Avalon Emerges From Longhorn's Mists

Microsoft has started to release the results of OS research it's been conducting behind closed doors for several years. The next version of the Windows client OS—code-named Longhorn and due for release in 2005—is early in its alpha cycle. Pieces of it that have emerged include WinFS, a file-organizing service based on Yukon (the code name for the upcoming version of SQL Server). The user will interact with WinFS, but it won't replace NTFS as the Windows file system. Longhorn's client UI—code-named Aero—should include significant new features, such as peer-to-peer networking and real-time collaboration.

A technology code-named Avalon (also called DCE) lies at the heart of the new interface and consists of its core graphics engine and APIs. Avalon makes the jump from Windows XP's UI to Longhorn's potentially as big as the one users and programmers experienced in moving from DOS to 16-bit Windows. You can compare the difference to interacting with a game as opposed to a business application. The realism of a game's graphics and sound, and a variety of subtle cues, immerse you in the game experience. You interact with the game almost instinctively, which lets you focus on its objective. Avalon restructures Longhorn's underpinnings to allow the OS itself to support this level of user interaction with all types of applications, enabling users to focus on doing their jobs better rather than on the software they're using.

Avalon abandons GDI/GDI+ and relies heavily on Microsoft's experience with DirectX and high-end gaming. The graphics possibilities are explosive, but the programming interface remains close to the simpler GDI+ interface. Avalon renders to multiple independent buffers, which solves many UI issues, including transparency and the way windows behave when moved. It's also blazingly fast, particularly in running video. The cueing and scaling aspects are exciting. Cueing lets your application interact with its users more subtly. (With luck, Microsoft will provide standard guidelines for cueing.) Avalon discards the idea of a pixel, moving to a resolution-independent unit to support scaling. Users will be able to change a form's scale, displaying smooth-edged elements at every size.

Longhorn relies on Moore's Law remaining in force—on the availability of sufficiently advanced hardware by 2005 to support the OS. Avalon will degrade in two steps to support significantly weaker hardware, but the full-blown Longhorn experience requires a beefy machine with 64 MB or better 3-D graphics.

Microsoft will integrate the .NET language runtimes into Longhorn at a low level and chop down the Windows API list—currently at tens of thousands. Programming the UI will be XML-based. It isn't clear yet how these aspects of Avalon will integrate with the .NET programming experience, but Microsoft will no doubt push the technology out to developers fairly soon so you can begin to deliver applications for the new OS.