Submitted by “According to a new report, there are as many as 6,000 references to MinWin in an internal Windows 8 client build. This may provide more clues and validation of Microsoft’s virtualization plans for the coming operating system release.”
“According to a new report, there are as many as 6,000 references to MinWin in an internal Windows 8 client build. This may provide more clues and validation of Microsoft’s virtualization plans for the coming operating system release.”
http://www.osnews.com/story/25026/Windows_8_Hyper-v_and_MinWin_A_Ga…
Its just someone else’s summary of the same base article. “MinWin” is at the core of every Windows OS since Vista. The title is windows speak for ” we are going to incrementally improve the kernel in the next release”. Shocking!
Anyone remember the original BeOS design. Have a hybrid kernel, with a bunch of servers / kits which can be used as required instead of being initialised all at once. Sounds exactly like what MinWin is trying to accomplish. To use a quote from the internet:
MinWin contains the bootloader, HAL (processor & chipset initialization, memory manager, etc), scheduler, most of the kernel infrastructure (sans most drivers, filesystems, etc) and some of the core Win32 subsystem upon which essential apps can be run.
To MinWin you can then configure an OS build by choosing which OS components you want to include from a catalog of items – NTFS, (ex)FAT, etc., TCP/IP & NETBIOS, SCSI/RAID, video, audio, printers, .NET, etc.
Yes, it does sound familiar… That’s how almost every operating system except windows has worked. People get way too worked up about min win. Its an internal ongoing refactoring of the windows kernel. Its major feature for end users is that it will be easier and faster for Microsoft to add new features to windows with less problems. If Microsoft actually releases windows 8 on time with the crazy new metro ui interface with programs written in html 5, that will be the testament of winmin’s success.
Well WindowsCE is also modular like you are describing.
Usually OEM vendors combine WindowsCE components the way it fits their devices.
As for the Windows kernel, the userland might have the problems we all know, but the kernel is quite good and it really shows the VMS heritage.
Don’t really understand which windows kernel you are talking about. We were led to believe that the NT kernel was pretty good and userland sucked, up until the disaster of longhorn, which was primarily caused by the mess that was the kernel. Since then they have been trying to reorganize it into a more modular design.
My understanding is that windows CE is not the windows kernel because the windows kernel was not modular enough.
The NT family. Actually reading this book is quite illuminating
http://technet.microsoft.com/en-us/sysinternals/bb963901
As well as reading old VMS documentation and seeing some patterns with early NT APIs.
But you’re also right, sometime along the way things got out of control and MinWin is a welcome change.
True, but from the applications point of view, most APIs are the same.
Well, WNT = ++VMS in Caesar cipher, after all…
“Way too worked up”
I think if one is to get worked up about anything in the world of MS Windows, this is as good a thing as any. These are base-level improvements upon which the OS can be more easily improved and evolved.
Exactly, this is just a case of “everything old is new again”. Dave cutler originally designed the WinNT kernel to be portable and modular but the machines of the time simply weren’t powerful enough and it caused performance issues, especially with graphics, so they moved more and more into the kernel.
Now that machines have gotten powerful enough that even a handset can afford the overhead they will be able to build a version of Windows for any arch and role they desire, simply by adding or removing modules.
let me say while the IDEA is sound from the looks of things the execution will be a giant FAIL. Why is that? Simple…right now on my local Craigslist there are plenty of ARM netbooks being sold cheap, because the shady dealer sold them with “Windows compact Edition” aka WinCE. Now folks see the XP desktop and they don’t know that WinCE will NOT run Win32, they just see the word Windows. When they found it wouldn’t run their stuff? they dumped.
Sadly you watch MSFT is gonna try to yet again take a page from the Apple playbook and use the minWin concept to push Windows onto ARM where it will fail horribly. Folks will see “Windows 8” think they can run their Win32, and when they find out they can’t they’ll be returning these things en masse.
Why they would want to damage the Windows brand like that is beyond me, but if it is one thing we’ve learned it is Ballmer isn’t qualified to shine Bill Gates’ shoes. They should stick with WinPhone on ARM, just as Apple uses iOS and NOT OSX on tablets. if Ballmer thinks magically using the word Windows will get him anything but a po’d consumer base when he puts it on ARM he has another thing coming. The only upside I see to MinWin is maybe Ballmer will finally be forced to ‘pursue other interests’ and they can bring in someone with skills (maybe Ozzie) as CEO.
