I think what's next for Windows is competitive x86-64 emulation to enable a new generation of ARM devices with strong battery life, and forcing Intel and AMD to drop prices to compete.
Windows can be the Android to iOS. Ultimately, most desktops and laptops run Windows, and if Microsoft is able to make it competitive on pricing (#1!), battery life, and performance, it's hard to imagine Chrome OS or the single user iPad seriously eroding Windows.
Most people in the world aren't affluent like the average HN reader. Their biggest consideration when buying a computing device is value, and Windows PC + ARM can win here.
Imagine a $499 laptop with a solid M1-level processor, a nice IPS display, good keyboard, 10+ hours battery life, and Windows 10.
That's serious competitor to both Chrome OS devices and Mac devices.
We should see with Alder Lake. Intel is adopting the same BIG.little approach, with a mix of high efficiency and high performance cores.
Of course, one thing Intel and AMD can't catch up on is pure transitor size. Apple has TSMC 5nm, Intel is on the equivalent of TSMC 7nm (Intel 10nm), and so is AMD.
This seems like a a weird thing to speculate about. Of course iOS isn’t exactly macOS, but it’s essentially the same OS with a few different feature flags and a different set of preferred UI frameworks, which are increasingly not so different. That almost exactly same OS is now running on exactly the same processor in an iPad. You can more or less compare that device to anything with similar thermals in an x64 machine.
Isn’t it interesting that Windows(and macOS for iOS) remains to be THE platform for creation and Android/Chrome OS/iOS to be consumption-only platforms? Combined with the fact that Intel and AMD each owns Altera and Xilinx...
This seems unlikely. I could see this make a bit more sense if Microsoft wouldn't have screwed up their mobile platform so badly, but if you just serve desktop/laptop, a switch to arm doesn't make sense. I expect x86 to catch up enough to not justify the performance impact of even really good emulation on arm. Usually if you go Windows it's because of its huge software library and backwards compat, so I wouldn't bank on having all the important apps be native arm after a year or two. Everyone else is already on a different platform.
I see people still use Windows only for old business apps (like bookkeeping and such). Almost all of their other work moved to web.
I successfully moved my wife's notebook to Linux as she don't need legacy apps. Photos, Movies, Music all works fine in Linux and web. Desktop apps she used are cross-platform.
Since using Linux all rant she had "aaaaarrrgh, Windows again updating and requires restart!" disappeared. And performance just rocket up agains Windows.
I think, importance of native desktop apps will fade away and Windows can lose this advantage.
Debian/KDE. I used to use it since 1997. KDE looks well for Windows user.
But this week I bought new Dell Vostro with Ubuntu 20.04 preinstalled. Although Ubuntu gave me some unpleasant impressions (controversial UI decisions, buggy as hell (can't imagine they call it LTS)) I gave it 2nd chance with fresh install of 21.04.
Don’t the already have x86_64 emulation? Pretty sure I have seen them demonstrating downloading a x86_64 binary from the internet and running it unchanged on Windows 10 ARM laptop.
I wonder why there hasn't been a Raspberry Pi with Windows on ARM yet, or at least something similar. This could kickstart the Windows on ARM ecosystem, so more SW would be available when the laptops/tablets come out. It would probably increase the sales of Rpi by a lot as well.
This... Was appalled by how much data the collect by default even when thinking I had unchecked everything during install and startup. Then I ran Windows Update and they automatically installed HP spyware.
The thing is that backwards compatibility is already developed, it called Windows and already paid for. That is the problem with all these ideas of rebuilding, it cost lots of money for ending up in at same place you already are.
I would be shocked if Microsoft abandoned the NT kernel but I wouldn't be too surprised if they open sourced a lot of the win32 runtime and turned it into some community maintained thing. They've already done this with .net.
What would adopting a Linux kernel solve? Nothing.
Problem here is not the NT Kernel, the problem is user space.
So building everything on top of the Linux kernel would end up in the exact same position as before because the need of backwards compatibility, this would cost money with no benefit.
And the thing is they have already integrated Linux with WSL2 and can now run Linux desktop applications anyway.
