“Microsoft said last week it plans to hire 4,000 to 5,000 new workers and to increase research spending by about 8 percent, to $6.9 billion per year. The company also said it has $49 billion lying around collecting interest. So what should Bill and Steve do with these wondrous resources? My suggestion: Start over.” It’s radical, but interesting. Check out the rest on zdnet.
ZDNET: Microsoft Should Rewrite Windows
About The Author
Adam Scheinberg
Technology Executive • Web Developer • Father • Foodie • Music Snob • OS enthusiast
Follow me on Mastodon @[email protected]
The NT codebase did start from scratch, perhaps not conceptually, but was a total rewrite from the prior codebase. And the UI was rewritten in 1995, then passed on to NT. I do agree that Windows needs to be drastically refactored and simplified, but much of the internals would probably remain the same. And as for major UI improvements, nothing much has happened on any OS for a LONG time ….
Starting over from scratch seems like a bit of a naive thing to suggest. Apple didn’t start over, Jobs decided that his NeXTstep OS was better (and it was) so they switched to that for their new machines. Of course if Microsoft wanted, they most certainly have the resources to implement win32 on another kernel, like the Linux or BSD kernel (as they demonstrated with their Unix version of Internet explorer 4; you didn’t think that used Motif did you?). If Netscape had open sourced their Netscape 4 codebase instead of starting over, they would likely have had much faster results (and probably more users) than they do now because they started over. Of course MS did start over they’d probably end up with a much better, more clean OS. But is that really the best thing to do? I’d wager not. Besides, if you dont call winNT “starting over”, then I dont know what you’d call it. The only thing that’s the same is the API.
This guy understands the problem facing MS. Their OS does need a rewrite with a completely different approach to computing compared to windows of current.
Make is clean, make it simple and make it powerful and alow the user to get their hands dirty if they want to but hide the complex settings away from Joe/Jane Average.
Hey whilst we’re at it could we have open file formats and adherence to open standards?
One can dream can’t they?
Microsoft cares too much about being profitable. Rewriting software is generally not a profitable venture. Sure Windows sucks, and Microsoft knows it, but it still makes them a lot of money.
What I find really interesting is how Microsoft’s upgrade gravy train has some to an end. Remember all the hype Windows 95 and 98 got? Normal people actually went out and bought the upgrades in droves. Now nobody gives a hoot because they have Windows 2000/XP and it’s good enough. I bet Longhorn will be an “Upgrade Version” flop, just like Server 2003 is turning out to be. People will just get a new OS with the new hardware. Yet even with the “Upgrade Version” ideal fading away, it seems to not even hurt Microsoft, and I find this surprising.
API-wise .net is “starting-over”. The OS is getting bloated and less-responsive (especially XP, even when running on good hardware). Win2K will always be the best Windows in terms of speed, reliability and simplicity. WinXP is icing on the cake and I sometimes question whether I should’ve even purchahed it.
But wouldn’t it be nice if at least some decent-sized group really loved it? I mean, couldn’t Microsoft create an operating system that users really like? They haven’t done so yet.
————
Who is he kidding?
ZDnet. Zzzz. Must be a slow news day.
Oh well, I hope he met his writing quota so he can get paid this week..
I’ll wait for Longhorn. It should be quite radically different…but we’ll see.
“If Netscape had open sourced their Netscape 4 codebase instead of starting over, they would likely have had much faster results (and probably more users) than they do now because they started over.”
heck no. NS4 was a bunch of speghetti code from what i hear why do you think they started over? becuase they couldnt use the same code to do what they needed to do! its not like you cant dl NS5a mozilla has the code for you to compile. maybe even binaries?
“The company also said it has $49 billion lying around collecting interest.”
Donate a few billions to OSS…….
Most of his complaints in the article are already being addressed. The .net framework is MS’s answer to these problems.
There is nothing wrong with the design of NT on the kernel level. Its a robust and proven design going back to the 70s.
There may be some bloat in the OS that could go but there is hardly a reason to start over from scratch. They already started over when they built NT the first time.
i want a modular os. interchangible default apps. open api to allow other apps to use said default apps.
i dont want IM, browsers, media players integrated into the os included is fine but not integrated! probably the only app i think should be integrated that has been is cd burning.
He says MS should use an Open Source *nix Kernel. After all the FUD MS has been spreading about Open Source and how EVIL it is. People would call them for that pretty quickly.
And he talks about the removal of the registry so applications are “easy to install”. While this IS better, MS needs the registry to prevent piracy. Look how easy it is to pirate OSX apps. Its just a matter of copy/pasting. I wish we didn’t have this problem, but we do.
The dude here talking about rewritting the OS is either a mad-linux fan or he needs a checkup. Why write something as beautiful as Windwos XP? Alright it needs higher PC configuration, it has some drivers conflict but so did Apple with OS X now after few days it has all setteled as developers have developed softwares for OS X. Yes OS X is beautiful & exellent to work but it also requires higher configuration (& last time I checked on Red Hat it also needs almost as same as XP config) & windows XP difficult to use? Don’t even get me started with that on Linux! I don’t want to sit in front of monitor starting my OS for 5 min just to check emails.
The point I am saying here is that MS has extra cash is because no matter what open source community says they have a beautiful OS at least for PC platform & Linux will require more at least 5 years to come in terms of XP by that time Windows would have gon way ahead.
& for those who needs speed then use Windows 98 why do you need new version of windows??????????????
Conclusion : Windows is the best option for PC. & I am pretty much willing to spend 200$ on it.
No OS will ever be all things to all users. Coursey should just buy himself a Mac if he is so dissatisfied with Windows. MSFT is designing a new GUI called “Aero” just for Longhorn anyway. MSFT isn’t about to rewrite their OS from scratch while they are experiencing double digit year-on-year revenue growth (11% as of the most recent quarter) just because some crank couldn’t think of anything new to write about today.
