> One big caveat to Polaris will be that it is unable to run Win32 programs natively. However, reports indicate that Microsoft will be bringing the Centennial Win32 app to Polaris employing virtualization.
So.. no gamers, powerusers, or most corporate deployments. Got it. Can't wait for them to kill this off in 2-5 years.
So for the sake of continuing to hold back technology we are held down by enterprises and DOSGaming.
In all seriousness we complain in the developer community that Internet Explorer and Edge holds back using more modern Javascript yet we make these excuses for OSes.
Almost all programs I use now have a 64-bit executable. If I am using something that needs to run legacy software then I probably don't want the latest version of something to run legacy stuff.
Also the article indicates you can still run 32-bit apps. Most older 32 bit games should run perfectly fine under virtualization. Even then it will be a few years before the majority of Steam user are even using Windows 10 as last I check over 50% were on Windows 7. So by the time gamers shift to Windows 10 i'd imagine their hardware will be more than adequate to handle virtualizing the 32-bit app, I can't imagine there will be much overhead to virtualizing 32-bit apps.
Well, the name made fine sense when it was introduced... Maybe it serves as a good warning to not include such things as dates and hardware versions in names at all.
Opps. Yah I messed up the naming, I meant exes, but just started typing 64 vs 32 which really is only a memory thing, if you simplify it.
Converting a Win32 exe using UWP Desktop Bridge is stupid easy though. Older exe with a lot of legacy dependencies may be more difficult or lack of source code. In which case they need a virtualization to run those legacy app.
So even though all games/apps fall into this category most companies could have a UWP build ready very quickly.
It just wraps the app, but underneath still calls those same win32 APIs. If you remove the APIs, "bridged" apps won't work. Hence why TFA mentions that not all win32 API can be removed and it isn't clear which will be.
So "bridging" your win32 app will not somehow make it immune to win32 api going away.
Thing is, in order to make this kind of cut over, you have to offer something in the new thing that the old thing doesn't have. In order for this to happen, Polaris will probably have to be much faster or MS will have to intentionally cripple regular Windows 10, if only by not supporting it with the same updates.
The more likely scenario is that nobody will end up using those APIs, especially if there are alternatives like Vulkan that work on older Windows versions.
Most DOS games were broken long ago just from clock speed changes. Many of them relied on things like spin loops for timing. Any person playing DOS games now is using DOSBox.
These machines have 64 bit processors, but not much ram (2 GB used to be common, 4 GB is a bit more common now). They also have tiny hard drives. It's possible to find machines sold today with just 32 GB drives. 64 GB is more common.
Very many people just need something as powerful as a phone but with a bigger screen and a physical keyboard.
Without MS providing something for this market they all move to Chromebooks and the Google suite.
MS have, for very many years, offered cut-down product or alternative product to home users. MS Works (a fucking awful product, especially the spreadsheet) or Outlook Express are some examples. If MS offer some kind of Office functionality, even if cut-down, we'll see a bunch of people using it.
It's also related to efforts in Xbox, HoloLens, Surface Hub, IoT, and even still relevant to Windows on phone handsets despite no indication that that is currently a main goal.
The goal has appeared to be since Windows 8 to have a modular Windows where you opt-in/buy-in to legacy backwards compatible systems as needs and hardware change. Polaris isn't so much a new idea as the next in a series of codenames for efforts that have been going on for some time.
Yep, as a gamer there is no chance I would willingly upgrade to this. I like being able to run 20 year old games natively if I feel like it. (e.g. Diablo 1, released in 1996, can still run without virtualization on Windows 10. Sometimes.)
I wish Microsoft was content with just keeping Windows working. I'd buy a special "we will only issue security and compatibility updates" version if I could. I hardly ever want the new features they decide to add. UWP is garbage. the Microsoft/Windows/Zune store is garbage. Cortana is garbage. The Xbox app/game bar is garbage. I don't want any of it, I want Windows to run my games and get out of my way.
> I like being able to run 20 year old games natively if I feel like it. (e.g. Diablo 1, released in 1996, can still run without virtualization on Windows 10. Sometimes.)
As I pointed out below, this isn't even 20 year old games. Name a release in the last year - if it's not a first-party Microsoft title, chances are it's Win32.
You mentioned Diablo 1, well with polaris Diablo III would need virtualization.
Fortnight? PUBG? Overwatch? All are Win32 games.
Excited for monster hunter in Q3? That'll be virtualized in polaris.
Can you explain what you understand will be going wrong when you imply that games needing virtualization will somehow "go nowhere"? Is there something about OS-managed virtualization (rather than application-managed virtualization) that you know that the rest of us don't? Why would a virtualized API even be noticable for modern games, which offload almost all the work to the GPU already anyway?
Only technical people have a clue what virtualization even means. Generally speaking, gamers haven't even heard of the word. Thus whatever virtualization soulution is used must be pretty much automatic. However if Windows already automatically handles this, then why mention that there's no Win32 support? Either people can keep launching their games on Steam or they can't. If they can't, then this new Windows won't win any fans among gamers. Even if there's automatic virtualization in place, as soon as you're losing fps, gamers won't be happy.
Yeah, that sounds like it might preclude any modding (as is the case with current UWP games) which will be a hard pass from me and many other PC gamers (I don't even mod much, usually just fan patches).
The Xbox One has come up a couple times here, and I don't think the people bringing it up understand that the Xbox is the exact opposite of what PC gamers want their PC to be. It's closed down, controlled, and limited compared to a full PC. It may be more secure, but often the beneficiary of that security is not the user but the console manufacturer, game publishers, and game developers.