Rebuilding Windows seems unlikely because there is no money in building operating systems, meaning for every operating system you build you need to maintain it indefinitely without getting paid for it.
What Microsoft need to do is to find a new market for their products. With a new market you can package Windows (or whatever) to fit that market specifically.
One question likely is if backward compatibility is a must-have feature or if you can figure out different ways to achieve it. Like could you run the legacy apps in some virtualized environment.
Windows used to be The platform you would use to run the apps developed with Microsoft technologies. Not sure about trends at large, but personally I would likely just go with Linux if I was creating something new with .net.
Last year the rumours said Microsoft was planning to launch Windows 10X without immediate support for Win32.
That would break the proprietary Nvidia and AMD graphics card drivers.
Just as a start.
Device driver universe on Windows may be a blessing or a curse, but entire industries have labored to make it sorta work, and to follow troubleshooting scripts (tech support) when it doesn't.
Aside from that, the leading edge graphics card drivers are their own, multiple OS universes. With their own collection of hacks for specific games.
I'm not familiar with the current state of high end gaming on Linux, but I believe Windows is still the platform for big game titles.
The Windows NT kernel is on par if not better than the Linux kernel. Replacing it will add no benefit, all the issues described in the article are well above the kernel level.
The very obvious benefit of outsourcing kernel development to Linux is cost avoidance.
When there is a volunteer army that can go after Meltdown and Spectre orders of magnitude faster than the corporate stable of developers, the business case seems to make itself.
If you step back and look at the historical arc, it won't be much longer until Windows is a desktop experience on a Linux kernel, much as OSX is upon Darwin.
A UI cleanup is needed. What Microsoft needs to do is pick a style and actually complete the transition. They aren't done with the transition to the Win10 look and feel. They never complete the metro migration. The Vista/Windows 7 look was never implemented through out the entire OS. In fact, even Windows XP never removed all the Windows 95/98/NT/2000 components.
At this point I also don't care if it's at the cost of backward compatibility, but they also need to fix the response times in the Windows UI. Given that high end games play very well on Windows, the OS is clearly able to process inputs fast enough. Yet for some reason when you're navigating the Windows desktop, everything seems to react pretty slowly.
Windows 2000 was the last version I was able to use without having my mouse hand cramp up. Something in Windows causes me to tense up my left hand (my mouse hand). It doesn't happen in macOS and it doesn't happen on Linux (Gnome/KDE/XFCE).
> And how do I move a browser window, now that my 8-15 browser tabs consume all of the title bar area?
This is the fault of browsers, not the OS. I think chrome was the first to do away with the actual title bar and put tabs there, then everyone followed. You can change this in the settings I think, there should be a small area you can drag between the minimize button and the last tab (on FF at least), and then you can always hit the "Alt" button to bring up the title bar you can then grab.
Remember when software used to have a manual to explain how to use it? Back when you could actually learn some UI principles and have that knowledge transfer between apps?
Now every application and OS component does its own thing however and whenever it likes, and the whole experience becomes impossible to learn other than as a giant list of special cases and exceptions, which will change with the next round of patches.
This is why I can’t explain to my mom how to use her computer any more, and generally tell her to use her phone instead.
> At this point I also don't care if it's at the cost of backward compatibility
Yeah, no. This is the heart and soul of MS. Apple breaks stuff all the damn time. MicroSoft keeps things working forever. Some people pay for upgrades. Some pay so they never have to upgrade. Consumers upgrade, and apple makes a lot of money. Businesses don't upgrade, and pay for the privilege.
Kinda sucks, but preserving how we got here is a huge deal. 20 year old systems are core to frighteningly important systems.
Sure, make another, good luck handling the zillion battle tested edge cases. Windows may be horrible now, but ancient code still works. Your hand cramp is less expensive than a rewrite.
(this might be a little aggressive. apologies, this is kinda a hot take)
No no, you're completely right. Fixing my complains about Windows comes at a horrific cost.
