Fresh Air For Windows?

jmcbain writes "The NY Times has an opinion piece on how the next Windows could be designed (even through Microsoft has already laid plans for Windows 7). The author suggests 'A monolithic operating system like Windows perpetuates an obsolete design. We don't need to load up our machines with bloated layers we won't use.' He also brings up the example of Apple breaking ties with its legacy OS when OS X was built. Can Windows move forward with a completely new, fast, and secure OS and still keep legacy application support?"

Comments

heh, normal version

[javy_tahu]

http://www.nytimes.com/2008/06/29/technology/29digi.html?_r=1&adxnnl=1&oref=slogin&ref=technology&pagewanted=print

Short answer: no

[Anonymous Coward]

Can Windows move forward with a completely new, fast, and secure OS and still keep legacy application support?

Based on past performance: No.

This has been another edition of Short Answers to Stupid Quesitons.

Re:Short answer: no

[Bill, Shooter of Bul]

The wise answer is "maybe". There are only two companies that have done something similar. Apple, tried doing it from scratch and basically killed itself in the process, had to adapt already written NeXT. Even that took forever and sucked for a couple of years before they got everything right. Microsoft did something similar with windows NT: a ground up modern rewrite that was mostly compatible with the existing windows, but there was a lot of time that passed between win NT 3.50 and win xp. So if they started right now from scratch, maybe in ten years they could have something that would be decent.

Re:Short answer: no

[bignetbuy]

Commercial versions of VMware allow multiple snapshots. The version you refer to is the freeware version.

Re:Short answer: no

[siddesu]

Yeah, indeed ;)

Or they could, like, ditch all their work done so far, fork wine and make the new OS run on top of linux+wine, possibly off a sqlite-based WinFS ;)

Then just port their platform libraries onto that, redo their visual tools as eclipse plugins -- and presto, you have best of both worlds.

And fast ;)

Re:Short answer: no

[v1]

Apple's policy is to provide approximately 100% transparent support ONE version back. They did an incredible job with classic (supporting OS 9 in OS X) and an even better job in the transition with rosetta. (supporting ppc on intel)

While it was fairly obvious you were running an OS 9 app in classic, almost no one notices a rosetta app running on an intel. Now notice, intels do NOT support classic. That's their "one hop" rule at work. And you can bet their next big one will drop support for powerpc.

So this can be done, but it's hard to get right. But when you get it right, nobody notices. And that's a good thing.

This is a bit like Windows. The problem they've had is that there's a lot more transition from dos to 95 to 98 to 2000 to xp to vista. None of those was entirely pleasant, and none of them were very transparent. Only half of them provided major new features, but all of them clung to numerous existing problems. So in the same timeframe, Apple has made just two massive leaps, with less "transition shock" in their two bumps that windows has seen in their five. The interim transitions (os 8 to os 9, 10.1 all the way to 10.5 really) were almost completely transparent.

They've got a lesson to learn here. XP probably would have been a good time to do a "major bump" such as mac did with 9 to X, but they dropped the ball. They chose to break less, but to fix less as a consequence. Eventually they have to bite the bullet and fix as many of the underlying design problems as they possibly can in one fell swoop. It's going to break stuff. Maybe a lot of stuff. But if they could provide something like Apple did with classic support for OS 9, it wouldn't be so bad. Apple proved that it's not necessary to just totally break all your old software if you can provide decent emulated support for your previous OS inside the new one, invisibly.

Sadly I don't see this happening with Windows anytime soon. Microsoft has never had a knack for making those internal transparent emulators like classic and rosetta. Unless they can get something like this together, it's either going to continue to be a wreck, or it's going to be a disastrous pill to swallow. Continuing to try to make these "baby step" fixes is going to drive the world crazy.

Re:Short answer: no

[lilmunkysguy]

I'm not sure how you were modded flamebite. I like your ideas.

Re:Short answer: no

[Zaiff Urgulbunger]

The installation size probably isn't an issue given that the target customer, corporates who have invested heavily in Win2K/XP, will be largely using high end hardware (as opposed to the "new" low-end hardware a-la Asus EEE).

Memory requirements might matter; but since we're looking at release two years from now, then 2GB is a reasonable requirement. If they base the "compatibility" code on XP rather than Vista, then it might be viable.

The biggest problem I see is what to tell people right now. Saying, "oh yeah, the next version of Windows will be completely different" is not likely to go down well, and is unlikely to encourage anyone to "upgrade" to Vista prior to Windows 7. But saying "Windows 7 will be based on Vista" isn't particularly inspiring either!