@jakebasile clearly a legitimate problem to be solved. The Xbox reference is merely to show that virtualized solutions can be very high performance. I've never had my Xbox drop frames or anything that would indicate it's not running with the full power of the machine.
Many games are still CPU bound. May applications are still CPU bound. Notice all the threads lately about how applications have so much more latency now than 20-30 years ago? Virtualization does have impact, and games still max out the hardware available to them.
If this is 100% seamless virtualization that doesn't make anything noticeably slower, it might be acceptable for business users. I doubt it's going to have zero impact though...
I highly doubt it's going to fly for power-users and gamers.
Hmmm. The only way I could get behind this is if there is a remaster of Close Combat 2: A Bridge Too Far. I think Slytherine owns the franchise now, but it was definitely a Microsoft game at the beginning.
It's funny how for many people who make this argument, they'll happily drop $500 for a top of the line gfx card to keep up with gaming, but <$200 for VMWare Workstation is somehow too much money to keep up "being able to play everything older than a few years" on the PC.
(That might not be you, but wanting an OS from 2018 to play games from 1998 or before natively sure makes it sound like it)
I went the other way: if I want to play my old games, I fire up a fullscreened VM, because I don't care what OS I'm hiding when I'm enjoying a nice bit of classic gaming, running in what it thinks is the OS it should be running in. Whatever atrocity the next windows is going to be, as long as VM makers release an update for their software, as far as gaming goes: I have no reason to care.
Once it's out? Pretty sure VMWare's entire business is about making sure their products work on "current operating systems"? Why would you think that VMWare won't update in lock-step?
In this case gaming could be a lot better if they spawned a minimal VM for just the game and hibernated the rest of the OS.
It would solve a lot of the windows gaming bullshit like Cortana popping an alert or filesystem indexing kicking off in the middle of Battlefield 1.
I'm not sure Microsoft's patchwork of fueding PMs will actually allow for something like that to happen but I can always dream.
PS: Apple already has technology to migrate GPU frame buffers across processes so that you can render video in one sandbox and display it in another. I used to work on it. Microsoft could be doing something similar.
>I wish Microsoft was content with just keeping Windows working. I'd buy a special "we will only issue security and compatibility updates" version if I could.
Windows Enterprise LTSB is something like that. Without UWP/Store and security updates for 10 years.
I can’t imagine that running games which are very old would be much more painful in a virtualized environment given the computational requirements are minimal compared to today’s modern applications.
In fact, it could even be a better experience. For example you wouldn’t need to patch it up to the latest version, just run the image with the latest code. Finished? Delete the image. No registry bloat.
So I feel you but it’s hapenning and you should probably get used to it because you’ll need the security patches unless you have an air gapped gaming machine.
But a move like in TFA wouldn't be just old games. It would be literally every game I own except Forza Horizon 3 and Gears of War 4 (the only UWP games I own). New games still max out hardware - that's what I buy the hardware for.
I own over 600 games on Steam and many others on Origin/UPlay/GOG. I would rather use an unsupported, unpatched OS to access those games than simply lose access to them because MS wants to "modernize" something that doesn't need changing.
> Yep, as a gamer there is no chance I would willingly upgrade to this.
It's not an upgrade for existing general-purpose (or any other) editions of Win 10, it's a new OS edition available only with new hardware; it's basically a successor to Win 10 S and Win 8 RT. And, while no doubt Microsoft thinks they've made the tweaks that will make it more attractive to some market this time (from other stories, they expect it to have enterprise and education market traction, which it's at least remotely plausible that virtualization for Windows API apps would provide a good enough solution for), it may suffer the same fate.
That's how it would start. But I don't believe for a moment that MS wouldn't jump at the opportunity to pull the plug on Win32 and decades of backwards compatibility if they could do so. In this case, Enterprise customers are the thing keeping Microsoft in-line for gamers and power users since they would have no tolerance for such a move.
From a technical point of view I can understand their desire to ditch Win32 all too well. And in fact I have wondered years ago why they did not use virtualization to achieve backwards compatibility by letting legacy apps run in their own special VMs with all the right drivers, libraries, etc.
However, if that does not work very reliably and seamlessly large corporate and government customers are going to skip that one.
That would be a huge chance for WINE to become the de facto platform for legacy Windows compatibility.
It already does wonders: https://appdb.winehq.org/
And that's when Windows 10 has pretty low hardware requirements as long as you use an SSD. It IS already pretty light, as modern general purpose desktops go. They must be desperate to get Windows Store going. Didn't they learn anything from Windows RT?
My 64 bit Windows can’t run 32 bit programs natively. WOW64 does quite a lot of work under the hood to make them run, it translates WinAPI, redirects file system and registry, etc. Works quite well in practice.
I expect something similar will happen with Win32 apps in that new Windows 10.
I think the point here is to break with legacy. It's expensive to keep it, error prone and it creates a lot of security issues. By shifting part of the work to software vendors, Microsoft forces the market to move on to newer APIs.
That “legacy” i.e. backward compatibility is the major reason why people buy and use Windows. While indeed expensive to develop the platform, I think breaking it will be a disaster. See what happened with windows phones and tablets.
Isn't this basically the strategy as the old Windows 8 on ARM? Ditch all legacy stuff, no backward compatibility... why would this succeed where Win8 on ARM failed?