MS has moved their platform to the next, called Dot Net.
You need to learn to write man!
And wtf this guy is a pro microsoft fan…
contrasutra: Security via obscurity = BAD.
If Netscape had open sourced their Netscape 4 codebase instead of starting over,
Ahem….they *did* release the code for NS4.x. It was just such a horrid mess that they really had to start from scratch.
I basically agree with this article, though. Windows is a clunky mess of code — code that was originally not designed for secure networks, or being multiuser.
Should it be based on a “Unix” kernel? No, not unless it’s Mach or L4. Although the fanboys will deny this to their graves, a monolitic Unix kernel these days is even more foolish now than it was in 1991. Many of the “problems” which preclude Linux and *BSD’s adoption on the desktop are directly related to the conventional model, which is not easily changed. Solutions tend to be either a) insecure (see: LindowsOS), or b) convoluted.
Using a microkernel (or nanokernel like L4) gives you a stable API and ABI for hardware drivers — all the kernel does is abstract the hardware into userland. You can’t do that under Linux. You can’t do that under *BSD.
Furthermore, it’s relatively easy to emulate other OS’s on top of such a structure. Linux runs on top of L4. MkLinux and the old BSDLites showed that they can run on top of Mach. OS9 runs on top of Mach — it’s just called “Classic.” So, you get this nano/microkernel, and on top you’ve got NuWin and OlWin running side-by-side for applications, using part of a FreeBSD kernel for the networking stack and posix compatibility….. And with new processers like the Athlon64, maybe the x86 architecture will cease to suck so bad at IPC and context switching.
But yes, it does sound like OSX.
While they’re at it at MS, they also need to work with somebody like Intel to jettison the standard PC BIOS. I again say that there is absolutely no reason a computer in 2003 should be able to boot DOS. It’s time to move to OpenFirmware or something like it.
Ugh. Here we go. I am not exactly a Microsoft Windows fan, but I have to tell the more misguided among you that there isn’t really anything wrone with the base NT architecture. That said, Windows as we know it could most certainly stand much improvement. Although welding Internet Explorer (and later Media Player) into the Exporer shell has the nasty side effect of taking the whole damned thing down when a serious error is encountered, the biggest reasons for the flaws in modern versions of Windows are lack of quality control (buffer overflows etc.) compounded by the fact that so many services are enabled out of the box.
If Microsoft were to put more emphasis on QC/QA than on shiny new features (although I am sure that they ultimately *could* do both), and shipped it with fewer servicies enabled, than Windows would be far less vulnerable to attack or bad luck (or less than technical husbands or wives
than it is now, and would certainly not require a complete rewrite.
Actually, I am pretty sure Coursey has a mac. It’s only one mans opinion, to which he is as entitled to as you are to yours, which I respect, but why should MS, or anyone else even care what he thinks? No disrepect, just a observation.
70’s ? Are we talking about the same NT kernel itself or it’s design here model ?
“70’s ? Are we talking about the same NT kernel itself or it’s design here model?”
I was wondering the same thing as well, but I believe that the parten post was refering to the idea of microkernels, as opposed to the actual design and implementation of NT, which began in the late 1980’s.
I may have my history worng, but I do believe that microkernels weren’t a big thing until at least the early-mid 1980’s.
70’s ? Are we talking about the same NT kernel itself or it’s design here model ?
Windows NT is basically a re-implementation of VMS. Many, many things are similar even today. David Cutler worked on both. So, the overall design does go back to the 70s, though they did move the interface to kernelland, etc.
Some links:
Windows NT Architecture, Part 1: http://www.winnetmag.com/Articles/Index.cfm?ArticleID=2984
Windows NT Architecture, Part 2: http://www.winnetmag.com/Articles/Index.cfm?ArticleID=3025
Windows NT and VMS: The Rest of the Story: http://www.winntmag.com/Articles/Index.cfm?IssueID=97&ArticleID=449…
“Parent” post, sorry. This is not my keyboard
Microsoft can do anything it wants with its 49 Gig $ reserve, the mere fact that they have that much must say a thing or two about their business capabilities. Personally, rewrititing is not the way to go, but refactoring, and I’m pretty sure that M$ is savy enough to be doing just that. Legacy support for the #1 OS vendor is still a requirement, they do not have the luxary to start fresh (like BeInc had with BeOS). So the OS may not be the most elegant, but hey, it works, and it’s good enough for 90% of the world. What should M$ do? The same thing they’ve been doing for the last 20 or so years.
I personally prefer the design of BeOS and the polish of MacOSX, but thats just me …
Why would Microsoft want to rewrite Windows? Their current version is good enough to give them 90+% market share. Never change a winning team, they say.
And what’s all that about using Unix? Is here something I’m missing? What advantage does being “Unix-based” offer that other modern OS designs don’t?
We are talking about more than the kernel here. OK, so maybe there’s nothing wrong with the kernels design, but it’s full of buffer overflows (I’m sure we’ve not seen the last one) and sloppy work. So even if the design of it is OK, the code still needs rewriting, if you ask me.
However, the kernel is not the only problem here. The GUI _is_ geting more complex, and the amount of errors the OS and belonging apps give are not decresing.
It’s not only the kernel here, but the whole shabang!
</nalle>
as for major UI improvements, nothing much has happened on any OS for a LONG time ….
So I guess you think OS X is just a revision of OS 9?
They look pretty different to me. Most long time Mac users were surprised by the different interface. It is just Windows that has maintained its UI.
With Microsoft it’s not about having the best platform, it’s about making money. They will only take the course of action which leads to the most money. It might not be in Microsofts interest to have quality software.