Microsoft can continue down the current path and Windows will still be hugely successful and widely used. It's will just be terrible and continue to get worse. Most people and businesses will just live with it. I mean if Windows is just the thing that launches Steam, then you don't really care. If it's just the shit SAP/Excel/what-ever launcher your IT department forces you to use, then fine. Windows just won't be a smooth experience or something people enjoy using.
The problem is right now the biggest thing going for Windows is the historic baggage. If Microsoft makes be choose between baggage and something new and possibly better, I'm also going to consider other options. In that case, Microsoft had better offer something amazing, that's worth the cost of removing history.
They've done it from time to time. The benefits of NT outweighed the loss of tools that required the windows 9x models; although not enough for many to switch to 2000; XP made some concessions to seal the deal. The benefits of 64-bit windows are generally worth the loss of 16-bit applications (although winevdm shows that they could have made that work).
But, I haven't seen much from Microsoft lately that leads me to think they'd be able to pull something like that off. If I had to make a choice, I'd probably go to something else that feels like it respects users.
Backwards compatibility is one of the selling points. Another is the UI, which has remained basically the same since Windows 95. They learned that lesson when they changed it for Windows 8, and immediately changed it back.
I stopped using it on my computers at windows xp (and windows 98 was the last version I spent any real time in) so the changes in newer versions of Windows are a lot more obvious (and confusing) to me. Here are just a few of the bigger ones:
The control panel has moved around a few times since then (and is way harder to find, I have to use the search feature to get to it) and now they have some windows RT replacement that I have no idea how to navigate. The user home folder (and even C:\) is harder to find now. Everything pushes you to use "one drive" which has non intuitive synchronization/corruption/compatibility/path behavior. The start menu is completely different and has ads now. I have no fucking idea how user creation works anymore but I think it's tied to some online service.
Linux has had a far more consistent UI. Xterm hasn't changed (as far as I can tell) since the version I had on my laptop when I was 12. Neither has FVWM (although they stopped calling it "FVWM2.") Bash has "deprecated" aliases but they still work. I guess GNOME has iterated on its shell a few times but it's not like Microsoft hasn't screwed around with explorer. Also caja is still there and behaves the same (does anyone actually use GNOME unironically anyway?)
Office has changed even more dramatically (which makes me wonder why people still stick with it, at this point libre office writer has fewer UI changes from word 98 than Microsoft word.)
If you want an example of "UI doesn't change much over half a century" look at coreutils. They make changes but people really freak out when they do because there's little difference between the UI and what is essentially an ad-hoc API.
I'm talking about the UI for corporate users, who don't use the control panel. They want a start menu that works, a bar along the bottom to get their apps, close/min/max buttons in the same place, a clock that shows the calendar when you click it, etc.
My experience is opposite, Linux desktop is garbage compared to Windows desktop, and I’m not talking about features or looks or anything like that, but core requirements like response time, handling of rouge processes (that can freeze your desktop), login/logout, virtual memory problems, scheduling of processes, memory leaks. List goes on.
I need to tinker with my Linux machine constantly. Linux is great a server, but desktop? Still a long way to go.
To some extend I don't disagree. In general I feel like response time is better on Linux, but I don't have any numbers.
Mostly I've given up on the desktop environments on Linux, not due to response times, but because I don't want to tinker with it, I just want to use the computer. My current home setup is just an OpenBSD machine with CWM, it manages windows and launches programs and that's all it needs to do.
I don't know what I do wrong (i.e. right) but these things just doesn't seem to happen to me.
Then again I use KDE Neon and only make to make software, browse the web and occasionally play CSGO.
My biggest annoyance is how everytime I boot either my Windows partition (weeks between) or KDE partition (anytime really) there's a huge stash of upgrades waiting for me.
KDE is the one desktop environment I haven't experimented with enough to give a fair judgement. I need to put that next on my list.
Updates on Windows is usually more annoying on Windows than on Linux because of the forced updates with reboots, in that regard Windows is more garbage than Linux.
I guess I must be pretty lucky or something. I've been running Debian testing with Gnome on a Lenovo X1C5 for a couple of years now (it used to run Win10 but I got fed up with the forced updates) and the only time I find myself tinkering with it is when I get bored and want to change things up. Go figure.