What software will it be able to run? I’m not a Windows users, but a future version of the Surface might be appealing. Love my iPad but would like Photoshop on it, for example.
Think this might go in devices used by the next 2 billion computer users? Emerging markets...
Why even call this Windows 10? It's a complete overhaul which has little backwards compatibility. Hell, I'd even say that it's false advertising to still call it Windows 10.
Who said it will be called Windows 10 when it's done? They're starting with their current codebase, which is Windows 10, and are now working to remove any and all legacy code from it that they can, to get to a reduced codebase for a new OS release. Like all projects, it has a codename for now, and when it's done (but right now, "if" it's done sounds a more realistic assertion) it'll certainly not be called Windows 10 anymore.
Just like when Windows XP was developed: they took the Windows NT 4.5 codebase, and started hacking on that until they'd changed enough to make it "Windows NT 5", but it was never released under that name. The underlying version number, however, was kept. So, it'll probably be Windows 11 (or rather, a Windows "something") with an underlying version number like 10.5.x or 11.x.x
Microsoft is the one who said it will be called Windows 10 when it's done. They've repeatedly stressed that Windows 10 is the last numbered update to Windows. That's not to say that the underlying technology won't change, even radically -
it's just that this will be done through Windows Update as an upgrade to the existing OS rather than a new OS that you won't have access to without purchasing an upgrade.
They realized at some point that they had several concurrent versions of Windows out there that they had to support, and many customers had no apparent incentive to upgrade to new versions. So going forward they ostensibly have a single version that auto-updates, and they provide the OS "as a service".
Right now there are a bunch of little forked differences between them all. Microsoft says they are working to merge all the shell differences into a single project (CShell; composable shell), and the point of getting to a clean core (the "Polaris" codename in this article) seems to be to make sure all the rest of Windows modularly slots in on top of that, and a benefit to that is that you could quite possibly more easily change between which modules are loaded/available.
From the Continuum project, we know that Microsoft finds it interesting to be able to switch between Mobile and Desktop editions as much "on the fly" as possible. Could mean they have ideas for switching between Desktop and Xbox modes, too.
> Microsoft says they are working to merge all the shell differences into a single project (CShell; composable shell), and the point of getting to a clean core (the "Polaris" codename in this article)
The clean core is Windows Core OS; Polaris is the desktop-specific composer on top of CShell and Windows Core OS and is also used in this article as the name of a UWP-only Windows edition incorporating the Core OS/CShell/Polaris-composer stack.
Do you have a source for your specifics? From the linked article I assumed that Polaris was just the new codename for the next big phase of Windows Core OS/OneCore/etc work and the article was mistaken or presuming much about a Core/CShell-only edition release plan.
It makes sense that there would be a codename for the desktop-composer on top of CShell, and that would be a far better explanation of things than the article here.
Here's an article which discusses the strategy in some more detail, explicitly identifying Polaris as the composer (but then also seeming to use it as the name for a desktop spin of Windows with Polaris in the stack):
This is a well-calculated PR move. They are trying to steal the momentum from the real Windows 10 and cast it upon their new offering (which is clearly doomed btw).
I'm not sure this deserves a down vote aside from failing to provide context. The major app compatibility problems in vista boiled down to two things:
1. Apps that hard coded which OS they supported (the number one compatibility fix was just too lie about the OS version)
2. DOS applications were now virtualized in an app we called "DOS box".
Though it seems silly to have such hatred pop up over DOS support in 2006, a lot of b2b apps still used it.
Given how much of a beast WOW is (the x86 on x64 later), dropping native x86 support excites me and makes me cringe.
Windows is my favorite desktop OS and hearing that it's going full-UWP kinda makes me feel sick. I don't use a single UWP app outside of the Settings app because they're so limited, ugly and annoying to use.
For instance, just highlight some text in Edge and bring up the context menu. Look how huge and ugly the context menu is and how few menu items there are. This is designed for touch users I guess. Compare with Chrome. No thanks Microsoft!
If Windows goes full UWP I'm going to have to move to Linux, something I've put off doing because I really enjoy the ease of use that I get with Windows where everything that I need to do just works.
I'm always fascinated by statements like these. Do you honestly thing that when it comes out, people won't be updating the applications that currently exist to make Windows 10 look and feel like a "classic" windows?
I run windows 10 with Classic Shell and 7 Taskbar Tweaker. The only time you know it's windows 10 is if you need to be in "PC Settings". For all other purposes it looks and feels identical to Windows 7.
I'm pretty sure when that new version rolls around, it'll take a well counted several days for cosmetic applications to show up to make the OS look less offensive to you, and as long as they get the virtualization right, on your 2019 computer you won't even notice that win32 apps (as in, non-uwp apps, not "32 bit applications") are running virtualized.
Some UWP apps offer mod extensibility (Edge extensions are the most prominent, but a couple UWP games have community mods and more are supposed to happen down the road), and Microsoft has shown interest in supporting general UWP application modding/patching, but haven't yet solved how they want to support that (and still try for security/reliability concerns; not just let it be a new malware vector).
If you think its about malware vectors and not about them having greater control and say over apps and app distribution on windows, I have a bridge to sell you.
You seem to think it's an either/or situation. It's definitely both. It's most correctly that the two concerns are directly related: app distribution is the primary source of malware. The "classic" app distribution platforms for Windows, EXE installers and MSI are terrible for preventing malware. After decades of people complaining that Windows is the most malware-ridden platform, can you blame Microsoft for wanting to make app distribution better?