At first glance the idea of rewrite sounds good. But it is largely unnescessry.
Microsoft should invest in code clean up. Win2K had 93 Million lines of code. They keep using the same code base, add some stuff here and there, and leave a large portion of the now defunct code still intact. For instance…why in the hell is the windows 3.1 program manager still available in XP.
I think a careful look at what they need is better suited. Designing an OS from scratch would be a very costly and time consuming effort (not that MS is strapped for cash).
As much as I hate Windows it isn’t totally broken…it has actually improved a lot.
The article brought up OS X. That wasn’t really a new os. It was based on NeXT. They had a working OS that was built up and refined and optimized with later additions to OS X. MS should consider a new interface (OS X vs. NeXT and Classic is a good example). The taskbar gets crowded way too easily.
Bottom line…just dust and polish, don’t buy new furniture until it is really needed.
Donate a few billions to OSS…….
Yeah…that will happen
There is nothing wrong with the design of NT on the kernel level. Its a robust and proven design going back to the 70s.
That may be true right now…but MS wants to weave IE into the kernel for future releases. That seems like a really bad idea considering how many security and stability problems are in IE now. Weaving IE into the kernel will likely break a lot of stability and introduce serious (not that there aren’t now) security issues to windows.
So I guess you think OS X is just a revision of OS 9?
No, it’s just a revision of OpenStep. I can’t think of any really revolutionary UI or fundamental architectural designs that have come about in a long time. OS innovation, for the most part, is dying, unfortunately.
Commercial companies don’t do OS’s anymore, because they’d have to compete with MS. This really is true for application software, too.
The Free Software world is busy copying or replicating the functionality of other things — not doing anything new. There are a couple of exceptions (i.e. DROPS and GNU/Hurd), but not many. The “killer app” concept of the past is gone. There needs to be a “killer feature” which will make running one of these so much more compelling than running Windows. It could be something like Sun did with NIS+ and NFS. It could be something that hasn’t been thought of yet. Then there’ll be a shift….
And if you still want to run standard Unix, there’s always NetBSD. 🙂
Yieeahh!! Start all over. This time, use FreeBSD as base (like OS x). 😉
With Microsoft it’s not about having the best platform, it’s about making money. They will only take the course of action which leads to the most money. It might not be in Microsofts interest to have quality software.
Agreed. They could care less what people really want as long as they keep upgrading.
Why should they re-write windows ? Even if it can improve windows by leaps and bounds they still won’t do a total re-write if that means that they are putting their illegal monopoly at risk. As others have said MS is all about the money ! Why do a re-write when they already have a full iron gripe of a monopoly in the software and OS market in the x86 platform ? This just does not make nickels and sense IMHO.
Microsoft would lose a lot of money if they provided too much quality, because quality has an association with control. Since when you purchase the Microsoft platform you are only renting the platform, the only person who owns the platform is Microsoft, it is in the interst of the controller to hand over software that is medium quality and dependant on upgrades that can only be done by the owner (Microsoft).
I think that you don’t understand these things because Open Source does not take advantage of it’s strengths yet (documenting the kernel conceptually).
…relinquishing power.
Is that you, topspeed?
I don’t think the world needs yet another Unix. If we’re always using the same old stuff, we’re never going to get anywhere. Just demanding Unix over and over again is very stubborn, where’s the room for new ideas?
That may be true right now…but MS wants to weave IE into the kernel for future releases. That seems like a really bad idea considering how many security and stability problems are in IE now. Weaving IE into the kernel will likely break a lot of stability and introduce serious (not that there aren’t now) security issues to windows.
That dosen’t make sense. There is nothing they would gain by integrating IE into the kernel. Its integrated into the OS already and they have the control they need.
The idea is to keep applications seperate from the kernel and beyond device drivers windows does this well.
The only application I can think of directly tied to the kernel is IIS6 in Win2003 server. It has a kernel level driver that tweaks Windows to dish massive amounts of HTTP requests. It seems to work quite well at this point.
“The only application I can think of directly tied to the kernel is IIS6 in Win2003 server. It has a kernel level driver that tweaks Windows to dish massive amounts of HTTP requests. It seems to work quite well at this point. ”
Web servers do not belong in the kernel! It blew my mind when one was included in Linux, though this one in Win2k3 is slightly less of a shock.
It’s not a simple job, but I am sure that there are so many learned lesson by made the mistakes.. The rewrite should give the much more cleaner API and many better stuff without have to poke and clean the millions line of code.
I don’t think the world needs yet another Unix. If we’re always using the same old stuff, we’re never going to get anywhere. Just demanding Unix over and over again is very stubborn, where’s the room for new ideas?
stew (IP: —.arcor-ip.net) – Posted on 2003-08-04 07:03:07
stew (IP: —.arcor-ip.net) – Posted on 2003-08-04 08:12:40
“I don’t think the world needs yet another Unix. If we’re always using the same old stuff, we’re never going to get anywhere. Just demanding Unix over and over again is very stubborn, where’s the room for new ideas?”
I don’t think the world needs yet another stewie post. If he’s always using the same old stuff, we’re never going to get anywhere. Just demanding over and over again is very stubborn, where’s the room for new ideas?
;p
>Windows NT is basically a re-implementation of VMS. Many, many things are similar even today. David Cutler worked on both. So, the overall design does go back to the 70s, though they did move the interface to kernelland, etc.
Nope… That’s just what the pro Microsoft people want you to think. And wouldn’t you say that winnetmag.com is pro Microsoft??
NT is based on OS/2. When NT 3.1 came out it was based directly upon what Microsoft was working on for OS/2 3. But since Microsoft and IBM broke off their arrangement to work together on OS/2, Microsoft released their work on the next version of OS/2 as NT 3.1
NT *is* OS/2 in fact, you can still run the really old 16bit OS/2 apps on XP.
