence, other browsers pretending to be WebKit because too many websites assume WebKit is the only browser in existence. Nothing could possibly go wrong.

The idea prompted some fairly heated responses from industry veterans like Remy Sharp, Christian Heilmann, Bruce Lawson and Eric Meyer.

To quote Remy Sharp: “this is bat shit crazy”. Hard to improve on that.

We have been here before, of course. More than a decade ago, Internet Explorer and Netscape were busy fighting the Browser Wars, each introducing its own proprietary tricks in the hope that gullible developers would choose sides. Innovation did happen, but interoperability and accessibility were frequently dragged into an alley and beaten with a blunt instrument.

Eventually the web standards movement arrived and convinced everyone that perhaps having one web, instead of several mutually hostile versions of it, was actually a good idea. And yet, here we are again.

This time the problem is the enormous popularity of WebKit, helped considerably by the explosion of mobile browsing. Too many developers are using -webkit- properties without bothering with equivalent properties or other prefixes. Build enough sites like that and suddenly a non-WebKit browser has a choice: render the web correctly according to the standards and look broken, or start understanding somebody else’s proprietary prefix.

It’s a little like arriving at a restaurant where everyone has decided to speak Italian because the chef once worked in Rome. You can insist that this is Sweden and Swedish is perfectly valid. You may even be technically correct. You’re still not getting lunch.

There doesn’t seem to be an obvious miracle cure. Browser vendors could eventually remove support for prefixed properties as the standards mature, but browser vendors are understandably reluctant to remove anything once thousands of developers have started depending on it. That’s one of the web’s less charming traditions: temporary solutions have an uncanny ability to become permanent infrastructure.

Presumably the clever people involved will find a way through this mess. Until then, perhaps developers could try the radical approach of remembering that the World Wide Web contains more than one browser. Crazy, I know.