Sure, you can, just don't mind me rolling my eyes at you. For more than a year now it's been as easy, if not easier, to sideload an APPX on Windows 10 as it is to sideload an APK on Android. (The same ease from a user perspective as using an EXE installer or an MSI file: double click, navigate the security prompts.)
"Windows is my favorite desktop OS ... I really enjoy the ease of use that I get with Windows where everything that I need to do just works."
I was in 100% agreement with this up until maybe a year or two ago, having flirted between Windows and Ubuntu for at least a decade. But now, I find Ubuntu 16.04 (MATE being my variant of choice) to work better out of the box than Windows, even with third party peripherals.
As one anecdotal example, I've a basic EPSON printer / scanner combo on my network. It's always been a fiddle to make work with Windows - you need to go to their website and download their driver and proprietary application so that Windows can find it and produce scans, and it's not all that reliable. The other day, I needed to scan something, and typed "scan" into my MATE menu fully expecting there to be a problem to solve. I opened up the Simple Scan application and not only was the scanner detected and online, but it just worked immediately with no quibbles. I get these sorts of pleasant surprises all the time now, to the point that I can honestly say it's easier to use than Windows.
All of the Applications I use have native Linux support (Firefox, Thunderbird, Spotify, Steam, Unity with VS Code, etc) and work as well as their Windows counterparts. Lots of things I find are easier to set up compared to Windows such as Python programming with virtualenvs, CUDA support for TensorFlow and so on.
Plus - no UWP, no forced updates, no dodgy telemetry, no obnoxious adverts for OneDrive and Office popping up in Explorer... add all that up and it's a done deal. No Windows in this house.
Personally I'd really love to move off of Windows, but I've never had this supposedly "good" experience with linux that people talk about. Everything is overly complicated and inflexible unless you want to strip it down to the kernel and start over, which is probably why there's a thousand distros out there.
I understand all the grumpy, snarly attitudes to this, but it seems imperative to me for MS to extricate themselves from the mire of 24 years of legacy OS. There's no sensible alternative for them, unless they decide to just be maintainers of a legacy OS.
Win32 is not completely legacy. Parts of the API have continued evolving and I don't see insurmountable technical limitations to it evolving further, except that MS prefers a fresh sub-system. The parts that were truly legacy like Win16 or MS-DOS or virtual 8086 have all been already removed.
USER/GDI haven't been removed, which is a big problem as they make proper UI and application isolation impossible. Removing this dependency is one of the major goals of WinRT/UWP/OneCore effort.
What is the end goal? A headless /dev/null? (pardon my exaggeration). As far as I can see, this direction won't attract a lot of customers. People want the simplicity of usage (and it does not necessarily means simplicity of underlying implementation).
App-to-app isolation is probably necessary in the end; at the moment a win32 app can just PostMessage() gaily into any other app whether it likes it or not.
That doesn't necessarily mean getting rid of the old API, but it might mean radically sandboxing it.
A good point. But this is mostly a solved problem. App can only PostMessage to another app running under the same user account. Regarding app sandboxing: it sounds good on paper, but in reality it becomes evident that sandboxing breaks interoperability between the apps returning all of us to the stone age of computing circa 197x.
The alternative is to break all existing compatibility, loose nearly all existing ties with established software publishers (by enforcing a 30% cut from Windows Store), and gain nearly nothing at the very end.
Do you really think somebody will ever write a single app for this all-walled-garden, dishonest effort? The current success of Windows relies on its openness. Add a walled garden and get a dysfunctional, slowly dying OS like macOS.
If they were being held back by the "legacy" parts of the OS, you might be right. But they're not being held back, rather, it's their unique selling point..
Arguably the legacy bits are the bits that work, the modern bits (UWP) are barely seeing any adoption. They're force feeding the modern bits, presumably because they've lost the human resources/mindset/talent to maintain the legacy bits?
This strategy of arm twisting coercion has failed them before and it will fail them again, and again, and again.
I feel this way too. The old shell is bad but I still prefer use it over Powershell. The Windows store stuff and feels like a flat joke that just never ends.
Most people just want an OS that keeps working for their stuff with the occasional security update.
WSL development for bringing in programmers as new users is nice and a feature that stays out of the way, but Microsoft wants to bring in tablet/phone users.
Universal apps so far seem kind of stupid, as an example Paint 3D is a single window only app. Pretty bonkers but I guess the API doesn't support multiple instances.
The one that has been the bane of my existence is Lync/Skype for Business. So often, doing development or even just in mundane business scenarios, its worthwhile to be signed into multiple unfederated Lync pools at the same time, or just different accounts on the same pool.
I've worked with the underlying RTC libraries directly, or through the managed .NET wrapper, and there's no great technical reason for this; it's just that the Lync client has a global single-instance mutex.
The thing is, a lot of the "legacy" stuff actually works fine, and much of the rest of it has suddenly started being addressed due to real competition from MacOS (for web developers) and mobile.
They're worried that people will stop making native windows apps and move to "cloud" or "mobile". Sure, but people are definitely not going back in the box of their proprietary store without a really good offer. What's the unique value proposition here? The value proposition of win32 is "you get to run all your old and new stuff, locally, without a middleman".
I agree. This sounds great, and people keep citing "older" software as the reason they don't want virtualization, which makes no sense. Old software is exactly where you won't see a performance difference.
Do people think that "virtualization" only means running a whole other desktop environment in a window or something? You can run virtualized and native apps side-by-side and most users wouldn't be able to see the difference.
