Thoughts on Longhorn: I agree with most of them. Aero/Avalon seems like Apple’s Quartz and Quartz Extreme, both of which are nice but not revolutionary. They make the windows look nicer, PDF composition trivial, and offload to the third-party video cards. You don’t hear Apple trumpeting the Quartz Extreme revolution.
WinFS is intriguing because it’s basically moving the file system to a database. BeOS did that and I liked the idea of real-time, super fast searching of text. Every time I hear about WinFS, though, it seems to consist of more and more Semantic Web hype. Semantic Web is nice and all, but it requires a lot of upkeep. Most users will never use it because the payoff is pretty abstract and long-range. I consider myself above average in willingness to include metadata—I add in acronym tags to these blog postings and title my iPhoto pictures—but I hardly ever put metadata in Word documents and I can’t imagine telling my file system more about my voluminous documents than a file name.
I especially liked his point about the requirement to use .NET managed code. Apple provides two major frameworks in which to program Mac OS X applications and is pretty agnostic about which is better. I would wager that Microsoft will learn from Apple’s example and still allow unmanaged code. They’ll probably end up doing it through a hack, though, because their Trusted Computing Initiative ostensibly precludes the use of unmanaged code.
Microsoft seems to be following their normal pattern of vaporware pronouncements that don’t quite make it to the final product. I also read somewhere that .Net My Web Services will be making an appearance in Longhorn, as will Palladium. Uh huh, methinks there won’t be any long lines ‘round midnight on some distant 2007 day.
[UPDATE (11/9/03): More on WinFS.]
[UPDATE (11/12/03): A Microsoftie perspective on WinFS; I hope they’re right about the ease of metadata entry because, as Tim Bray said, “there’s no such thing as cheap metadata.”]