Windows "Polaris"


I have been reading some news on Windows Polaris on Windows Central. As they report, it is stripped down OS to components where they will remove the legacy elements and make Windows a true modern OS. Polaris on the other hand will be a light weight Windows OS desktop without the Win32 stuff.

I haven’t read any article on so far on this. I would love to know what Thurrot and the team thinks about this.

Link below

Comments (53)

53 responses to “Windows "Polaris"”

  1. TheJoeFin

    Brad and Paul talked about this on First Ring Daily today (Friday) and mentioned how this story seemed odd, perhaps this isn't a complete picture of what is going on. Also the concept of 'pulling out the Win32 stuff' is sort of impossible because it makes up the foundation of UWP. So, maybe there is a better way to describe what Microsoft is doing, but for the most part it is not nearly as simple as mentioned in that Windows Central article. Also maybe there is some confusion with Polaris and Andromeda or maybe both are two sides of the same coin. Who knows.

    • jimchamplin

      In reply to TheJoeFin:

      Yeah, but they could build a new set of foundation libraries that support only the calls used in UWP without exposing the legacy and deprecated elements.

      Also, eliminating Win32 may refer more to removing Win32-based system components and replacing them with newer UWP, or at least managed .NET versions with a modern UX.

      • skane2600

        In reply to jimchamplin:

        I'd be surprised if they removed all the Win32-based system components. Despite all the .NET promotion over the years, the Windows group has never embraced .NET except as a layer on top of the mostly C-based system components.

        It makes sense IMO, you want to minimize the overhead of both size and processing at the lowest level of an OS.

        • jimchamplin

          In reply to skane2600:

          Well, what would be really technically most impressive is if they wrote a new personality for NT from the ground up to run UWP. Instead of keeping any of the legacy code, they just rebuild it all from zero.

          However it it would be the software equivalent of H-bombing Windows.

  2. martha

    Well, I would not think that it will be popular

  3. Dan1986ist

    Has anyone compared the write-ups over this Polaris thing on MSPU to those of Windows Central?

    • RR

      In reply to Dan1986ist:

      I think MSPU is more of a peripheral player here, and generally too, most of the time. I think the site? Writer? has a volume for eyeballs strategy and just pumps out secondary data, often without regard to quality. In this case the writer just read Windows Central like everybody else, and decided to counter program with a sort of half report (you have to look hard for the attribution) half editorial that goes opposite to the Windows Central hype. I wouldn't waste too much time analyzing that for any meaning.The MSPU writer has no independent leads, just his derivative on the same report he read, just like you. If he had written the same blog as a comment in the Windows Central comments section like everyone else, you would not have thought of it beyond that.

  4. jimchamplin

    I'm really starting to get a lot of different viewpoints, and honestly, a ground-up rewrite of Windows would be a good place to address these issues.

    There are two major things that contribute to Win32 "feeling old-hat." Let's discuss them.

    The first is the security issue. That would be solved by rewriting the shared libraries that the OS uses to execute Win32 code. If they're rebuilt from the ground up to run Win32 code in a secure way that doesn't break the software, then that issue is solved. If the OS executed each Win32 process in its own sandboxed space, then it would be instantly as secure as UWP.

    Apple did this without changing their program model. Mac App Store software executes in a secure sandbox and the software didn't have to be rewritten, simply submitted to the Mac App Store. The App Store disallows certain things like deep OS hooks, but does have a lot of permissiveness for utilities. It's not perfect, but in the end it's better than having to rewrite the software from the bottom up.

    The second is the "look and feel" issue. That's way more subjective and honestly very secondary. I really wish that that everything on Windows 10 looked like UWP software. I prefer the cleaner, more resolution-independent style. Others may not. In the end, if Microsoft would simply let legacy software follow the light/dark theme choice set by the user then it would go a LONG way toward fixing the discontinuity.

    As it is, Win32 applications call the same MSStyles-based theming engine that was implemented for Windows XP. Back in XP, if one selected the Olive theme, then older apps that didn't use the Luna theme would STILL get recolored to match. If you chose silver, then everything matched the Silver theme. Why can't Microsoft create a dark theme for Win32 software and switch to it when the user chooses the dark app mode?