This seems like a good way for Microsoft to free itself from legacy while maintaining compatibility, including for games, for people who need it.
> Old software is exactly where you won't see a performance difference.
For 'current' old software yes, because Moore's Law held until a few years back, but now we can no longer count on that. So a game from this year will presumably become 'legacy' by 2021, but will hardware power have advanced sufficiently that it will run on 2021 virtualized container with nearly the same FPS as today on bare metal?
What I want most from Microsoft is to, in a weird way, go backwards to the days of DOS.
Well, OK, I don't mean that literally. There's one very specific aspect of older operating systems that seems to have been lost forever: OS/App independence.
Our engineering workstations require somewhere in the order of two weeks to fully setup from scratch. It's quite an investment in time, money and pain-in-the-ass factor. In a typical transition we might build a parallel system, keep the old one around, start using the new one and refer to the old one if anything was left behind (small apps, settings and configurations, customizations, etc.)
Because of this we do not upgrade hardware or OS as quickly as we might like to.
If applications installed with independence from the OS this would be very different. What I mean by that is that your "Programs" directory could be moved to a completely new computer with a fresh OS install and you are up and running instantly. In our case we would probably devote an entirely separate physical drive to applications, just as we do for data.
In other words: Why do I need to reinstall MS Office (or whatever) from scratch, ever?
If this were possible we might update all of our computers every two to three years rather than every six to eight. The pain and expense of dealing with upgrades would be gone. An upgrade would be as simple as buying new hardware, moving the physical data and app drives over, running through some form of authentication and off you go.
There are a million different ways to accomplish this wile protecting the rights and licenses of software companies. I don't care to have that conversation because that is a solvable problem. The problem they have to address is this question of having to reinstall all of your software every time you upgrade hardware and OS. That, to me, in this day and age, is ridiculous.
I truly believe the approach taken by MS so many years ago has actually cost the computer industry untold billions of dollars. A company with hundreds or thousands of computers can't even begin to consider upgrades given the non-trivial cost of managing them. If applications were independent from the OS in a portable manner the decision to maintain hardware and OS would be far simpler to make, even at large scales.
The issue goes farther than that. If I need to travel and have access to the $100K in engineering software on my workstation through my laptop I am screwed. Sure, there's remote desktop but that sucks or is nearly impossible to use with some of these applications. Why is it that I don't have the ability of run my apps on any machine I want. I mean, a processor is a processor, is a processor. Outside of requiring specific GPU configurations, I should be able to grab a drive with my apps, grab my laptop, get on a plane and go. I should then be able to use any and all of my applications natively on that machine, even if it is running a different (yet compatible) version of the OS.
In the end I don't care one bit about the operating system. I care about the applications. That's what I use to do my work. If Microsoft understood that they would create an operating system that would help rather than hinder the users with unnecessary complexity, toys and garbage nobody wants.
It's understandable that MS engineers would enjoy working on a greenfield OS, but what MS are selling is legacy compatibility. If they stop offering that product, customers will just walk away.
They should keep Windows Windows and offer the new one as a "boot to Office", not mentioning any reference to operating systems at all, to set expectations right. That one could really fly with a certain market segment (RDP thin clients are a thing). That's assuming they can get Office proper to work on the new API, it might be the one codebase that is more entangled with legacy windows APIs than any other.
I generally agree. UWP may be great tech under the surface actually, but the skin is just not suited for big screens. Too much whitespace, monochromatic colours and everything is so clunky and pared down. One UWP app I do like is the Xodo PDF reader.
I'm sure the UWP will continue to expand, but it probably will never reach the full power of Win32 because MS no longer wants to let app developers get that kind of control easily. At least not regular app devs.
The font rendering is ugly too, for some inexplicable reason. Microsoft haven't even the courage of their own convictions to convert all of Control Panel to UWP.
Parts of the old control panel are public API, like the mouse control panel. You can add tabs to it that fit the old dimensions of the dialog.
And of course it’s Microsoft so half of the company doesn’t want to switch to the next new technology the other half is going to abandon in a few years.
Maybe things have changed, but last time I checked you can't easily publish UWP apps without going through the Windows Store, and the Windows Store requires you to give 30% of your revenue over to Microsoft. That's a huge change from Win32.
My assumption is that it's more of a "sprint milestone" codename for the MinWin/OneCore "next phase" than an entirely new project. (It was also rumored to have the milestone codename "Andromeda" recently, and I wouldn't be surprised that they switched to Polaris to avoid Google's similar codename.)
Reading all these comments makes me feel weird for actually liking UWP apps. Of course certain things don't make sense for them, like games, but I definitely prefer UWP apps over web-apps for things like Messenger and Todoist. I honestly wish more programs were UWP instead of Electron since they would work way better most of the time.
149 comments
[ 5.3 ms ] story [ 177 ms ] threadSo.. no gamers, powerusers, or most corporate deployments. Got it. Can't wait for them to kill this off in 2-5 years.
In all seriousness we complain in the developer community that Internet Explorer and Edge holds back using more modern Javascript yet we make these excuses for OSes.
Almost all programs I use now have a 64-bit executable. If I am using something that needs to run legacy software then I probably don't want the latest version of something to run legacy stuff.
Also the article indicates you can still run 32-bit apps. Most older 32 bit games should run perfectly fine under virtualization. Even then it will be a few years before the majority of Steam user are even using Windows 10 as last I check over 50% were on Windows 7. So by the time gamers shift to Windows 10 i'd imagine their hardware will be more than adequate to handle virtualizing the 32-bit app, I can't imagine there will be much overhead to virtualizing 32-bit apps.