“NT *is* OS/2 in fact, you can still run the really old 16bit OS/2 apps on XP.
”
While you are absolutely correct on NT’s OS/2 roots, the reason one can run OS/2 programs on WinXP (and Win2k, etc.) is the fact that it has an OS/2 subsystem as opposed to the fact that it (WinXP) is directly descended from OS/2.
Thinking of it the way you put it is like saying that WinXP can run Win95 apps because it is directly derived from Win95, and that’s just not true. It can run Win95 apps because of the Win32 subsytem that provides WinXP’s primary set of APIs.
>NT *is* OS/2 in fact, you can still run the really old 16bit OS/2 apps on XP.
And you can run win 3.x apps on XP as well … just like I always suspected … there must be DOS 8.1 or something under the hood
In the real world just as most people are not landlords, but instead are renters or pay a long term mortage, the same holds true in information technology. There are far more people that are willing to give up control than there are those willing to take control and the associated responsibility. Businesses are pressured into using information technology because of shareholder interest. A public corporation is required to meet certain standards and it is pressured to spend on IT, even if the IT investment does not add long term value. A large orgainization can acquire tax relieve for such spending so the people who pay for half of it are unknowing individual taxpayers. Microsoft makes money for America. That is the only thing that matters because money curculates in the economy, in othe words, you are using Microsofts money if you take a loan from your bank. The Americans make money when that money comes from outside of the country, that is their priority.
The fact is that if you can not document the details of all of the changes that you want to happen in the source code, the new architecture, the new interface, and provide the reasons why. Unless you know exactly what you are talking about than how do you know that you are recieving value? You just want changes because that’s what you desire, you would rather watch a new movie than watch a movie that you already saw even if you know it is a good movie. It’s human nature…but you will not know about quality software I tell you, until you can identify the problems, you need to be able to conceptualize the system from a high level view, than be able to have the control, so that you can use lower level views of the system and identify the modules that are responsible for that requirement. If you don’t have any control than you don’t know what you want, just something…anything to change.
” If MS had 49 trillion dollars, they still couldn’t write an OS that was quality work. They are complete money grubbing pigs and they will never, and i mean never, surrender that urge to make room for quality. Microsoft is the #1 scumbag in the world. They supply shit software to total morons, and they’re proud of it. Shills of stupidity are not interested in quality. ”
In all honesty the only moron here is yourself, there are numerous people who have managed to post quality comments on this article but there always has to be the moronic Linux thrall who simply repeats the same old line. If your going to make swepping alegations provide some form of proof, personally I’m quite a fan of MSN messenger, MS Word, Visual Studio and .NET, why? because they are all usefull quality peices of software which make my life easier and I’m yet to see an OSS equivlant which is anywere near as good.
But back to the discussion (and to repeat a well repeated line). MS doesn’t need to rewrite windows, NT was the rewrite, .NET adds a layer of simplicity and compatibility to different versions of windows running on differen’t architecutres (an essential feature if they want to succesfully move to 64bit proccesors) ontop of the Win32 API (which is horribly ugly), MS is already redoing there GUI as well in Longhorn with support for hardware proper hardware acceleration, as well as that MS is in fact putting into Longhorn what looks to be an intresting (and ‘kinda’ novel) feature, WinFS which if it succeds will make everyday workflow much simpler and quicker. So after all that why would MS wan’t to redo the kernel and aim in a radical new direction.
BTW. For all those who hate the idea of having part of a Webserver integrated into the kernel (Ala, Linux/Windows
) , I’d like to ask why? after all the system is only going to be used for serving up data not for any other tasks so whats wrong with making it faster? and if you read up on it carefully you’ll find MS did EXTENSIVE quality control on the TINY code module they integrated into the Kernel so I’m thinking it should definetly be safe.
BTW. Why would MS want to implement windows ontop of Unix??? Or is this just crap from the linux is the best bunch, Let me enlighten you the windows code base is bigger and in many ways more mature than the Unix/*BSD/Linux code base, it has many features that Unix lacks including such things as asynchronous I/O, a modular kernel framework which alows simple loading of processes into the kernel rather than the Linux HACK known as LKM, and many things integrated which aren’t integrated on Unix I.e. GUI.
Lastly what MS should do is provide a more logical division between Home and Pro editions of Windows currently all I can see Pro as is a more expensive version of Windows with a few extras slapped on. I say give the pro users what they really want proper configurability and less dumbed down wizards instead of the patronisingly dumbed down ones windows currently has and let the home users have everything really simple, its a logical system if you ask me.
Nope… That’s just what the pro Microsoft people want you to think. And wouldn’t you say that winnetmag.com is pro Microsoft??
NT is based on OS/2. When NT 3.1 came out it was based directly upon what Microsoft was working on for OS/2 3. But since Microsoft and IBM broke off their arrangement to work together on OS/2, Microsoft released their work on the next version of OS/2 as NT 3.1
No Windows NT has an OS/2 subsystem and API but it is not in fact OS/2 itself.
Why would Digital sue MS if NT was OS/2 ? Unless you want to believe that OS/2 itself was based on the original design of VMS as done by Cutler.
NT *is* OS/2 in fact, you can still run the really old 16bit OS/2 apps on XP.
Just as you can run some POSIX stuff but that dosen’t make NT POSIX fully.
I’d suggest you be clear on whether you mean Linux or Unix. There are many Unix systems that have had the features you cited for years before WinNT was around. Solaris for instance.
I have a feeling what you are going to be seeing from Microsoft in the future are any and all attempts to create relationships with hardware vendors so that these vendors provide drivers only for Microsoft and absolutely nobody else. You will see Microsoft also moving horizontally into various hardware.
