Microsoft Ponders Windows Successor
InfoWorldMike writes "Before Vista is even out of the gates, a Microsoft exec was talking Wednesday about Windows' replacement at a VC conference. Speaking at The Venture Forum conference, Microsoft's Bryan Barnett, a program manager for external research programs in the Microsoft Research group, said multicore architectures are of particular interest when weighing what to put in future operating systems at the company. "Taking full advantage of the processing power that those multicore architectures potentially make available requires operating systems and development tools that don't exist largely today," Barnett said. Well, with Vista in the pipeline as long as it has been, you must admit it is not surprising Microsoft is taking the long-term view. And it won't be built overnight: There is no timetable for a Windows successor right now. But early work on this effort has not yet been organized, with five or six small projects afoot in various places throughout the company, Barnett said."
Vista won't fail, it doesn't really matter if it's far better than XP or not. Will Vista be more stable or secure than XP? Probably not. Will that matter? Probably not. It will look different and all the PC manufactures will preinstall it on their machines, everyone that buys a new PC will get and use it, and within several years after it's release it will be used by the majority of PC users (since the majority will have bought a new PC by then). Meanwhile the Mac lovers will call it a cheap ripoff of Mac OS X (which it probably is) and the Linux users will say you can get that stuff for free (watch the demo of Novell Linux 10 with xgl, it demonstrates all the cool windows effects MS is saying will be in Vista, and then some). But the majority of PC users won't know or care. To them it's a new feature when it shows up in MS Windows. The only thing that will break the cycle of everyone adapting MS's newest OS is the ability to effortlessly run Windows apps on Linux, or Mac. It's sad but true.
I think the key point to keep in mind here is not that Microsoft is looking for a successor to Windows, but that these statements came from "a program manager for external research programs in the Microsoft Research group". This is what Mirosoft Research does. They come up with blue-sky ideas like replacing Windows entirely, and then the product groups integrate those ideas into real, shippable products. As an example, the "Drivatar" AI used by Forza Motorsport came directly out of MSR. The researchers had grand plans for the technology (get real motorsport "legends" to generate drivatars based on their driving style, learn from the player as he's playing, etc), while the implementation in Forza was more practical (the main AI was based on pre-release training and didn't learn from watching the player, there were no "professional" drivatars, the player had to actively train his drivatar in specific sessions rather than having it learn while he plays, etc). That's not a bad thing, and it's still a damn sight better than most other racing game AI out there (Gran Turismo, I'm looking at you. Damn retarded bumper car AI ...). Researchers are good at coming up with crazy ideas and sample implementations that don't take into account the rest of the system (back to Forza, there's only so much processing available in an Xbox to handle all of the physics and AI, which means that real-time drivatar training wouldn't be feasible). If you know what to look for, you can see many Microsoft Research contributions in shipping products (speech, grammar checking, natural language processing, etc in Office; anti-phishing in the MSN/Windows Live Toolbar and IE7; pretty much the entire backend for MSN/Windows Live Search; and so on), but it's only bits and pieces. Go poke around, look at the many areas of research going on at MSR. Take a look at their sample code. And then remember that when you see a similar but less-grandiose feature 5-10 years from now in a real, shipping product.
Note: I'm neither a Microsoft researcher nor a Forza developer, so all of the information above is what anyone can deduce from the sources I cited.
Microsoft has already done this to a fair extent with Terminal Server. The main thing to keep in mind is that the main bits in kernel space really are drivers, not the UI framework (and even that's changing with Vista). Terminal Server is very much Microsoft's X. Do you remember the "Fast User Switching" feature in Windows XP? Yeah, that's Terminal Server, and what it really means is that every time you use the Windows UI (in XP and 2K3) you're actually interfacing through a local Terminal Server session (just like X!). Of course, TS will have its little differences when running over a network, like not supporting video overlays or 3D acceleration, but in most case
The next Microsoft OS is quite likely to be based entirely on interpreted/dynamically compiled languages, obviously the CLR. The actions over the last 2 or so years seem to indicate that Microsoft wishes to deprecate native code. They would probably run existing x86 Windows programs in a sandbox so that untrusted code (aka all native code) cannot damage the system. The OS would deny even the computer owner the right to run native code with any authority unless it's signed by Microsoft. We can already see this coming with Vista: unsigned code cannot run in the kernel at all in x64, and in all versions unsigned code cannot request that dialog box to ask the user permission for admin access. (This last one was never announced by Microsoft and was slipped into a build. Developers filed it as a bug; Microsoft declared "as design" with no comment whatsoever.)
.NET programs could run unsigned. (They'd probably require signing to do anything interesting like write files to disk.)
It works great for DRM, because sandboxed code cannot manipulate other code. If implemented correctly, something that Microsoft has shown to be possible with the 360 (though with native code), it would be unbreakable other than at the hardware level. Microsoft would make it so that only Microsoft-signed programs are allowed to run natively, whereas
This is terrible and I hope Microsoft meets a lot of resistance.
Melissa
"Screw Sun, cross-platform will never work. Let's move on and steal the Java language." - Visual J++ Product Manager
Didn't Vista start out pretty much the same way? "Total rewrite from the ground up", everything shiny and new, new paradigms for file system handling and coffee making?
Look what we ended up with.
History repeats itself, repeats itself, itself...
Assorted stuff I do sometimes: Lemuria.org
Yes.
And no.
Barnett's quote of "Taking full advantage of the processing power that those multicore architectures potentially make available requires operating systems and development tools that don't exist largely today," is meant to obfuscate the fact that there are OS's that handle multiple processors very well (Linux and OS X, not to mention other unix variants).
Microsoft has a vested interest in not doing PR work for the 'nix community. And they certainly don't want to imply that Vista won't get the most out of the current crop of processors when other OS's will.
Mark my words folks, we're currently watching the Fall of the Roman Empire. Nero (Ballmer) is fiddling (throwing chairs during temper fits, screaming "Developers!" repeatedly, etc.) while the city of Rome (Redmond) is burning to the ground.
I guess the capitalists were right, leave the marketplace alone and eventually it will find a center and select a survivor. In the OS wars, my money is on unix (in any flavor, take your pick) as the eventual winner. I'm sure Bill Gates knows this, that's why he's bailing while he can, just as he bequeathed the empire to Ballmer years ago when the DOJ was breathing down MS's neck. Gates is a lot of things: Stupid isn't one of them.
Nitewing '98
Everything works...in theory.