A quick clarification on Molly’s otherwise excellent post on Microsoft’s fear of updating IE: Mr. Schare doesn’t preclude improvements to IE’s standards support altogether. Indeed, neither Molly nor Tristan say otherwise — though my slow brain did get that impression at first.
The truth is, Mr. Schare says quite the opposite:
Right now we’re aiming for Longhorn for that because we think it affords us the opportunity to say, “Okay a few things have changed, if you want your apps to work with Longhorn you may have to make a few changes.” Versus just blanket upgrading the installed base with some new features, and “Oh by the way we broke a bunch of stuff.”
That’s an encouraging statement, if a bit disappointing. On the one hand, it means Microsoft is going to update IE’s standards support. So far as I know, that’s the first time we’ve gotten a commitment to updated standards support from Microsoft. Yay!
On the other, it means we’re going to have to wait unitl 2006 for Longhorn to get it, and we’re likely going to have to wait much longer for even a slim majority of Windows users to upgrade, because Mr. Schare has effectively ruled out rendering engine improvements before then. Drat.
As for his argument that updating IE would ‘potentially break a lot of things’, I’m less enthusiastic about DOCTYPE switching than Tristan is. Think about this: what if Microsoft failed to fix even just a few of the glaring bugs in IE, but did add support for child and attribute selectors? That would break a fair number of my sites, DOCTYPE switching or no. And whatever else you might say about Microsoft, they most definitely work hard to avoid that sort of thing.
What I do find utterly nonsensical, however, is the idea that IE’s support for stadards is deficient only in ways that, if remedied, might cause such breakage. Sure, suddenly changing IE’s faulty overflow model might break a site or three. And even fixing something so obviously wrong as the Magik Creeping Text bug might trigger a chain reaction of new bugs and oddly altered behavior throughout the gnarly old Trident rendering engine. But how on earth would adding support for alpha transparency in PNGs break anyone’s site?
Back when WaSP was formed, we heard one question over and over from Microsoft developers, team leads and product managers: show us even one site that breaks because IE doesn’t support X. This drove some of us into french-fried fits because of course we couldn’t show them any broken sites — we were all spending 20% of our time working around those very bugs so our sites wouldn’t break, and we were tired of it. That was the whole point!
Well, now the shoe’s on the other foot guys. Show me even one site that would break if IE started supporting alpha transparency in PNGs. Show me one site that would break if IE started supporting position: fixed. Or hover: on non-link elements. Or border-spacing. Or generated content. Or. Or. Or. You get the idea.
Mr. Schare might well argue that adding those things might have other, unintended consequences in the rendering engine. Given that the Trident rendering engine is now over seven years old and has suffered considerable entropy, he might also be right. But that’s not a case of ‘changing the platform’ and breaking people’s sites because it no longer behaves as they expect. It’s simply the risk of introducing bugs by making changes to the code that Microsoft faces — often quite successfully — every time they release any sort of patch or update.
Update: Be sure to have a look at Eric Meyer’s thoughts on the matter. With apologies to Molly and Tristan, it’s the best-thought-out, best-argued response I’ve seen to date.