Linux. Open Source developers need to focus on a desktop oriented OS or they will never succeed.
I hope that windows continues to dominant the desktop market well into the year 10000. i hope it crushes, destroys, obliterates linux, into tiny little bits, so i can feed them to my dog.
That may be true right now…but MS wants to weave IE into the kernel for future releases. That seems like a really bad idea considering how many security and stability problems are in IE now. Weaving IE into the kernel will likely break a lot of stability and introduce serious (not that there aren’t now) security issues to windows.
That dosen’t make sense. There is nothing they would gain by integrating IE into the kernel. Its integrated into the OS already and they have the control they need.
The idea is to keep applications seperate from the kernel and beyond device drivers windows does this well.
That is true. It does not make sense for Microsoft to weave IE directly into the kernel…but that IS what they have announced their plans are. Further, IE will no longer be available as a stand alone application. We can not expect to see IE 7 available for download…ever. The only way you can recieve new versions of IE is by upgrading Windows.
Like it or not…whether it makes sense or not…it is what Microsoft has announced they are doing. I think it is a really dumb move myself. I guess they think they can magically solve the security weaknesses of IE by weaving it into the kernel. I don’t think it will really work though. I think one of two things will happen.
1. They will utterly fail at weaving IE into the kernel (why the hell you want a web browser in the kernel seems baffling to me).
2. They will succeed in weaving IE into the kernel but produce a vast array of security and stability problems in Windows itself.
Either way…that is what MS has said they are doing. I agree with you though…it doesn’t make sense to me. Apps should be totally seperated from the kernel…but MS doesn’t seem to think so.
as well as that MS is in fact putting into Longhorn what looks to be an intresting (and ‘kinda’ novel) feature, WinFS which if it succeds will make everyday workflow much simpler and quicker
Actually WinFS is not novel. I am glad they are doing it…but it isn’t novel. This was already done with BeOS. Granted, the implementation of WinFS will be nicer, of course a lot of time has passed since Be did this.
Further, Mac OS X 10.3 “Panther” will be able to perform the same search capabilities as are expected with the WinFS/Yukon based search tools…and it will be out long before Longhorn. If you are getting ready to disagree with me about this, don’t bother. The developers preview release that we have at work has already demonstracted the abilities.
It may be ‘kinda’ novel for MS…but that is just because MS is often behind the curve on developing things that are innovative and useful for the masses. I hate to say that since I am running XP right now…but everything I have seen that was hailed as a “new” feature for Windows I have seen implemented elsewhere first and usually better.
The believe that the new filesystem design originated with data models of Database Management Systems (DBMS’s).
I don’t know if you’ll ever get to read this comment, but OS News is rapidly turning into Slashdot.
It does not make sense for Microsoft to weave IE directly into the kernel…but that IS what they have announced their plans are
Link please.
Not really the point of the article, but it seems unlikely the new troops will have anything to do with writing Windows software. Longhorn is already well underway.
It’s more likely that Microsoft is hiring for Xbox 2, expanding it’s army of lawyers, or creating a soft-drink division.
MS will be trying to close the doors on the competition, primarily through incompatibilities, but also lots of corrupt deal making. That’s what the research is all about. Find ways to stifle the competition so that they can maintain control.
Just a word to the otherwise – Jebus, I’d suggest you rethink the grounds for your claim for Microsoft and Windows.
Size does matter in software development, as much as size matters in, for example, writing fiction. If you use two or more words to make the impression that one word could’ve made just as well, Strunk & White’ll haul you up by the toenails and make you rewrite – and rewrite, and rewrite, and rewrite, and rewrite … until you get it right.
In fact, one of the complaints many Unix users had about System V Release 4 was that it was bloated. Surprise, surprise, that is also one of the most common complaints made against Microsoft Windows.
As far as Windows being a more mature code base than Unix, I somehow doubt that. If you take the position that Windows NT (3.1 -> 5.2 +) is a direct descendant of VMS, you might have a point, or the beginnings of one – but Unix was developed on the DEC PDP 7 and PDP11, and then ported to the 32bit DEC VAX. VMS was developed by DEC for the VAX. Guess which machine DEC made first? (Hint: it wasn’t 32bit.)
As far as Windows being better than Unix because of all the stuff it incorporates into its kernel, allow me to dissent – again. I think the problem with Windows over the last few years has not been the number of stuff it has incorporated, as the manner in which it was done. It’s only been recently that Microssft has had the guts to acknowledge that what everyone has been saying for years is true – that MS’s quality control is abysmal and its security is obscenely and hilariously minimal.
I think MS designed .NET carefully to offer a good migration path.
I’m not sure about the kernel. But all the stuff at the higher levels is likely to be C++, and will be recompiled to unmanaged byte code.
When this works, code will be moved piece by piece to managed code. This will expose or eliminate buffer overflows and similar threats to stability and security.
After that, the code will be redesigned and simplified by making use of the advantages of the .NET platform (Goodbye COM and MFC). The new code will be written in C#.
At least this approach seems to make most sense to me. The use of managed code will lead to a more stable system. It will be more easy to support additional hardware platforms. The APIs will get more elegant. And all this can happen in a smooth transition over several releases.
So I think that this is what M$ is aiming for.
I think they are aiming also for more control on binaries than they have now.
1) Before .net, Microsoft had 100% control over the Windows implementation and the user had 0% control, and after .net, Microsoft still had 100% control over the implementation. So we know one thing for certain about the future.
2) They will be moving toward developing tools, so that the users are more dependant on Microsoft and it will be easier for Microsoft to lock in developers who are dependant on tools rather than programming skills. If you program for Microsoft it will be through specialization (solutions) rather than generalization (systems).