I have used the same Gentoo/i3wm setup for ... a lot of years. Probably 10 or more by now.
I guess we maybe changed the way package-installer (portage) profiles worked during that time, and I sort of recall something about how repos are configured being changed. Those were each at most a command or two run as root.
Otherwise, I don't really get a sense that I have had to do any tinkering other than what I specifically want to mess around with something, as you say.
E.g., I changed from grub to efi-stub booting, which was fun and took a bit of work. I tried out Conky and some different status bars (cool, but not really my thing).
Now, i3 is really pretty barebones. I can imagine that some desktop environments maybe require more tinkering than i3, but if so then that's not really Linux's fault, is it.
Right. You can’t blame the horse for having a poorly fitted saddle. Unfortunately sometimes we get horse saddles that try to copy elephant saddles, and that’s when we get something like Gnome.
I'm on dev release channel (in order to use WSLg, very nice so far), and it has an updated look&feel in Explorer.
But Settings / Control Panel needs to be reversed. Just got a new computer, and having to look both places to find a knob is annoying. Settings is also too dumbed down and spacey to be usable. And actually slow to navigate.
But don't think they should or can break compatability as you suggest. Firstly, not having shit break every update is one reason I moved off MacOS and back to Win. But also for MS's part that if thing breaks and they have to do big upgrades, users might as well look elsewhere. People keep using Win because it works for them and the incentive to switch vs cost is low.
I sooo agree with this, they f'd the UI up so much with the switch to whatever app crap they were thinking of when they "improved" the control panel. 95-Win7 made sense and I could find stuff even with the consolidation that happened in XP & Vista (well metro). The 8/10 consolidation is crap and hides features in a needless button maze.
Yes, some sort of containerization could solve all this. Drawback is size on disk, however with today’s disks like m2 I don’t think it should be a major problem.
I seem to recall that was sort of the plan starting with Windows 8.
On the inside the "metro" core runtime (can't remember the name) was not depending on the historical win32 parts, and on the GUI the win32-dependent desktop appeared as a tile i.e conceptually an "app" on the start screen. Like a nested VM, but at the library level. I could have readily imagined multiple desktop tiles being there, one per legacy Windows version, to maintain some sort of perfect backwards compatibility without emulation, Wine style?
You could see a path forward where the start screen OS was not a layer atop the desktop, but the main thing with the legacy desktop boxed inside, possibly in the future only installed when a legacy app required it, and pruneable anytime. It made me think of the Win95+ transition ultimately leading to the MSDOS dependency being turned around and dropped.
Of course mistakes were made, and people cried foul about losing the desktop, so back to the desktop it went, with its legacy cruft.
Maybe indeed it can still move forward that way by using a more "Fusion" like approach where foreign legacy Windows appear on the desktop (like WSL2+GUI).
Windows faces a dilemma - its most valuable users value it for its deep and broad compatibility, but this compatibility is provided by geological layers of legacy cruft, which puts a severe drag on modernization initiatives. Case in point: the bizarre inconsistency of the current Windows 10 UX.
The way forward is probably to partition off those compatibility layers somehow, and sell support for them as add-on services, with a price escalator balanced so as to encourage migration to modern APIs, while presenting no cliff-edge that would push customers to competitors.
I’m thinking wider than just UI, but in this example, I think the main reason the “old” UI is perceived as superior is because it was, once, finished, whereas the “new” UI is a perpetual work-in-progress mish-mash of different generations’ worth of UI teams’ competing ideas clashing with each other and creating a total mess.
I think the “new” UI is fixable, but the way to do this is to finish it and make it a coherent system, not to add another layer while keeping the old layers hanging around.
One huge glaring example of this is how "settings" are handled in Windows. Half the stuff is in fancy new "metro" UIs that are gimped and practically useless, all web-styled with padding everywhere and choked full of "UX". Whilst we still have the functional and familiar Control Panel. The day they remove the Control Panel is the day I switch to Linux permanently.
The new Settings is indeed a trainwreck. They thought they could get away with a coat of paint, leaving “advanced” stuff to the old UI (too much shit to do, bro!), but the result is an annoying and unusable mishmash.