The marketing solution will likely be to not really give any concrete answers for as long as possible whilst telling people Windows 7 will build on their existing investment. If they don't do this, people might start looking elsewhere!!

Re:Short answer: no

[bcrowell]

The wise answer is "maybe". There are only two companies that have done something similar. Apple, tried doing it from scratch and basically killed itself in the process, had to adapt already written NeXT. Even that took forever and sucked for a couple of years before they got everything right. Microsoft did some thing similar with windows NT: a ground up modern rewrite that was mostly compatible with the existing windows

It may be dangerous to reason by historical analogy, because the hardware situation is qualitatively different now. CPUs are no longer showing the kind of Moore's-law growth in power that they used to. Meanwhile ram and hard disks are ridiculously cheap. For the typical user who just uses a computer for websurfing, email, and word-processing, it's kind of silly to spend any significant amount of money on a new system. They already have more ram and disk space than they need, and the CPU isn't going to be that much faster. We're seeing perfectly reasonable desktop hardware now for $200, and it won't be long until you can get that same hardware for $50.

If I was one of the people at the helm of Microsoft, I'd be really worried about this, because when the hardware is $50, there's not going to be much room left for profit on the OS. Most retailers have been reluctant to sell cheap hardware, because their own margins on it are thin, but it's just a matter of time until that changes. Fry's sold $200 Great-Quality-brand machines for years, and WalMart is now selling the gPC online for $200. Once people realize that they can get a computer for $100, or $50, the dam is going to have to break, and retailers are no longer going to be able to sell machines at prices of $500 or $1000. It's going to be like the transition from the radio as a big wooden box to the transistor radio that you could carry with you to the beach, and throw in a dumpster if it got sand and water in it.

In this new landscape, there's very little reason for MS to exist. One of the few reasons left for them to exist is that people have money invested in software, and they don't want to have to buy new software. The insane success of the eeePC -- and even at much higher prices than they originally thought they could get --- shows how vulnerable MS is. There are a lot of users out there who just use their computers for word-processing, email, and websurfing. Maybe first they buy a $50 Linux box for their kid to use to write her high school papers. That works out okay, and pretty soon the kid is like, "Mom, are you crazy? You're talking about spending $400 for a new computer? Just buy one like mine."

Re:Short answer: no

[peragrin]

It is already beginning. I submit the EEpc, OLPC, and the sudden burst of real computers with real OS''s being shipped for under $400 right now Windows is holding back more development than anything else, especially with the intel atom processor. sorry you can't get a $100 OS onto a $400 device.

why do you think msft is still selling XP for only low powered devices that Vista couldn't run on if it went on a diet. Why do you think MSFT is intentionally trying to limit the specs of such devices when they are already as powerful as any computer of 6 years ago?

Re:Short answer: no

[smallfries]

Erm, when did CPUs stop showing exponential growth in performance? Was that a memo that nobody sent to Intel?

Although clockspeeds are stuck because it is no longer economical to raise them, performance and transistor density are still scaling at the same rate. If anything we are in a period of performance increases that is slightly above trend, because now that the horrific NetBurst ISA has been killed off the Core2 replacement is rather lovely. Clock-for-clock it runs twice as fast as the old ISA because of shorter pipeline stages that have reduced instruction latency, and so far Intel have doubled the number of cores every 18 months. Given that they are ready to scale up to new fabs that can handle 2B transistors I would assume that they can continue to do so for the near future.

It would be a seismic shift for the industry if processor performance flatlined but I don't see that happening for a long time. What we are seeing with the introduction of the Eee Pc et al is actually a trend that has been going on for decades. Roughly every ten years a new form factor is introduced at the bottom of the market, with the same performance, but with the price halving each time.

So although your analysis of what changes are happening is way off, your final paragraph is quite accurate about what it means. The amount of performance that people actually require for most day-to-day tasks was exceeded when processors passed the Ghz mark. Now we are seeing cheaper and cheaper devices that deliver that (roughly) constant power. The effect on Microsoft is likely to be as you predict.

Re:Short answer: no

[Tangent128]

I don't know about the "Eclipse plugin" part, but I do suspect Microsoft has a contingency plan of that sort.

Move to a nix-y kernel, release a full .NET port; maybe fork wine, or just use some more dog-foody compatability layer.