3) Vendors will be able to program constraints on data through meta data that is stored centrally in a database catalog/Dictionary.
well at least that Mono guy , see its coming…
Turn .NET into a portable distributed computing layer, ensure that it really is portable (e.g. java like), ensure that it also can run multiple languages (e.g. whereas java is one language, .NET can handle C++, C#, F# and so on), also make .NET work with varying different substrates (e.g. classic von neumann architecture, rna/biological computing), fully develop the operating system and the applications (e.g. office) so they are distributed and can connect and use external resources (e.g. at computing centres), ensure that part of .NET is OSS’d, so the whole community has confidence in Microsoft, but don’t OSS it all so as to lose competition.
From what I see inside Windows, it is “sort of” migrating to the right direction, but as anyone who works with legacy systems will realise, evolution is hard, and even harder if you have millions of customers and millions of varied sorts of software and applications.
Some posters here say that the tools are better on Micorosoft. It’s a more productive development environment, etc. It’s only a more productive environment if you are speeding to be locked into a situation where you have no control. Where you will not be able to acquire knowledge about how to solve problems on your own. You will not be able to get off of the speeding train once you get on board because you will not have the right skills. Instead you’ll be constanty trying to keep up with the changing environment that is motivated by sales much more than it is movtivated by value or quality because those composites are only attainable through control.
if microsoft really wants to fix the design flaws in windows there is no way around the redesign of the core system.
http://security.tombom.co.uk/shatter.html
The Microsoft platform is not a platform, it’s much more like a product. It is only a platform if you control the source code and the packaging, so from Microsofts point of view, it is an open platform, however to everyone else in the world, it’s just a product, because there is no stability if you have zero control over the future. So when you rent the Microsoft product (platform) you use Microsoft tools or else you move on.
and you don’t build systems…only specializations (solutions).
Your wrong applications sell OS’s not features. This is one of the reasons why Linux will not catch on. Look at how people were using windows 3.11 when OS/2 was around. Clearly OS/2 was a superior OS but it did not have any developer support anywhere equal or near to windows at the time. Hell BeOS was a better OS then windows 95 in it’s time but that lack of application support was what killed it off. Like Steve Blamer once said it’s all about “DEVELOPERS, DEVELOPERS, DEVELOPERS !!” because they make the applications people want, will us and which will sell an OS for you easliy.
People are more then willing to put up with bad design, security hassles, stability issues and clunk-ness ( I.E. Windows 95-ME ) of an OS as long as it has the commercial application support to make it useful. If Linux gains wide hardware support along with substantial commercial application support then it will move up in a major way. Until then it will be a niche OS. As it is now Microsoft is in a great postion to control that flow of software development and thus has no need or is in no hurry to add “OS features” to compete with other OS’s on the x86 platform.
He is right ! Do a search on OSNews ( “Microsoft Drops the Ball with Internet Explorer” ) and you will find it yourself. It was posted here when they made the announcement that IE 6 would be that last version that would be disturbed on the net for people to install and use or even capable of removing. The next version of IE called IE 7 will and shall be intergraded heavily into the OS and will only be available through the purchasing of that OS ( AKA LongHorn ) and will not be removeable at all !
…is a funny man. He should write about the SCO issue on http://www.theonion.com
Isn’t MS going to clear out older unused API’s and also abandoned drivers for older hardware, because this new OS is going to run on only new hardware. Isn’t that enough of a gamble for MS.
Anyway reading through the comments pages most people have already mentioned the issues about bloat etc.
Anyway win2k runs fast on a 450Mhz PC, and linux with GNOME or KDE or any other WM does not. Flame away boys and girls.
On Linux, the user has to have more responsibility to understand how the platform works, especially how to configure and install/make, as well as more time and education. That does not attract the masses of users.
The way for Linux to gain popularity is for it to become a product like Microsoft, rather than a platform.
On the other hand, the way for Linux to become sucessful is for Linux to develop it’s strengths as a research platform, and that is by making access and learning easier for beginner developers. Organizations like Red Hat might do very well if they became a knowedge base for system implementation services. If Red Hat offered visual tools and documentation that focused on the strength of open source, the fact that the user has control and can generalize (build systems). Beginners should be able to see the kernel as high level views, and than transend to lower level views. The documentation should be relative to the view. DOCUMENT, DOCUMENT, DOCUMENT.
He is right ! Do a search on OSNews ( “Microsoft Drops the Ball with Internet Explorer” ) and you will find it yourself. It was posted here when they made the announcement that IE 6 would be that last version that would be disturbed on the net for people to install and use or even capable of removing. The next version of IE called IE 7 will and shall be intergraded heavily into the OS and will only be available through the purchasing of that OS ( AKA LongHorn ) and will not be removeable at all !
Just because they won’t offer it in a standalone version dosen’t mean it’ll be integrated into the kernel of NT. There are tons of integrated apps that ship with windows that aren’t part of the kernel.
you don’t honestly think that Notepad has kernel level access do you ?
“Conclusion : Windows is the best option for PC. & I am pretty much willing to spend 200$ on it.”
So you are a s****d
When NT was only an infant is got quickly known as “Nice Try”. That had nothing to do with flaws in the design of the kernel but everything to do with bad code and design decisions in userland stuff such as IIS etc.
Fact: The NT kernel is neither a monolithic kernel nor a microkernel but rather a modified microkernel because they built in the VM Manager and the process sceduler into the kernel as they saw big performance benefits. Same reason they later included some graphics stuff and most lately a http handler into the kernel of 2k3 server. You can be absolutely sure that this code is not causing problems.
Now the kernel IS NOT based on OS/2 (people are just assuming “guilty by assissiation” here).
