32 MA XIMUMPC OCTOBER 2004
T
o date, we know of five major chang-
es to Longhorn: a new, improved 3D
user interface, a new database-like
file system, features that will make gaming
more accessible and convenient,
revamped digital rights management
(DRM), and a recently announced “Castle”
feature, which is an interesting and easy
new way to share user profiles and data
across small home networks. All this new
technology makes Longhorn the biggest
operating system upgrade since Windows- Let’s start with the 3D user interface.
Introducing Avalon
Longhorn’s 3D graphical user interface,
code-named Avalon, could revolution-
ize the way you access your PC. As you
might imagine, Microsoft has kept the
details regarding Avalon under tight
security (to read about how it might
look, turn to page 34), but we know this
much: Avalon isn’t just a 3D GUI for
Windows. It’s actually a whole new way
to draw user interfaces for applications—
in 3D and 2D.
Currently, programmers of non-Avalon
Windows applications create their interfac-
es using a special program called a form
designer. Using this form designer, every
single button and field is associated with a
function in the program. Then the form is
saved as part of the application. In WinXP3D content drawn by a Direct3D-compatible
3D accelerator requires a totally different
set of programming rules, so programmers
have thus far been unable to add 3D effects
and elements to a 2D user interface.
With Avalon, user interfaces (and even
some simple programs) will be writtenusing a new language called XAML—
extensible application markup language.
This means that instead of using an eso-
teric program designed specifically to cre-
ate user interfaces, developers will be able
to write and edit XAML files with a simple
text editor. Even cooler, the Avalon XAMLWINDOWS LONGHORN
Say goodbye to the old, flat Alt-Tab experience. The new 3D task switcher lets
you see the contents of your windows while you flip between them.Goodbye Direct3D—Hello
Windows Graphics Foundation!
Because Longhorn throws out the old-fashioned 2D Windows
Desktop in favor of a new, 3D-rendered Desktop (every Window
will literally be a 3D object!), it’s crucial that Windows’ 3D-render-
ing pipeline be rock solid.
This is where the Windows Graphics Foundation comes in.
In some ways, WGF is nothing more than the next version of
Direct3D, but it sports a lot more functionality than earlier, gam-
ing-oriented versions of Direct3D. No longer is speed the primary
concern; with WGF, reliability and stability are vital.
Until 2002, most graphics processors were designed with
a singular task in mind: running games as quickly as possible.
Multitasking, recovering from crashes, and being a responsible
Windows citizen were minor concerns compared with the
almighty frames-per-second. But because the 3D-rendering pipe-
line is going to be responsible for the brunt of Windows’ force, it
must be subtly reshaped now—before Longhorn ships in 2006. Inparticular, there are two new improvements to the graphics pipe-
line that Microsoft must perfect if WGF is to be ready on time.
First, the driver and graphics card must work together to
provide seamless fault recovery. If the video driver crashes, it
needs to restart and resume operation without missing a beat.
Similarly, if a 3D application crashes, Windows needs to be able
to recover and quickly resume operation. Forcing the user to
restart Windows because of a graphics crash is not an acceptable
solution, as people familiar with Windows 3.1 and Windows 95
can attest.
Second, the GPU needs to be accessible to multiple apps.
There needs to be a way for the OS to schedule batch process-
ing on the GPU. Ideally, the OS will be able to add graphics
tasks to a queue, and possess the flexibility to preempt less
important tasks (like re-rendering a background window) for
more important tasks (like redrawing the foreground window)
whenever necessary.
WGF 1.0 should also give game developers new levels of
flexibility such as we’ve come to expect from major new revi-
sions of Direct3D.