Longhorn Trimmed

Today Microsoft released some more concrete information on their plans for Windows "Longhorn". You can read the press release over on Microsoft's site, but I suggest you fasten your seatbelt first; the spin on this one is pretty severe. Let me enumerate the salient points without the hype:

  • Broad availability for the Longhorn client operating system in 2006
  • WinFS removed from Longhorn; to be in beta when Longhorn client is released
  • WinFX (Avalon + Indigo) decoupled from Longhorn. Will be available for Windows XP and Windows Server 2003 in 2006
  • "Anticipated availability" for Longhorn server in 2007

So, what's it all mean, Mr. Wizard? Well, in the first place, it's very clear that Microsoft is throwing features overboard in order to make some kind of reasonable ship date for the next client. The most substantial of these is WinFS; long-time observers will recall that this technology was first promised as ObjectFS in "Cairo" a decade ago. That may be some sort of vaporware record. Also, by separating WinFX (Avalon being the new display technologies, Indigo a communications subsystem) from Longhorn, Microsoft has created some wiggle room if ship schedules get even tougher. They could, for example, ship rudimentary forms of these technologies in Longhorn and then brush them up in a service pack that coordinates with the anticipated shipment for XP/2003. Personally, I wouldn't put too much stock in that 2006 date for WinFX either; Microsoft's record of hitting ship dates is not great.

Second, there are gonna be some upset developers out there. Microsoft handed out pre-alpha bits are the PDC about a year ago, and got everyone pumped up about Avalon, Indigo, and WinFS as the fundamental "pillars" of Longhorn (along with a fourth, "basic" pillar that covers all the miscellaneous improvements). Since then, there's been a steady stream of hype in MSDN Magazine and elsewhere, designed to convince people that WinFS is real technology that will be useable soon. Now it's slipping backwards.

Finally, I'm skeptical of the plan to make WinFX available on older operating systems. At first blush this sounds like a way for applications that use WinFX to have a much wider reach of potential client machines. But I fear this is going to run up against the limitations of the installed hardware base. It seems pretty clear that Avalon is going to require a hot-shot video card that you won't find in many existing computers. As for Indigo, well, there are certainly some people excited about the latest Web Services standards, but on the whole I still think that's a niche market. For the most part, WinFX is going to be about creating exciting new user interfaces - which will require exciting new hardware no matter what operating system support is available.

Does it really matter? Well, for many of us, this announcement is just a confirmation of something we already knew: Microsoft has an incredible amount of trouble coming up with plans for a Windows release that they can actually follow through with. If you flash back to PCD 2001, you may remember Bill Gates confirming Longhorn as a Windows client release for 2003 that would contain many new graphics and peer-to-peer technologies. But what's three years of slip between friends?