Post a Comment
It is based on singularity, and singularity has some very modern and innovative ideas. NT is based on VMS, which is fairly conservative as kernels go.
The big shame about singularity was that it has been a research project for years now. It went under MS-RL a few months ago, but that still kept people from making a working implementation (reference license basically means read-only source code)
If this actually takes off, I would say it would be the most interesting (serious) project in the OS space since the BeOS microkernel.
you don't have to buy it, singularity is free to download and run/compile.
It obvious that Microsoft has to look ahead, we have newer technologies and newer methods and needs for using the technologoies.
I agree that the NT Kernel and most the NT systems are good, however how long can Microsoft keep extending these systems to take advantage of newer tech.
I think the biggest requirement for an OS is to take advantage of mutliple cores and to provide an easier way for applications that run on the OS to take advantage. We see apple going this way with Grand Central, intel have stated that they see future chips with hundreds and thousands of cores, we have GPU's which contain a wealth of computing power.
Microsoft really excell's when they are put under pressure, although sharing the same kernel the difference between Vista and Server 2008 is night and day, proving that the userland/UI is the biggest problem in Vista, we have Mac OSX and Linux (unix's) that will hopefully push Microsoft to get out of this medocracy pattern they have been in for the past 8 years and start giving some real value for money, everything after Windows 2000 seems like a can't be bothered slap some new paint on it approach.
Edited 2008-07-10 08:46 UTC
If you read the article, you would see
A) It is not a rumour
B) it is based on singularity
Chances are if v1 is done in 9 years, and we use NT adoption as a reference, we are talking 16 years before this would become mainstream. That means this is a "looking ahead" type of project.
I have a feeling that we'll start to see this in niche situations; possibly in the portable area first, then followed by maybe HPC, then work its way to the middle - then again, given the new devices these days, I have a feeling that eventually the desktop's days will be numbered.
I'd love to see this as a success - even as a Mac user, the added competition will improve product quality over all. Its too bad, however, that Microsoft does have a reputation for taking a wonderful idea and completely butchering it. Case in point would be Windows NT, and the silly things they did in NT4 (moving graphics into kernel space).
Actually, if you look at Linux, only a *VERY* small amount exists in kernel space, infact, it is more of a 'door way' for the Xorg server driver to speed up talking to the video card.
The problem is that with Windows is they threw the WHOLE driver into kernel space. Rather than keeping the heavy loading in user space, and have a small short cut, they improved 'teh snappy' through brute force. Windows Vista has changed things, and from what it appears, more of the driver now sits in user space, which allows graceful recovery rather than a BSOD.
As for Mac, I'm unsure how things work in the Mac world given I haven't really investigated all that heavily into how the whole graphics layer works. What ever the case, it seems to do the job required with minimum fuss and bother.
NT was originally very microkernel-ish and sensibly put a lot of driver stuff in userspace, only later on moving graphics into the kernel. MacOS, which itself tries to be microkernel-ish, these days does what Windows does now and throws graphics into the kernel, for performance reasons.
Linux is much the same, though as a monolithic kernel it bears no shame for putting graphics stuff into the kernel. *nix systems in general try to leave as much as possible to X, but as I understand it a good deal winds up in the kernel anyway.