Photoshop, Visual Studio, Chrome.. you name it. Pretty much ALL executables fall into this category. Unless it's a UWP app.
Almost all modern games will fall into this category.
EDIT: To further explain, You still use Win32 for 64-bit processes. It's just a stupid name.
Converting a Win32 exe using UWP Desktop Bridge is stupid easy though. Older exe with a lot of legacy dependencies may be more difficult or lack of source code. In which case they need a virtualization to run those legacy app.
So even though all games/apps fall into this category most companies could have a UWP build ready very quickly.
It just wraps the app, but underneath still calls those same win32 APIs. If you remove the APIs, "bridged" apps won't work. Hence why TFA mentions that not all win32 API can be removed and it isn't clear which will be.
So "bridging" your win32 app will not somehow make it immune to win32 api going away.
Brb gonna go play more overwatch.
I still encounter a fair numver of new games that are built on DX9, not even DX11.
These machines have 64 bit processors, but not much ram (2 GB used to be common, 4 GB is a bit more common now). They also have tiny hard drives. It's possible to find machines sold today with just 32 GB drives. 64 GB is more common.
Very many people just need something as powerful as a phone but with a bigger screen and a physical keyboard.
Without MS providing something for this market they all move to Chromebooks and the Google suite.
MS have, for very many years, offered cut-down product or alternative product to home users. MS Works (a fucking awful product, especially the spreadsheet) or Outlook Express are some examples. If MS offer some kind of Office functionality, even if cut-down, we'll see a bunch of people using it.
The goal has appeared to be since Windows 8 to have a modular Windows where you opt-in/buy-in to legacy backwards compatible systems as needs and hardware change. Polaris isn't so much a new idea as the next in a series of codenames for efforts that have been going on for some time.
This is aimed at making the Windows Store more relevant. MS is looking to be the gatekeeper with an appstore
I wish Microsoft was content with just keeping Windows working. I'd buy a special "we will only issue security and compatibility updates" version if I could. I hardly ever want the new features they decide to add. UWP is garbage. the Microsoft/Windows/Zune store is garbage. Cortana is garbage. The Xbox app/game bar is garbage. I don't want any of it, I want Windows to run my games and get out of my way.
edit: All that said, Edge is kind of OK.
As I pointed out below, this isn't even 20 year old games. Name a release in the last year - if it's not a first-party Microsoft title, chances are it's Win32.
You mentioned Diablo 1, well with polaris Diablo III would need virtualization.
Fortnight? PUBG? Overwatch? All are Win32 games.
Excited for monster hunter in Q3? That'll be virtualized in polaris.
This is going nowhere.
Worth noting: the Xbox One has used virtualization since day one for video games. Done right, gamers shouldn't even have to know that this is what's happening. https://en.wikipedia.org/wiki/Xbox_One_system_software#Syste...
Disclaimer: Microsoft employee, not working on Windows 10 (aka I have no idea how accurate this is).
The Xbox One has come up a couple times here, and I don't think the people bringing it up understand that the Xbox is the exact opposite of what PC gamers want their PC to be. It's closed down, controlled, and limited compared to a full PC. It may be more secure, but often the beneficiary of that security is not the user but the console manufacturer, game publishers, and game developers.
If this is 100% seamless virtualization that doesn't make anything noticeably slower, it might be acceptable for business users. I doubt it's going to have zero impact though...
I highly doubt it's going to fly for power-users and gamers.
64 bit games, and even 64bit _only_ games are increasingly common.
Turns out for many modern AAA games 3GB of RAM isn't enough.
All Windows APIs relevant for game developers are available to UWP apps.
(That might not be you, but wanting an OS from 2018 to play games from 1998 or before natively sure makes it sound like it)
I went the other way: if I want to play my old games, I fire up a fullscreened VM, because I don't care what OS I'm hiding when I'm enjoying a nice bit of classic gaming, running in what it thinks is the OS it should be running in. Whatever atrocity the next windows is going to be, as long as VM makers release an update for their software, as far as gaming goes: I have no reason to care.
It would solve a lot of the windows gaming bullshit like Cortana popping an alert or filesystem indexing kicking off in the middle of Battlefield 1.
I'm not sure Microsoft's patchwork of fueding PMs will actually allow for something like that to happen but I can always dream.
PS: Apple already has technology to migrate GPU frame buffers across processes so that you can render video in one sandbox and display it in another. I used to work on it. Microsoft could be doing something similar.
They don’t even need a full-blown VM, windows 10 has pico processes. That’s how windows subsystem for Linux works: https://blogs.msdn.microsoft.com/wsl/2016/05/23/pico-process...
Windows Enterprise LTSB is something like that. Without UWP/Store and security updates for 10 years.
In fact, it could even be a better experience. For example you wouldn’t need to patch it up to the latest version, just run the image with the latest code. Finished? Delete the image. No registry bloat.
So I feel you but it’s hapenning and you should probably get used to it because you’ll need the security patches unless you have an air gapped gaming machine.
.. what? You have to get this image from somewhere, and given the size of modern games that's going to be an awfully large image.
> it’s hapenning and you should probably get used to it
Words that start customer rebellions everywhere.
I own over 600 games on Steam and many others on Origin/UPlay/GOG. I would rather use an unsupported, unpatched OS to access those games than simply lose access to them because MS wants to "modernize" something that doesn't need changing.