For that matter, Windows Phone 7 not only exhibits the same incompatibility with Win32, it also… doesn’t seem to actually have any windows ;p (as in, GUI concept)
Maybe MS should hijack more completely generic, kinda concept-related, words. Considering all the panning action, Microsoft Curtains (too bad the roll-down ones don’t seem to have a “singular” word in EN), or Shutters / Blinds (likewise, both roller ones) might fit. Maybe MS Awnings having less “wrong” connotations (Sudare could be decent, too). And for ARM netbooks – relatively small, portable, always moving windows – MS Porthole / Windscreen / Sunroof?
It *is* really annoying. All the real essence behind “MinWin” is:
1. no special concept at all, Linux has it since ages, other modular designs are much more sophisticated
2. already “in use” since Windows Vista (claimed by MS itself after people complained about Vista missing it)
3. nothing the end user will ever get in touch with — you can compile Linux yourself, but not the Windows kernel
I think even MS folks are ashamed of their marketing department picking it up and milking it for some time. What is it with OSAlert, jumping on the boat again and again all the time? So eager to find *anything* maybe stunning about MS OS development?
I used to do it when Linux came in floppies, I was still at the university and had nothing better to do.
Nowadays?! There is a whole world outside filled with better things to do then compiling kernels.
Yes, last time I compiled my own kernel was in 2002 to get my TV capture card working.
Thank god it is not needed anymore.
However, there are hobbyists out there. And if you want to get in touch with a working example of a modularized kernel (however, not a microkernel), then you can with Linux. You won’t ever get in touch with “MinWin”.
…and there are hobbyists who find the progression of the Windows kernel changes interesting. Don’t like the hype? That’s fine, everyone has an opinion.
My point was exactly that this “news” is _not_ about the progression of the Windows kernel.
Or can you put any substance to it?
p.s.: Just to note that me personally I am highly interested in the progression of the Windows kernel, or any major OS kernel.
Edited 2011-08-09 17:21 UTC
I cannot substantiate anything without access to code. Neither can you. It is a progression if they are continually modifying the kernel, moving previously in kernal api elements out of kernel. Reporting on changes to reference counts, or renaming of components, etc. is of interest on OS News.
So can you prove that it is not a progression? And what “news” would you rather see? You claim you are interested in the progression of the windows kernel or any major os kernel yet you immediately dismiss any apparent (and perhaps unsubstantiated) claims that references in the kernel have changed and might indicate something positive in the kernel design.
I see what you want to say, but doesn’t that boil down to mere statistics then? Or the assertion, that _something_ is changed in the kernel (instead of it being stalled)? Or what?
What I find interesting news about kernels:
+ new features
+ design decisions
+ security/performance overhauls
+ … anything that is tangible
Like: New filesystem (WinFS, although it died), subsystems in user-space/kernel-space (Windows Vista sound architecture), security improvements (randomized logical address space, nx bit), throughput vs. responsiveness (linux scheduler debate, sorry I have no windows example here)…
But this. What is this? There is no information leaked ever about how the modularization of “MinWin” is done. Is it like in the linux kernel, more like in a micro kernel, a new design? There is no information about when the transition took place and how it affects *anything* in the operation of the kernel. All that was said is first “what a great feature of windows vista this will be” and then “oh see, you didn’t find any minwin in windows vista, well it is there!!”.. and now this speculation, like if you would read from the stars: “may there be minwin in windows 8?!”.
What can you really learn from this? Nothing.
Here are a couple of links:
http://arstechnica.com/microsoft/news/2009/11/inside-minwin-the-win…
http://www.betanews.com/article/Mark-Russinovich-on-MinWin-the-new-…
From this you can learn several things:
1. Minwin is a user space refactoring project; it doesn’t change anything in the kernel.
2. The refactoring was done by introducing the concept of “virtual DLLs”, or API sets.
3. The transition happened in win7. There is no minwin in Vista.