They would have spent their time better by reskinning win32.
The question was rather, what would the future of Windows have looked liked with Windows 10x. I'd say pretty bleak. If Win32 support wasn't perfect at launch (and AFAIK it was planned to not even support Win32 initially) it would be dead on arrival, everybody would stick with their "proper" Windows version instead of upgrading. New buyers would wonder why their new Windows PC doesn't run any games and either return the PC or "downgrade" to Win10.
The whole idea didn't make any sense from the start. There are better devices and operating system options out there that don't run Windows applications. Why should I pick Win10X?
Two things that annoyed me about this article are that that 1. A "Modern OS" is an OS that is vendor locked to hell. 2. A graphical re-work is comparable to actual features.
The person who wrote this article seems to think that progress in OSs taking control away from the users. I'm not a big fan of Windows, but I give it a lot of respect for compatibility and allowing users to modify it to some extent.
This could be my command line bias coming through. But how does a graphical rework and new icons compare to real-world features?
WSL1 and WSL2 are both impactful tools added to the Windows desktop. While reading this article would seem to have you believe that a graphical re-fresh is the most cutting edge feature that Microsoft has for Windows.
I am not a fan of the Windows Operating system, but this article seems to criticize and ignore everything I think it's doing right while ignoring everything that it's doing wrong. Like spying on users or not supporting better file systems.
Yes, as for the OS-es that the author calls "modern", the user is not really the owner of the device - the vendor is. Fortunately there are still many people who understand the value of unlocked platforms.
I agree with you. He even mentions no direct access to file system. IMO direct access to file system is one of the main things differentiating operating systems from content consumption devices such as mobile OSs.
I do see value in having applications compartmentalized (which is something the Linux world is moving towards too with Flatpak and Snappy). But the user should have full access to it, which is is not the case on mobile.
I have to admit this is the first time ever I’ve heard about Windows 10X.
Reading about it, it seems ambitious but also kinda risky or unrealistic to pull of, much in the same way Windows 8 with the “modern” apps failed to garner developers and reboot the Windows developer-platform and ecosystem.
Maybe Microsoft concluded the same and decided to can it before investing too much in it?
78 comments
[ 3.3 ms ] story [ 183 ms ] threadWindows can be the Android to iOS. Ultimately, most desktops and laptops run Windows, and if Microsoft is able to make it competitive on pricing (#1!), battery life, and performance, it's hard to imagine Chrome OS or the single user iPad seriously eroding Windows.
Most people in the world aren't affluent like the average HN reader. Their biggest consideration when buying a computing device is value, and Windows PC + ARM can win here.
Imagine a $499 laptop with a solid M1-level processor, a nice IPS display, good keyboard, 10+ hours battery life, and Windows 10.
That's serious competitor to both Chrome OS devices and Mac devices.
If Apple ported iOS to x64, I wonder what the battery life on a device would be like compared to ARM?
Of course, one thing Intel and AMD can't catch up on is pure transitor size. Apple has TSMC 5nm, Intel is on the equivalent of TSMC 7nm (Intel 10nm), and so is AMD.
> That's serious competitor to both Chrome OS devices and Mac devices.
$500 is a tough sell vs a $200 Chromebook if Chrome OS does what you need.
Since using Linux all rant she had "aaaaarrrgh, Windows again updating and requires restart!" disappeared. And performance just rocket up agains Windows.
I think, importance of native desktop apps will fade away and Windows can lose this advantage.
But this week I bought new Dell Vostro with Ubuntu 20.04 preinstalled. Although Ubuntu gave me some unpleasant impressions (controversial UI decisions, buggy as hell (can't imagine they call it LTS)) I gave it 2nd chance with fresh install of 21.04.
Anywhere in the world you will find business laptops running windows and the primary work will be done in outlook, excel, word, and PowerPoint.
I think this must be one option that is (or was) on table if they are also thinking about rebuilding whole windows.
They may not have the resources to make everything work seamlessly but that's not a problem for Microsoft.
Problem here is not the NT Kernel, the problem is user space.