    That would go SO far toward fixing the discontinuity. Maybe change some of the metrics, make the fonts larger, give it the thicker borders and flatter style so that Win32 software LOOKED like it fit.

    Software that renders its own UI, like Adobe apps, Office, and Visual Studio? Well, they don't get forced into it.

    If they're going to rebuild Windows from scratch essentially, why not acknowledge their strengths and keep them, but design it so that the strengths remain strengths?

    • skane2600

      In reply to jimchamplin:

      If each Win32 process runs in its own sandboxed space, compatibility would be broken because many programs rely on the ability to anonymously interact with other programs as well as accessing cross-program data in a common registry. It might be possible (or maybe not) to design new approaches that still support this kind of interaction and integration but legacy programs wouldn't be able to use them even if such work-arounds existed without being rewritten.

      • jimchamplin

        In reply to skane2600:

        Yeaaah. I didn’t need consider that. It’s possible that such processes could be virtualized within the sandbox so that it’s transparent to the application. But in reality, Windows is handling the request securely.

        Another approach could simply put all Win32 software in a single sandbox. That keeps them compatible, and protects the system from them. Hyper-V could be used to create an independent Windows instance, while the operation of the software remains completely unchanged to the user’s perspective.

    • hrlngrv

      In reply to jimchamplin:

      . . . If the OS executed each Win32 process in its own sandboxed space, then it would be instantly as secure as UWP. . . .

      That presumes there's no benefit to old-style interprocess communication. Session-level scripting, a la AutoIt, wouldn't be possible, nor presumably would VBScript automating multiple different applications.

      . . . The second is the "look and feel" issue. That's way more subjective and honestly very secondary. . . .

      Agreed it should be secondary, but is that how MSFT views it? OTOH, there's the related issue of high res displays and scaling. MSFT's approach is that it's only available for UWP is disappointing.

      As for MSFT itself observing Light vs Dark, set the Light theme then check whether the [Win]+X menu observes it. [Submitted months ago as Insider feedback.]

      . . . why not acknowledge their strengths and keep them, but design it so that the strengths remain strengths? . . .

      Do you honestly believe MSFT still views making Win32 software more consistent with Windows 10 as a benefit to MSFT? That is supports where MSFT wants to go with Windows?

      • skane2600

        In reply to hrlngrv:

        Thanks for mentioning AutoIt. I've used VBScript for Office automation and Visual Test for testing back in the day but wasn't aware of this tool. I don't have any specific need for it right now but it could come in handy some day.

  5. warren

    At the very least, they're going to have to finish killing off Control Panel before this can happen.

    • hrlngrv

      In reply to warren:

      Agreed. Gotta wonder why it took MSFT more than 3 years (from the initial Windows 10 Insider builds) to move Fonts from Control Panel to Settings.

      As for other things remaining in Control Panel through Build 17083,

      • AutoPlay (also in Settings, but more comprehensive in Control Panel)
      • BitLocker
      • Color Management (separate dialog, will this remain in Windows?)
      • Credential Manager
      • Date and Time (separate dialog)
      • Default Programs (also in Settings, but more comprehensive in Control Panel)
      • Ease of Access Center (also in Settings, but more comprehensive in Control Panel)
      • File History
      • Fonts (also in Settings, have to wonder why the redundancy)
      • Indexing Options (separate dialog)
      • Infrared (separate dialog, will this remain in Windows?)
      • Keyboard, Mouse (2 items, but related, raises the question how easy will it be for 3rd party peripheral hardware makers to add user configuration sections to Settings)

      I'll stop there, roughly half way through. I wouldn't include Administrative Tools or Backup and Restore because those are more separate utilities and accessible from the Start menu. Nor would I include File Explorer Options or Internet Options because those are specific app configuration accessible through File Explorer and IE menus.

      The interesting question is whether Infrared and Phone and Modem are going to make it to Settings at all.