I suspect they'd introduce/keep their own API, though. I wouldn't expect X Windows to be bundled with (let's say) "Windows X"; they likely would use the transition to more strongly push Windows Forms over the older system, though.

And of course, don't expect their addons to be Open Source, even if they do adopt the Linux kernel.

In short, see OS/X.

Re:Short answer: no

[Anpheus]

Eclipse? Fast?

Re:Short answer: no

[postbigbang]

I worked with the code from OS/2 and from the original WinNT SDKs (55 floppies of it). Sorry, but conceptually, Cutler had little choice but to take the OS/2 APIs and turn them into Microsoft analogs. I have the code; Cutler had marching orders to one-up IBM and he did it. No argument except citing anything from the WSJ as a technical history source.

Re:Short answer: no

[MrEricSir]

Compared to NetBeans, Eclipse is a formula one car.

But to be fair, compared to NetBeans, a dead snail is a formula one race car.

Re:Short answer: no

[cjsm]

Let's see what's wrong with Vista: 1) lack of hardware drivers, rendering many machines obsolete

...which is a problem with the hardware and/or the hardware manufacturer who decided not to support their hardware in Vista. It is NOT a problem of Vista any more than some hardware without a working Linux driver is a problem of Linux.

Why is this a problem with the hardware manufacturers? Microsoft made major changes to the driver model, partly to implement their DRM, so writing new drivers is time consuming and expensive. So Microsoft gets to make a fortune selling Vista; but the hardware manufacturers are supposed to spend a fortune writing new drivers for old equipment they sold years ago, so everyone upgrades to Vista and Microsoft can make even more money. Microsoft should pay the hardware manufacturers to write new drivers.

Re:Short answer: no

[Bert64]

Modern OS's do sandboxing already, ie running everything in its own memory space with copy-on-write shared libs, and no write access to the kernel space or other apps. It helps stability a lot compared to the older OS's with a flat memory space, but it hinders performance too. AmigaOS ran in a flat memory space, and was very fast, but one program could easily crash the whole system. On the other hand, the inherent instability forced app developers to write decent code instead of relying on the OS to bail them out.

As for wine, it's not so much further abstracted, as abstracted in a different way. Your not running windows game on top of windows on top of linux (ala vmware) as the quote suggests, your running windows game on top of wine on top of linux... Which is really no different than running the game on top of win32 on top of NTKRNL. If the code implementing wine is more efficient than the code implementing win32 on top of NT, or if the linux kernel does things more efficiently than NT, or if your drivers do their work more efficiently, then the linux/wine combination can be faster.
If you look just at the kernel level, windows has far more complexity relative to linux, all that added complexity comes at a performance price.

Time to Get Rid of The Gates Borg Icon

[Anonymous Coward]

Now that Bill Gates is retired from Microsoft, the editors should get with the times and lose that dated, painfully unfunny logo they use for Microsoft.

Most people probably wouldn't get the Borg reference to begin with, and now Bill Gates era at MS is officially in the past.

Only MS gets this ridiculous logo..now its finally the time they get rid of it.

Re:Time to Get Rid of The Gates Borg Icon

[Spy der Mann]

I vote for a chair breaking a Window :D

No, I'm serious. Get a picture from the Microsoft Headquarters, and from a building, add a chair breaking a window and falling to the floor. Cartoonize it, and you're done! :)

Re:Time to Get Rid of The Gates Borg Icon

[bigstrat2003]

If anyone doesn't get the Borg reference, they don't belong here, they should be at the geek office turning in their geek card (or educating themselves, whichever they prefer).

Re:Time to Get Rid of The Gates Borg Icon

[Renderer of Evil]

Now that Bill Gates is retired from Microsoft, the editors should get with the times and lose that dated, painfully unfunny logo they use for Microsoft.

That icon is there for legacy purposes.

Re:Time to Get Rid of The Gates Borg Icon

[techno-vampire]

Considering that Microsoft is often considered ot be the 800 lb gorilla in the software world, there's a better idea: King Kong on the Empire State Building fighting off biplanes.

Re:Wine?

[Brian Gordon]

WINE just provides a reverse-engineered implementation of the Win32 API. Microsoft has the real original code.

Not gonna work / we already have it

[orkysoft]

Apple could do that because they were much smaller than Microsoft, and had a small but relatively loyal customer base, and their rewrite did pay off, as people are generally very happy with OS X and don't care about the incompatibility with OS 9 and older anymore.