So building everything on top of the Linux kernel would end up in the exact same position as before because the need of backwards compatibility, this would cost money with no benefit.
And the thing is they have already integrated Linux with WSL2 and can now run Linux desktop applications anyway.
Rebuilding Windows seems unlikely because there is no money in building operating systems, meaning for every operating system you build you need to maintain it indefinitely without getting paid for it.
What Microsoft need to do is to find a new market for their products. With a new market you can package Windows (or whatever) to fit that market specifically.
Windows used to be The platform you would use to run the apps developed with Microsoft technologies. Not sure about trends at large, but personally I would likely just go with Linux if I was creating something new with .net.
Last year the rumours said Microsoft was planning to launch Windows 10X without immediate support for Win32.
https://www.thurrott.com/windows/windows-10x/238014/windows-...
Just as a start.
Device driver universe on Windows may be a blessing or a curse, but entire industries have labored to make it sorta work, and to follow troubleshooting scripts (tech support) when it doesn't.
Aside from that, the leading edge graphics card drivers are their own, multiple OS universes. With their own collection of hacks for specific games.
I'm not familiar with the current state of high end gaming on Linux, but I believe Windows is still the platform for big game titles.
When there is a volunteer army that can go after Meltdown and Spectre orders of magnitude faster than the corporate stable of developers, the business case seems to make itself.
Which might be among the reasons Mr.Softy has been a Platinum Parter of the Linux Foundation the last five years => https://www.itpro.com/strategy/27583/microsoft-joins-linux-f...
At this point I also don't care if it's at the cost of backward compatibility, but they also need to fix the response times in the Windows UI. Given that high end games play very well on Windows, the OS is clearly able to process inputs fast enough. Yet for some reason when you're navigating the Windows desktop, everything seems to react pretty slowly.
Windows 2000 was the last version I was able to use without having my mouse hand cramp up. Something in Windows causes me to tense up my left hand (my mouse hand). It doesn't happen in macOS and it doesn't happen on Linux (Gnome/KDE/XFCE).
Is this why every app on Win10 seems to have a different title bar color? And even different active/inactive behavior?
And how do I move a browser window, now that my 8-15 browser tabs consume all of the title bar area?
Keyboard shortcuts?
Alt space brings up the "restore, move, size, minimise, maximise, close" menu.
Windows Key + cursor keys will snap the window to different places.
This is the fault of browsers, not the OS. I think chrome was the first to do away with the actual title bar and put tabs there, then everyone followed. You can change this in the settings I think, there should be a small area you can drag between the minimize button and the last tab (on FF at least), and then you can always hit the "Alt" button to bring up the title bar you can then grab.
Now every application and OS component does its own thing however and whenever it likes, and the whole experience becomes impossible to learn other than as a giant list of special cases and exceptions, which will change with the next round of patches.
This is why I can’t explain to my mom how to use her computer any more, and generally tell her to use her phone instead.
Yeah, no. This is the heart and soul of MS. Apple breaks stuff all the damn time. MicroSoft keeps things working forever. Some people pay for upgrades. Some pay so they never have to upgrade. Consumers upgrade, and apple makes a lot of money. Businesses don't upgrade, and pay for the privilege.
Kinda sucks, but preserving how we got here is a huge deal. 20 year old systems are core to frighteningly important systems.
Sure, make another, good luck handling the zillion battle tested edge cases. Windows may be horrible now, but ancient code still works. Your hand cramp is less expensive than a rewrite.
(this might be a little aggressive. apologies, this is kinda a hot take)
Microsoft can continue down the current path and Windows will still be hugely successful and widely used. It's will just be terrible and continue to get worse. Most people and businesses will just live with it. I mean if Windows is just the thing that launches Steam, then you don't really care. If it's just the shit SAP/Excel/what-ever launcher your IT department forces you to use, then fine. Windows just won't be a smooth experience or something people enjoy using.
They've done it from time to time. The benefits of NT outweighed the loss of tools that required the windows 9x models; although not enough for many to switch to 2000; XP made some concessions to seal the deal. The benefits of 64-bit windows are generally worth the loss of 16-bit applications (although winevdm shows that they could have made that work).