It's not an upgrade for existing general-purpose (or any other) editions of Win 10, it's a new OS edition available only with new hardware; it's basically a successor to Win 10 S and Win 8 RT. And, while no doubt Microsoft thinks they've made the tweaks that will make it more attractive to some market this time (from other stories, they expect it to have enterprise and education market traction, which it's at least remotely plausible that virtualization for Windows API apps would provide a good enough solution for), it may suffer the same fate.
From a technical point of view I can understand their desire to ditch Win32 all too well. And in fact I have wondered years ago why they did not use virtualization to achieve backwards compatibility by letting legacy apps run in their own special VMs with all the right drivers, libraries, etc.
However, if that does not work very reliably and seamlessly large corporate and government customers are going to skip that one.
ducks, runs away
Still don't see GNU/Linux systems on sale at the shopping mall.
Hence centennial project aka desktop bridge.
I expect something similar will happen with Win32 apps in that new Windows 10.
Hence why the only way Linux (the kernel) only got successful for consumers with either a browser OS or a Java OS on top of it.
This is even worse. So dangling old code just hanging around to make the system unstable and insecure. Yay?
Project Centennial is an amazing piece of virtualization but it has its caveats (https://docs.microsoft.com/en-us/windows/uwp/porting/desktop...): it won't be able to convert every legacy software (why isn't Office on the Windows Store for example ?).
At best, it will target locked down hardware like consoles and tablets, but isn't what Windows S was created for ?
Think this might go in devices used by the next 2 billion computer users? Emerging markets...
That changed within the last week, apparently. http://www.zdnet.com/article/microsoft-brings-its-core-offic...
Make an easy migration path for SAP, Oracle and Visual Basic and the problem is mostly solved.
Just like when Windows XP was developed: they took the Windows NT 4.5 codebase, and started hacking on that until they'd changed enough to make it "Windows NT 5", but it was never released under that name. The underlying version number, however, was kept. So, it'll probably be Windows 11 (or rather, a Windows "something") with an underlying version number like 10.5.x or 11.x.x
They realized at some point that they had several concurrent versions of Windows out there that they had to support, and many customers had no apparent incentive to upgrade to new versions. So going forward they ostensibly have a single version that auto-updates, and they provide the OS "as a service".
Right now there are a bunch of little forked differences between them all. Microsoft says they are working to merge all the shell differences into a single project (CShell; composable shell), and the point of getting to a clean core (the "Polaris" codename in this article) seems to be to make sure all the rest of Windows modularly slots in on top of that, and a benefit to that is that you could quite possibly more easily change between which modules are loaded/available.
From the Continuum project, we know that Microsoft finds it interesting to be able to switch between Mobile and Desktop editions as much "on the fly" as possible. Could mean they have ideas for switching between Desktop and Xbox modes, too.
The clean core is Windows Core OS; Polaris is the desktop-specific composer on top of CShell and Windows Core OS and is also used in this article as the name of a UWP-only Windows edition incorporating the Core OS/CShell/Polaris-composer stack.
It makes sense that there would be a codename for the desktop-composer on top of CShell, and that would be a far better explanation of things than the article here.
https://www.windowscentral.com/windows-core-polaris
This Windows Central article actually seems to be the source for most of the other recent coverage, so it may be the most complete piece out there.
Though it seems silly to have such hatred pop up over DOS support in 2006, a lot of b2b apps still used it.
Given how much of a beast WOW is (the x86 on x64 later), dropping native x86 support excites me and makes me cringe.
For instance, just highlight some text in Edge and bring up the context menu. Look how huge and ugly the context menu is and how few menu items there are. This is designed for touch users I guess. Compare with Chrome. No thanks Microsoft!
If Windows goes full UWP I'm going to have to move to Linux, something I've put off doing because I really enjoy the ease of use that I get with Windows where everything that I need to do just works.
I run windows 10 with Classic Shell and 7 Taskbar Tweaker. The only time you know it's windows 10 is if you need to be in "PC Settings". For all other purposes it looks and feels identical to Windows 7.
I'm pretty sure when that new version rolls around, it'll take a well counted several days for cosmetic applications to show up to make the OS look less offensive to you, and as long as they get the virtualization right, on your 2019 computer you won't even notice that win32 apps (as in, non-uwp apps, not "32 bit applications") are running virtualized.
I was in 100% agreement with this up until maybe a year or two ago, having flirted between Windows and Ubuntu for at least a decade. But now, I find Ubuntu 16.04 (MATE being my variant of choice) to work better out of the box than Windows, even with third party peripherals.
As one anecdotal example, I've a basic EPSON printer / scanner combo on my network. It's always been a fiddle to make work with Windows - you need to go to their website and download their driver and proprietary application so that Windows can find it and produce scans, and it's not all that reliable. The other day, I needed to scan something, and typed "scan" into my MATE menu fully expecting there to be a problem to solve. I opened up the Simple Scan application and not only was the scanner detected and online, but it just worked immediately with no quibbles. I get these sorts of pleasant surprises all the time now, to the point that I can honestly say it's easier to use than Windows.
All of the Applications I use have native Linux support (Firefox, Thunderbird, Spotify, Steam, Unity with VS Code, etc) and work as well as their Windows counterparts. Lots of things I find are easier to set up compared to Windows such as Python programming with virtualenvs, CUDA support for TensorFlow and so on.
Plus - no UWP, no forced updates, no dodgy telemetry, no obnoxious adverts for OneDrive and Office popping up in Explorer... add all that up and it's a done deal. No Windows in this house.
https://news.ycombinator.com/item?id=16200863
That doesn't necessarily mean getting rid of the old API, but it might mean radically sandboxing it.