Microsoft has a huge userbase with much less loyalty, and generally a huge existing investment in software.

We don't need a MS Windows rewrite, we've already got Ubuntu, because that's essentially what the article author wants: an operating system that Just Works[tm], even at the expense of compatibility. That's a pretty good description of any popular Linux distribution.

Windows done right from the ground up

[BadAnalogyGuy]

WinCE. Pity about the name, though.

Yes, it's been done before

[LoTonah]

Windows NT had an emulation layer that handled 16-bit apps. OS X had Rosetta and the Classic environments. And Microsoft now owns Virtual PC.

They have the technology to make Windows a clean OS with emulation errors for doing whatever legacy OS you want. They just seem too lazy to do it.

Fluff piece

[ejdmoo]

He really doesn't know anything about the internals of the Windows kernel or the Mach kernel, he's just assuming that since the NT kernel is "monolithic" and the Mach kernel is a "microkernel" then the latter must be better, and the reason it's better is it is "smaller."

If you want to know where the real problems with Windows lie, they're in the API and the shell, not the kernel. The NT kernel is perfectly fine. See this Ars write-up by someone knowlegeable:
http://arstechnica.com/articles/culture/what-microsoft-could-learn-from-apple.ars

I'd like to point out that Microsoft employs one of the original authors of the Mach kernel, Rick Rashid. He runs Microsoft Research. Look it up.

Re:Fluff piece

[Tumbleweed]

I'd like to point out that Microsoft employs one of the original authors of the Mach kernel, Rick Rashid. He runs Microsoft Research. Look it up.

Being put in MS 'Research' is the kiss of death if you want to make something that MS will ship. They seem to hire those brilliant people and give them massive funding only to keep them happy and prevent them from working for a competitor who might want to actually SHIP something brilliant they would come up with. Rather like IBM, only substitute incompetence in place of amorality as motivation.

Re:Fluff piece

[bcrowell]

The Ars Technica piece is interesting, but I'm pretty skeptical about this whole idea of making radical changes in Windows and breaking backward-compatibility.

One thing you have to keep in mind is that there's a huge downside for the user when you break backward-compatibility. Apple actually did an amazing job of maintaining backward-compatibility when they made the switch from 68000 to powerpc, but when they brought out MacOS X, the backward compatibility was lousy. You could still run classic apps on X, but they typically worked very poorly -- some features wouldn't work, apps would crash, and it took a really long time to start up the classic environment. Essentially Apple expected you to buy all new applications. Then Apple kept on bringing out frequent point-upgrades to MacOS X, and every single one cost a significant amount of money. My wife bought one of the early lamp-shaped iMacs, and we stayed on the upgrade treadmill for a while, but it really got old spending money every six months or so for a new version of the OS, so at this point we're still running an old version of MacOS on that (expensive) machine. Now we basically can't run any new software, because it only works on newer versions of MacOS X.

It's also worth looking at it from MS's point of view. They're a monopoly, and their interest is in keeping users sucking at the tit. Maintaining backward compatibility has worked very well for them. One of the main things keeping Windows users from jumping ship for another OS is that they know their apps will continue to work. It's actually kind of amazing. I tech at a community college, and some of my colleagues are still using an old DOS shareware planetarium app. It still runs on Windows XP.

Sure they can!

[ThorGod]

Just switch to Mac and get parallels :P

Yeah, I know, not very funny. But does every comment have to be great?

Re:Existing legacy support. Wait, what?

[Anonymous Coward]

Any software that was created in the past few years which vista 'broke' were most likely poorly designed or were associated with managing or doing the functions expected of the OS itself (with a few exceptions.)

Vista really isn't that 'buggy.' It is top heavy and uses way too much resources if you are only using it for limited things, but as a general purpose OS it really isn't that bad. I would still prefer Windows XP on new computers simply because I can get away with more power with a smaller investment in hardware, but I'm not necessarily 'against' Vista.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Wine?

[larry bagina]

the windows NT kernel is fine. Moving to BSD or linux, or QNX etc won't improve it. OS X wasn't just a move to BSD, it was also a move to OO via Cocoa. The toolbox/Carbon is/was strictly procedural, much like the Win32 api. DotNet is OO, but so was MFC.

Challenge accepted

[Anonymous Coward]

Is that a challenge?
Static
Animated GIF

Re:oh come on

[Darkness404]

But... I can run the same thing on Linux in RAM with 512 MB (or less) of RAM and a 700 MB CD. When I install it it takes perhaps 2 GB of HD space for the exact same functionality.