Fact: The NT kernel is heavily based upon VMS. It has an OS/2 subsystem allowing it to run software that calls OS/2 APIs. Nota bene: only the 16bit OS/2 subsystem was implemented making NT unable to run 32bit OS/2 software (OS/2 Warp and onwards).
Now, all hardware is accessed through a HAL so not even the kernel can access hardware direcly and this makes it much easier for developers to write drivers as they write their drivers to interface with the HAL API. When the kernel (or any other software for that part) accesses hardware it calls the HAL which in turns calls the driver. This is just as a microkernel. New design needed here? Don’t think so. Code check? Maybe.
Should the Windows userland be cleaned up and in some parts rewritten. H**l yeah! As was pointed out in a previous commant why are old Win 3.1 commands still included?
So in essence I’d say that as usual the ZDNET people have no clue, again, what they are talking about when it comes to Windows. Forget that article quickly and move on.
Here are a few suggestions to MS: make the Win 3.1, 95, and 98/ME software run in Virtual Machines instead of as API subsystems and remove all that cruft from Windows system folders. This can only make compatibility better (especially since connectix with their VM software was bought). Make those VMs installable so that users can choose whether they want then installed or not (I don’t run that old legacy stuff any more so I have absolutely no use for it).
Rewrite components in 32/64bit safe C++/C# (not C# for .NET but compiled against a newly created set of standard libs for C#).
Make Windows more modular and configurable. Drop the stupid and insulting “My Shit” naming scheme. If I don’t know that it’s MY shit then I don’t deserve to use any tools above the technology level of a hammer (which btw is just a redesigned club (refr: 2001)). Why can’t you use useful names on files. Windows have had support for long filenames since 93 (release of NT -93, 95 in the Win9x series) so why still use 8.3 naming conventions? ntoskrnlexe?! At least name it NTOSKernel.exe (OTOH, the name Kernel.exe is more than sufficient). Use descriptive names. Why not call taskmgr.exe TaskManager.exe cos that’s what it is.
Make a Windows license for hone users with more than 1 computer (I have 4 at home incl. the laptop). Make it possible for people to be legal (I am as I run FreeBSD and linux on 3 of mine). Drop the Home/Pro divergence. The problem is not Home users but rather that corporate customers are starting to look elsewhere (OSS Desktops etc).
I have lot’s of more good, IMVHO, ideas for changes but that should do for a while.
They (MS) only said that further changes to IE would require changes to the underlying system (IMHO bullshit!) making it impossible to support a standalone version of IE.
They did not say ANYTHING about integrating the whole or even parts of IE into the KERNEL!!!
http://www.osnews.com/story.php?news_id=3684
I’m going to write a book detailing all of the clean up that Microsoft did and I’ll tell you how hard they worked and how such and such change increased performance. Of course none of this will be true, but you won’t know any difference. And there is no way to prove that it isn’t true, but it doesn’t matter, because it sounds good and we can all sleep because we are so smart.
“Buy Oracle and put Larry out to sea. “
<sarcasm>Yeah, that would be great for everyone.</sarcasm>
-bogey
Great idea. Condemn 80% of the computer using public to another ten years of beta testing, and break most of the software on this planet.
If you use windows, you are stuck with it. It’s OK, but it’s not going to get much better for a long time.
1) IE is going kernel. It has to be moved to the kernel inorder to support MSFT’s DRM concepts; this means that IE will have to be secured and able to run in the secure memory partition. It will also have to be protected from other applications; DRM requires that all the images and sounds playing on the web page can’t be grabed by some other application.
2) MSFT is already rewritting a lot of Windows. .NET is the future for all MSFT software. Win2K3 will only run about 80% of the current software written for windows. Longhorn will run even less. Why? Because MSFT has started removing old APIs and fixing long time security holes that a lot of current software depend on.
Windows is too complicated and I agree that Longhorn will only make it more complicated. By starting over from scratch, Microsoft could design an OS around the needs of today’s hardware, applications and users instead of retrofitting Windows with new technologies.
Though, I doubt Microsoft will do this because businesses desperately need backwards compatibility. There are many old Windows applications that companies need to run that they don’t have the source for and the companies behind those applications may not even be in business anymore. So, Windows will go down the road of backward compatibility/bloat/retrofitting because its what Microsoft’s customers require.
Actually David Coursey does own a Mac and a PC that’s why he knows Windows needs to be rewritten.
And release a real operating system.
MS should rewrite the OS from scratch. They could ensure that they don’t make any of the mistakes they made with Win9x and WinNT. The new OS would be simple, elegant, and ultra-responsive, even on older equipment. They could call it…BeOS! (Oh wait, someone has already done that). 😉
What I think the world needs now is to have a single API for all OSes and no special features for anyone. This way, they’ll compete for the best and fastest internal plumbing and as a result- an efficient OS. This way all OSes may have to re-write something for conformance.
Make is clean, make it simple and make it powerful and alow the user to get their hands dirty if they want to but hide the complex settings away from Joe/Jane Average.
It seems Apple beat Microsoft to this goal. As far as clean, simple, and powerful goes, the (Free|Net|Open)BSD people got there first, too. And, of course, for people who also want the kitchen sink, there’s always a Linux distribution a download away.
With that extra money lying around it should consider investing in security. I think with XP it has taken it to the extrme of user friendlyness, to a point *I* can’t stand. Doesn’t mean other people don’t enjoy it, but with that extra cash it should consider securing its os. Hire more coders to attack these problems. There is also nothing wrong with re-rewriting or starting from scratch but then people will bitch about how there is : XP,2k,2k3,and whatever else it pulls out of its ass. Next issue people will argue would be, which OS do I want to choose today? Revamp security, ability to hog less ram, be more resourceful would be nice.