      Given how long it's taking MSFT to move things from Control Panel to Settings, it's hard not to conclude that it's one helluva lot harder to add things to Settings than it has been to add things to Control Panel. [deleted stuff] ADDED: shouldn't have deleted it: Settings's Date and Time section only handles changing time zones. Setting date and time still requires Control Panel. Didn't Windows phones have some way to changing date and time?

      • maethorechannen

        In reply to hrlngrv:

        The interesting question is whether Infrared and Phone and Modem are going to make it to Settings at all.

        There's no question that they have to keep them. Otherwise people won't be able to beam files off their PDAs and then upload them to a BBS.

  6. davidblouin

    You're talking about a website that talk about Windows Phone like it's still a thing.

    For me this discussion start and ends there.

  7. skane2600

    IMO, Microsoft hasn't had a consistent vision since Windows 7. It's like they're flailing around trying to latch on to whatever flotsam and jetsam they can find. The last thing Microsoft needs is another Windows variant.

    • Tony Barrett

      In reply to skane2600:

      That is very true. Win7 was Microsoft's last real 'vision' for Windows. Vista done right. Developed for the users (not Microsoft's) benefit. Monolithic 3 year dev cycle. Feature rich API. Rock solid. It works. Now MS are chasing their tale going round and round. Now we have the potential for yet *another* version of Win10. Disaster for everyone, inc MS.

  8. jimchamplin

    Yeah. Windows Central has been really selling this. I was really concerned if it was... you know? Real?

    I'm not accusing Dan Rubino of making things up, but perhaps just they got what they thought was a scoop and it's really just... Not what they think? They're talking about it in huge, sweeping ways. It sounds simply like what Windows 10 S is supposed to be.

    • irfaanwahid

      In reply to jimchamplin:

      One thing I know, Paul/Brad usually write up something once they have concrete evidence or authentic source about some information. I assume so far they haven't got any heads up from their sources on this.

      For Dan and Zac, they do hype up things a little bit, but I doubt that they are shooting the arrow in the dark with this. They have got pretty elaborated explanation on this in their article.

  9. stelaban

    As they record, it's miles stripped down OS to components where they'll cast off the legacy factors and make windows a true modern OS. Essay Writing Service I'm not accusing Dan Rubino of creating matters up, however perhaps just they were given what they thought became a scoop and it's surely simply.

  10. ponsaelius

    As a journalistic exercise this is all good stuff. In some sense there is a lot of glib comment. I am part of that. There is also dis-belief in Microsoft's strategic analysis because, with the exception of the enterprise, Microsoft visions of the future have tended to benefit others - mobile as an example.

    The starting point is the removal of "legacy" code (read WIN32 here) and replace with UWP. UWP, to Windowscentral, is not actually UWP apps per se but the wrappers like Centennial that puts apps in the store. Spotify is an example of this. Other UWP considerations are the web apps that are placed in the store. UWP, as apps on their own, seems to be a mess with few "must have" killer apps. The universal apps platform is more likely to be web apps based on PWA.

    So from a political perspective forcing users into the Microsoft Store, taking a cut from sales, advertising and the like is all good for Microsoft. Upselling store visitors will monatize Windows and, as I am sure they will say, be a more secure experience for users looking for apps.

    The new light UWP WIndows is a commercial decision as well as a technical one. People have not really been using the store so Microsoft will force them as part of the platform re-design.

    Win32 will be in Centennial apps and perhaps some kind of appv container layer or maybe even in the cloud. There are options. However you new Windows will be essentially a UWP experience removing the 25 years of backwards compatibility and "modernising" it in perhaps a more thoughtful way than rouch first Windows 8.

    All good so far. Each of these things is problematic in it's own way but, for the general public, who use less productivity software, are using the cloud, arguably spend most time in smartphone apps or the browser all of it may make sense. By the browser we mean Chrome. Google obviously has a browser based OS so the case to run a light Windows OS without Chrome for people who mainly use cloud services is something I struggle to understand.

    The next core problem I see is that this imagined new Windows seems to be for "normal" users. IT pros and the enterprise are likely to continue to need tools that need access to the command line and the WIN32 api. Many enterprises have their own software repositories, use System Center for deployment etc and bypass the store completely. The security is often policy based via Group Policy or other tools like Appsense.