-A file manager - There is Thunar in Xubuntu, Nautilus in Ubuntu and Konqueror in Kubuntu
-A web browser - Firefox
-Multiple filesystem support - Ubuntu can read/write more filesystems then Windows can
-Most extensive driver library in existence - Except for the fact that on 90% of hardware I can get Ubuntu to get everything to work out-of-the box except for proprietary drivers for ATI/nVidia cards and Ubuntu makes that easy, Windows is a pain to install without like 10 driver CDs or an OEM restore disk
-Office tools like Mail, WordPad, Calendar, Calculator, Contacts, Paint? - Thunderbird, OOo, a calendar program, a calculator program, various contacts programs and The GIMP
-Full command line environment (DOS) - Full UNIX shell (BASH) -Complete media architecture in DirectX.. that's DirectSound DirectInput -has Linux equivalents though I can't think of them off the top of my head -DirectDraw.. a LOT of big packages if it was Linux. Also Windows Media Player/Photo Gallery - Totem/Amarok for WMP
-Graphics APIs and rendering engines -Again, found on Linux
-Remote desktop - VNC/SSH
-Labyrinthe configuration utilities and applets -Don't really know what that is, a Wiki search returned nothing -Monster domain features.. detailed ACLs on every resource, complex user permissions, domain controls enforced on clients (integrated securely right into the interface). - UNIX-style permissions, secure by default


Just about everything you said is included on Linux on a *Buntu default install, or can be added without going over what Vista has installed. Sorry to say, but really Vista is just pure bloat. Lets see what is in a default * Buntu install that Windows doesn't have...

Full Office Suite - OOo
Photoshop Replacement - The GIMP
Various network services - Telnet, SSH, etc
(*real*)3-D Desktop - Compiz-Fusion
Multiple Desktops
PDF Reader
Various support for files that Windows doesn't have by default (Ogg, FLAC, etc)

As you can see, Windows just can't compete with Linux when it comes to programs per storage space. In 5 gigs of a Vista install you get just about only the default install, in 5 gigs of a Ubuntu install, you get the default install, plus some of your files, some development tools, some more games, a few more applications, etc.

Re:Existing legacy support. Wait, what?

[Toll_Free]

Having run Vista32 on this laptop when new, and just recently moved to Vista X64, I agree.

I turned most of the "eye candy" off on 32 bit, but 64 doesn't seem to get bogged down nearly as bad with the eye candy turned on. NOTHING else was changed, only the OS.

Anywho, yes, Vista is fine. Pisses me off that I can't run Win16 apps on Win64 (like, install C&C, for instance), but oh well.

I think I'll try 64 bit linux next.. Never tried a 64 bit rev... Any suggestions? I've always run Slackware since my first install, but it's not always the most "hardware friendly". It's a HP DV2000 based laptop, x64 1 gig ram.

--Toll_Free

Sorry, but

[postbigbang]

No. The code bases were to merge at Windows 2000 Professional. Windows 95/98/ME were based on DOS. Win2K was the merge point at server and 'desktop'. XP came after Win2K, sealing the fate. At Vista, support for 8/16-bit code using DOS functionality essentially died. Try Duke Nukem II if you're unsure.

Re:Sorry, but

[postbigbang]

There was NT 4 workstation, and then 2000 Professional, and then XP. If you're talking about 'home' operating systems, XP was probably the one. The code base for developers merged at 2000. Look it up in 'historical' mags like Windows Magazine, or in other archives. I wrote seven books on Windows from 95-2000, not to mention others.

Re:oh come on

[Nefarious Wheel]

The extra 30GB is to hold the fine print on the new EULA.

Obligatory Pratchett Quote

[cp.tar]

Today's Apple isn't the same Apple. When you replace all of your computers internal components and reinstall a completely different OS, isn't it a completely different computer, even though it has the same case?

"Fakes?" said Vimes. "They were all fakes?"

Suddenly the King was holding his mining axe again. "This, milord, is my family's axe. We have owned it for almost nine hundred years, see. Of course, sometimes it needed a new blade. And sometimes it has required a new handle, new designs on the metalwork, a little refreshing of the ornamentation... but is this not the nine-hundred-year-old axe of my family? And because it has changed gently over time, it is still a pretty good axe, y'know. Pretty good. Will you tell me this is a fake too?"