If any of the people who say that Windows is not easy to use are also those that proclaim Linux as the new OS for the masses, they are insane…not dumb, just insane. If anyone who uses linux can say that it is easier to use than windows, then their hate for Microsoft and extreme attachment to linux has blinded them to the fact that windows is actually easier to use than linux.
So everyone is saying “ohh windows is hard to use” and “oh windows is too bloated”….so what OS is easier to use? OS X would be my first consideration, but like many others, I don’t have $2000 to dish out for a computer. I do however agree that the whole start menu taskbar thing needs to be redone….
No, it’s just a revision of OpenStep. I can’t think of any really revolutionary UI or fundamental architectural designs that have come about in a long time.
http://www.apple.com/macosx/jaguar/quartzextreme.html
What I think the world needs now is to have a single API for all OSes and no special features for anyone. This way, they’ll compete for the best and fastest internal plumbing and as a result- an efficient OS. This way all OSes may have to re-write something for conformance.
I am praying to the gods of the computing right now that this was a troll or a joke.
Microsoft only needs these few changes to be taken seriously in the server room.
1: No registry
2: Separation of user, application, and system configuration.
3: No graphics in the kernel.
4: No putting everything in the c: drive( virtual memory, logs, home directory, program files)
The best idea would be to streamline the code. Take advantage of the money and market dominance to make Windows the most compact and fastest OS ever. If it means hand coding in assembly, Microsoft is the only company that has enough resources to do it. Imagine if Windows XP/2003 was only 10 or 20 megs. No competitor would ever be able to match this feat. Otherwise it is only a matter of time before OSs such as Linux catch to Windows in all aspects.
Gill Bates, the editor already owns a Mac. Personally, I don’t understand why the editor keeps harping on about Windows as its future doesn’t affect me or infact any Mac user in the slightest.
As for the re-write, this is what I would do. Rip out and start again:
FreeBSD 5.2 as the core, however, use *POSITIVE* aspects of NT, such as NTFS, which can work with FreeBSD as NTFS is a fully POSIX compliant filesystem. On top of that port the full .net framework over form Windows and have that as the default API, then once that is done, build up from there.
However, with that being said, I don’t see anything radical happening. Why? because there isn’t a great need for it.
MacOS *NEEDED* and overhaul, however, Windows NT need some parts overhauled, but not the whole lot.
I thought Microsoft was rewriting the os to prepare longhorn for TCPA (or whatever these guys call themselfs today).
I remember reading somewhere that they are desinging a new kernel and so the system services for longhorn, but I’m not quite sure. Does anyone has more information on that?
Beside that longhorn will get a new graphics subsystem similar to fresco (feature wise) and . (Hopefully this will get kde and gnome to adapt fresco to keep up with ms.)
Registry is the worst part of windows. The idea seemed good at the beginning, but… Almost all the technical points I hate in windows come from registry : having to resintall it every three months, slower and slower because of the growing size of the registry, weird shot happening with no solution, etc…
It is strongly linked to the lack of transparancy of windows : if something is going wrong, there are almost no ways to correct it, etc… It would be good to have two modes, a bit like in Mac OS X : the joe user mode ( everything is GUI, etc.. ), and the guy who knows a bit about what he is doing ( for server, etc : command line, a real shell, and a way to understand wht is going on in windows ), etc…
But I don’t see that happening, because of the microsoft policy. It would be like having modules to handle ext, XFS ( I don’t know any details of the kernel structure of NT, but I am sure it is technically possible ) file systems, etc… It is against MST’s policy.
>>If MS had 49 trillion dollars, they still couldn’t write an OS that was
>>quality work. They are complete money grubbing pigs and they will never,
>>and i mean never, surrender that urge to make room for quality. Microsoft
>>is the #1 scumbag in the world. They supply shit software to total morons,
>>and they’re proud of it. Shills of stupidity are not interested in
>>quality.
>
>In all honesty the only moron here is yourself, there are numerous people
>who have managed to post quality comments on this article but there always
>has to be the moronic Linux thrall who simply repeats the same old line.
>If your going to make swepping alegations provide some form of proof,
>personally I’m quite a fan of MSN messenger, MS Word, Visual Studio and
>.NET, why? because they are all usefull quality peices of software which
>make my life easier and I’m yet to see an OSS equivlant which is anywere
>near as good.
messenger, word, vs, and dot not? what? those are examples of quality? are you high?
Gotta love messenger for becoming a spam delivery tool. MS word is the only editor that corrupts it’s own documents and then blows up when you try opening them again. VS is pure shit: lack of edit-and-continue debugging, half-baked server explorer, the IDE doesn’t expose all options, etc. I
usually need to close down VS.NET and re-open it at least once per day because of some weird thing it is doing. Sometimes the problem doesn’t go away and I eventually learned that deleting the VSWebCache fixes the problem. I even went so far as to create a batch file to do this automatically and I added it to the External Tools menu in the IDE.
The alternatives? Chat clients and word processor apps are for LUsers. The only use for a word processor is to format a resume. And even abiword will do that. Programming editors are the way to go. The features of programming editors put any word processor to shame. Kylix is much better than VS; borland has always made better compilers/IDE’s than MS, and it’s a cross development platform.
You are obviously a lightweight user who makes no real demands on the apps. It doesn’t bother you when an app or the whole OS freezes because you’re not doing anything of value!!! You’re just sitting there yanking your pud as
you type more gibberish into messenger and word. Whoa, another blue screen, time for a cig/fag … is your attitude. Luser.
BeOS was, and is, a far more user friendly OS than Windows. Perhaps MS should use their huge amounts of cash to buy the code from Palm and make that the basis of a new “BeWindows”
Visual C++ has edit and continue debugging