Do you really think somebody will ever write a single app for this all-walled-garden, dishonest effort? The current success of Windows relies on its openness. Add a walled garden and get a dysfunctional, slowly dying OS like macOS.
Arguably the legacy bits are the bits that work, the modern bits (UWP) are barely seeing any adoption. They're force feeding the modern bits, presumably because they've lost the human resources/mindset/talent to maintain the legacy bits?
This strategy of arm twisting coercion has failed them before and it will fail them again, and again, and again.
I could believe the flashing (when running over RDP), glitchy, sluggish Outlook 2016 is using some of the new stuff. It's not a success in my view.
WSL development for bringing in programmers as new users is nice and a feature that stays out of the way, but Microsoft wants to bring in tablet/phone users.
Universal apps so far seem kind of stupid, as an example Paint 3D is a single window only app. Pretty bonkers but I guess the API doesn't support multiple instances.
I've worked with the underlying RTC libraries directly, or through the managed .NET wrapper, and there's no great technical reason for this; it's just that the Lync client has a global single-instance mutex.
The legacy is their crown jewels.
They're worried that people will stop making native windows apps and move to "cloud" or "mobile". Sure, but people are definitely not going back in the box of their proprietary store without a really good offer. What's the unique value proposition here? The value proposition of win32 is "you get to run all your old and new stuff, locally, without a middleman".
GNU/Linux setups are nowhere to be seen on consumer shops, let alone Chromebooks (at least in Europe).
So that leaves Windows for the regular consumer, regardless how Microsoft sells it.
Do people think that "virtualization" only means running a whole other desktop environment in a window or something? You can run virtualized and native apps side-by-side and most users wouldn't be able to see the difference.
This seems like a good way for Microsoft to free itself from legacy while maintaining compatibility, including for games, for people who need it.
For 'current' old software yes, because Moore's Law held until a few years back, but now we can no longer count on that. So a game from this year will presumably become 'legacy' by 2021, but will hardware power have advanced sufficiently that it will run on 2021 virtualized container with nearly the same FPS as today on bare metal?
Well, OK, I don't mean that literally. There's one very specific aspect of older operating systems that seems to have been lost forever: OS/App independence.
Our engineering workstations require somewhere in the order of two weeks to fully setup from scratch. It's quite an investment in time, money and pain-in-the-ass factor. In a typical transition we might build a parallel system, keep the old one around, start using the new one and refer to the old one if anything was left behind (small apps, settings and configurations, customizations, etc.)
Because of this we do not upgrade hardware or OS as quickly as we might like to.
If applications installed with independence from the OS this would be very different. What I mean by that is that your "Programs" directory could be moved to a completely new computer with a fresh OS install and you are up and running instantly. In our case we would probably devote an entirely separate physical drive to applications, just as we do for data.
In other words: Why do I need to reinstall MS Office (or whatever) from scratch, ever?
If this were possible we might update all of our computers every two to three years rather than every six to eight. The pain and expense of dealing with upgrades would be gone. An upgrade would be as simple as buying new hardware, moving the physical data and app drives over, running through some form of authentication and off you go.
There are a million different ways to accomplish this wile protecting the rights and licenses of software companies. I don't care to have that conversation because that is a solvable problem. The problem they have to address is this question of having to reinstall all of your software every time you upgrade hardware and OS. That, to me, in this day and age, is ridiculous.
I truly believe the approach taken by MS so many years ago has actually cost the computer industry untold billions of dollars. A company with hundreds or thousands of computers can't even begin to consider upgrades given the non-trivial cost of managing them. If applications were independent from the OS in a portable manner the decision to maintain hardware and OS would be far simpler to make, even at large scales.
The issue goes farther than that. If I need to travel and have access to the $100K in engineering software on my workstation through my laptop I am screwed. Sure, there's remote desktop but that sucks or is nearly impossible to use with some of these applications. Why is it that I don't have the ability of run my apps on any machine I want. I mean, a processor is a processor, is a processor. Outside of requiring specific GPU configurations, I should be able to grab a drive with my apps, grab my laptop, get on a plane and go. I should then be able to use any and all of my applications natively on that machine, even if it is running a different (yet compatible) version of the OS.
In the end I don't care one bit about the operating system. I care about the applications. That's what I use to do my work. If Microsoft understood that they would create an operating system that would help rather than hinder the users with unnecessary complexity, toys and garbage nobody wants.
They should keep Windows Windows and offer the new one as a "boot to Office", not mentioning any reference to operating systems at all, to set expectations right. That one could really fly with a certain market segment (RDP thin clients are a thing). That's assuming they can get Office proper to work on the new API, it might be the one codebase that is more entangled with legacy windows APIs than any other.
Guess we're going to have Polaris upgrade nags now.
I'm sure the UWP will continue to expand, but it probably will never reach the full power of Win32 because MS no longer wants to let app developers get that kind of control easily. At least not regular app devs.
Also, they are currently working on moving more and more settings from the Control Panel to the UWP Settings app.
And of course it’s Microsoft so half of the company doesn’t want to switch to the next new technology the other half is going to abandon in a few years.
https://www.google.com/amp/s/arstechnica.com/information-tec...
https://news.ycombinator.com/item?id=16259539
https://www.pcworld.com/article/3194946/computers/chromebook... Chromebook shipments surge by 38 percent, cutting into Windows ...