    Microsoft's core business is moving away from consumer to enterprise. So far UWP and enterprise don't go together. Consumers don't seem to be attracted to the store.

    Technically WIndowscentral may be onto something. Whether it amounts to anything other than a reboot of the ideas in Windows 8 or Windows RT is to be seen.

    • skane2600

      In reply to ponsaelius:

      Perhaps I'm being pedantic, but Google doesn't have a "browser based OS", but they do have an OS whose only application is a browser.

      • hrlngrv

        In reply to skane2600:

        If you're pedantic, I'll be picky.

        Chrome OS has a Linux kernel and various Linux subsystems, but the desktop environment and window manager are unique to Chrome OS. The Chrome browser is the only major application, but there are some others, a few bundled with Chrome OS. They're the ones that when you right click on them in the launcher, the menu shows no option for open in window. That means they only open in separate windows and never as tabs in the Chrome browser. Examples: Files, Calculator, Camera. There are also a fair number of 3rd party standalone apps (in the Chrome OS context) in the Web Store, e.g., DOSBox, which is decidedly not just a browser tab.

      • ponsaelius

        In reply to skane2600:

        You are being pedantic. However I guess I was taking the observation from the perspective of the normal user. So I was being technically inaccurate. Correction accepted.... <s>

  11. hrlngrv

    The big question remains where the value of Windows lies. Is it in MSFT's preferred new direction, stripped down, all software from the Store and nowhere else, security and simplicity above all? Or is there remaining value in all the Win32 software sold in the last few decades, much of which will never make it into the store because it's too old for the rights holders to bother doing so? That is, how much of the value of Windows is bound up in the value of older 3rd party software which can run under Windows? Assuming some value, some Windows users may find more value in that old software than in new Windows's security and simplicity.

    I've seen Rubino's YouTube summary of Polaris in which he says (admits) that new Windows isn't meant for people who need to run such things as AutoCAD or PhotoShop. It reminded me of Jensen Harris's comments at Build 2011 in which he stated that the then-Metro UI wasn't mean for dense UIs, that there'd continue to be a place for software with dense/rich UIs. I can't help believing that no matter how much MSFT and its fans want it otherwise, Windows without Win32 support is just another iteration of Windows RT.

    On a different tack, if complex software migrates to running in the cloud on application servers, as long as all the major local device OSes have clients which can connect to those servers and handle I/O, the OS running on local devices becomes far less relevant or becomes far more reliant on the apps ideally suited to running on simple local devices, and that means MSFT would exchange the remaining value of accumulated 3rd party Win32 software for the value of the MSFT Store and all the wonderful UWP apps in it.

    • skane2600

      In reply to hrlngrv:

      IMO, it's a trade-off that was never necessary to make in the first place. For years Windows has satisfied the needs of both casual and sophisticated users. If casual users found UWP apps significantly easier to use than Win32 apps, they'd be flocking to them. After all, it's the lack of sophisticated apps that are the primary criticism of the store.

      It's really a question of whether Microsoft wants to make Windows just another commodity OS and compete with MacOS, Android and iOS with no legacy advantage, or continue to enjoy the unique benefit of being the only OS that can run most industry-standard and special-purpose programs.

      • hrlngrv

        In reply to skane2600:

        . . . with no legacy advantage . . .

        Almost an oxymoron. In tech generally, computer software particularly, legacy carries bad connotations. IMO, the accumulated mass of 3rd party Win32 software is Windows's greatest advantage.

        Instead of calling Win32 legacy, I'd call UWP new & improved!

        • skane2600

          In reply to hrlngrv:

          I can't tell exactly what your position is from that post, so what I'm going to say you may already agree with:

          I would argue from a professional developer's perspective, "legacy" and "new" carry no implicit connotations, it all depends on the context. The idea that "new" is always better or "legacy" is always bad is more of a enthusiastic beginners perspective.

          • hrlngrv

            In reply to skane2600:

            I'll be clearer. Win32 is legacy for Windows like Shakespear, Austen and Orwell are legacy for English literature. To be even clearer, the accumulated software which eventually finds its way into the MSFT Store will be lucky to reach 1/10 the value of all Win32 software so far.

            That's from a user's perspective. No doubt these days MSFT regrets the perceived value of Win32 software since it makes it so difficult to move on to the Next Great Thing.

            • skane2600

              In reply to hrlngrv:

              The "perceived" (and real) value of Win32 software is why the Mac never achieved 50% or higher market share. Perhaps if they had become a minor OS player they'd be able to work on the Next Great Thing without that burden of being a highly-successful company.

              • hrlngrv

                In reply to skane2600:

                The "perceived" (and real) value of Win32 software is why the Mac never achieved 50% or higher market share. . . .


                Historically a huge benefit for MSFT, but now it seems MSFT and too many MSFT fans view it as a burden, mere legacy. UWP is no longer necessary, and whether it presents any advantages in MSFT's PC-dominated universe seems speculative at best. PWAs as a remedy for sparse MSFT Store offerings would only promote Chrome OS nearer to parity with Windows 10 S specifically and Windows generally.

                Pity MSFT no longer has any interest in playing to its strengths and prefers chasing the mirage of a 30% cut on all 3rd party Windows software sales.

          • longhorn

            In reply to skane2600:

            I'm not a developer, but it's hard to see the advantage of UWP compared to Win32. This is what I believe and if I'm wrong please correct me.

            Win32 has unprecedented binary compatibility for consumers; 15+ years

            Win32 programs even fairly complex ones don't need to touch the Registry.

            Only bad programs/installers leave junk behind

            Win32 programs can be packaged and used with for example Chocolatey

            No technical reason to not have Win32 in Windows/Microsoft Store

            Win32 can be sandboxed; Chrome has its own sandbox which protects the program

            Windows is protected from Win32 by UAC and antivirus software. By scanning Win32 programs for malware and putting them in a Store the risk of malware infecting the OS is small.

            You always have the option of not installing Win32 programs if these scare you

            Portable versions can be useful on a USB-stick, but exe-files can also be blocked (for example in the workplace)

            UWP apps are sandboxed by default so forget useful system utilities

            UWP APIs are limited so apps will be limited compared to Win32

            UWP apps can in fact leave junk behind just like Win32 apps

            UWP APIs are written in flaky C++ compared to Win32 API's more stable C code

            I'd say that for 95 % of Windows users (business and consumers) Win32 is the reason to use Windows. Why is Microsoft obsessed with creating Windows RT again and again? Since 2012 users have rejected the Modern/lesser parts of Windows just because of the legacy advantage.

            • skane2600

              In reply to longhorn:

              I agree with a lot of what you said although I disagree with some of it. I don't know if UWP APIs are written primarily in C or C++, but the latter isn't flaky. In fact sometimes C++ is faster and often less error-prone than C. C++ is a superset of C so there's really nothing that can be written in C that can't be just as efficiently written and executed in C++.

              Sometimes people imagine that C++ implies complicated class hierarchies but not necessarily. At a minimum classes are similar to C structs with the addition of functions. No additional indirection is required unless you add features to the class that requires them.

              • doofus2

                In reply to skane2600:

                My guess is that the code is in minimal C++ like you mentioned. In system components you have to be careful when allocating memory (e.g. control where it is and its persistence across processes).

  12. doofus2

    The whole Win32 vs UWP debate is stupid. All MSFT had to do was add a simple scalable UI system to Win32, add an installer/updater system that works with third-party stores, disallow usage of older Win32 technology (DDE, etc), and have devs agree that they couldn't touch forbidden areas (protected registry areas, file system areas, etc.). All of that should be enforced at the kernel level anyway, not some bogus "broker" system that results in orders-of-magnitude lower performance. This "Modern Win32" scheme could have been backported to Win7 and had a chance of success.

    We didn't need a pile of new APIs to read and write to disk. We didn't need stupid "async" APIs that actually complicate storing information reliably. The whole process of reading/writing to disk needs to be "async" (Open, Read/Write, Close) not each step along the way (OpenAsync, Read/WriteAsync, Close). Win32 already had the best multithreading APIs and system of all the operating systems and they threw all that away. It's ridiculous.

    Pop quiz: how many dollars do you end up with when you take 30% of zero dollars?

    • hrlngrv

      In reply to doofus2:

      . . . disallow usage of older Win32 technology (DDE, etc) . . .

      Some of those older features provided functionality. Personally, I never found much benefit to DDE, especially after web queries became built into Office. OTOH, OLE still serves a purpose.

      As long as there are options, I'm fine with one of those options being lock down everything as long as there's another for allow everything. I don't have a problem with the former being the default as long as it doesn't require an account with Administrator privilege to allow most features.

      Finally, if you believe MSFT has any interest in backporting anything to Windows 7, you haven't been paying any attention to MSFT over the decades. MSFT wants Windows 7 and 8.x to die off ASAP, ideally well before EOS.

      • doofus2

        In reply to hrlngrv:

        If they had made "Modern Win32" available on Win7 back when WinRT8 was introduced it would have had a chance to succeed. Instead, the "new" thing was only available for Win8, so it had zero users and zero developers. Devs would have had to maintain two separate code trees for Win32 on Win7 (where they made money) and WinRT8 on Win8 (where they made no money and would only receive 70% of any future sales). More work for less profit. It was DOA.

        • hrlngrv

          In reply to doofus2:

          Windows RT and Windows 8 were what they were as a means to force the numerous Windows PC user base to become familiar with Windows phones. MSFT did that either in the hope that the benefits of live tiles were just so great that people would flock to Windows phones or (cynical) to inure PC users to the phone UI and hope that force of habit would lead to increasing Windows phone sales.

          The last thing MSFT wanted to do was make Win32 and Windows 7 stronger relative to new & improved Windows.

          MSFT wanted to reverse Android's relative growth, and MSFT may have been right that the best option for doing so was Windows Phone and the WinRT API for PCs in Windows 8. I'm sure MSFT believed making Windows 7 stronger (or making Windows 8 more like Windows 7) wouldn't have produced significantly higher Windows PC sales and would only have hurt Windows phone sales. Unfortunately for MSFT, that failed. There are now more devices in use running some version of Android than all devices in use running some version of Windows. That's not likely to change within the next decade. A reborn Windows RT called either Windows 10 S or Polaris won't change that. It may be the direction MSFT wants to take Windows, but that doesn't mean it's the direction Windows PC users would be willing to follow.

    • longhorn

      In reply to doofus2:

      "It's ridiculous."

      I agree. Give it a "Modern" label and hope people will look no further. The right way to develop something, both software and hardware, is to gradually refine a product into a more and more mature state. Apple has been good at this.

      At Microsoft there is not much respect for the past. Old geniuses have left and the buzzwords are "reimagine", "hit refresh", whatever... instead of building on a solid foundation. New management, new developers... it's their territory now and they want to piss everywhere. To be fair this isn't unique to Microsoft. Many companies behave like this. It was actually the old guys that started this with the hastily produced Windows 8. The saddest thing is that Microsoft as a company learned nothing from Windows 8. The Win32 eco-system is a potential goldmine for Microsoft, but the company is determined to venture into the wild where no developers or users can be found.

      It's ridiculous.

      • doofus2

        In reply to longhorn:

        The best comparison is with the OS/2 debacle in the late 1980s. MSFT tried to kill Windows then and it failed miserably. Weise and the others resurrected it internally and it caught fire. Are there any people like that still at MSFT???

        Once the company finishes killing their main platform, Windows, they'll be adrift on a pseudo-platform, Office, and easily replaceable "cloud" system. Cloud systems are remote and independent of the users out in the field (do you care where a website is located? Nope!). A entrenched local platform in the field, like Windows, is very difficult for a competitor to defeat.

        Note that UWP is just WinRT10, an evolution of WinRT8. It has been out and promoted for over five years and is going nowhere. Time to kill it. The damage is already irreparable.