But, I haven't seen much from Microsoft lately that leads me to think they'd be able to pull something like that off. If I had to make a choice, I'd probably go to something else that feels like it respects users.
Backward compatibility is the entire selling point of windows.
>Windows 2000 was the last version I was able to use without having my mouse hand cramp up.
Hardware accelerated graphics? I've noticed these tend to add a few frames of latency and have a similar reaction.
I stopped using it on my computers at windows xp (and windows 98 was the last version I spent any real time in) so the changes in newer versions of Windows are a lot more obvious (and confusing) to me. Here are just a few of the bigger ones:
The control panel has moved around a few times since then (and is way harder to find, I have to use the search feature to get to it) and now they have some windows RT replacement that I have no idea how to navigate. The user home folder (and even C:\) is harder to find now. Everything pushes you to use "one drive" which has non intuitive synchronization/corruption/compatibility/path behavior. The start menu is completely different and has ads now. I have no fucking idea how user creation works anymore but I think it's tied to some online service.
Linux has had a far more consistent UI. Xterm hasn't changed (as far as I can tell) since the version I had on my laptop when I was 12. Neither has FVWM (although they stopped calling it "FVWM2.") Bash has "deprecated" aliases but they still work. I guess GNOME has iterated on its shell a few times but it's not like Microsoft hasn't screwed around with explorer. Also caja is still there and behaves the same (does anyone actually use GNOME unironically anyway?)
Office has changed even more dramatically (which makes me wonder why people still stick with it, at this point libre office writer has fewer UI changes from word 98 than Microsoft word.)
If you want an example of "UI doesn't change much over half a century" look at coreutils. They make changes but people really freak out when they do because there's little difference between the UI and what is essentially an ad-hoc API.
I need to tinker with my Linux machine constantly. Linux is great a server, but desktop? Still a long way to go.
Mostly I've given up on the desktop environments on Linux, not due to response times, but because I don't want to tinker with it, I just want to use the computer. My current home setup is just an OpenBSD machine with CWM, it manages windows and launches programs and that's all it needs to do.
Then again I use KDE Neon and only make to make software, browse the web and occasionally play CSGO.
My biggest annoyance is how everytime I boot either my Windows partition (weeks between) or KDE partition (anytime really) there's a huge stash of upgrades waiting for me.
Updates on Windows is usually more annoying on Windows than on Linux because of the forced updates with reboots, in that regard Windows is more garbage than Linux.
I guess we maybe changed the way package-installer (portage) profiles worked during that time, and I sort of recall something about how repos are configured being changed. Those were each at most a command or two run as root.
Otherwise, I don't really get a sense that I have had to do any tinkering other than what I specifically want to mess around with something, as you say.
E.g., I changed from grub to efi-stub booting, which was fun and took a bit of work. I tried out Conky and some different status bars (cool, but not really my thing).
Now, i3 is really pretty barebones. I can imagine that some desktop environments maybe require more tinkering than i3, but if so then that's not really Linux's fault, is it.
But Settings / Control Panel needs to be reversed. Just got a new computer, and having to look both places to find a knob is annoying. Settings is also too dumbed down and spacey to be usable. And actually slow to navigate.
But don't think they should or can break compatability as you suggest. Firstly, not having shit break every update is one reason I moved off MacOS and back to Win. But also for MS's part that if thing breaks and they have to do big upgrades, users might as well look elsewhere. People keep using Win because it works for them and the incentive to switch vs cost is low.
So why not bring the new Windows Core OS and have backward compatibility with integrated VM?
On the inside the "metro" core runtime (can't remember the name) was not depending on the historical win32 parts, and on the GUI the win32-dependent desktop appeared as a tile i.e conceptually an "app" on the start screen. Like a nested VM, but at the library level. I could have readily imagined multiple desktop tiles being there, one per legacy Windows version, to maintain some sort of perfect backwards compatibility without emulation, Wine style?
You could see a path forward where the start screen OS was not a layer atop the desktop, but the main thing with the legacy desktop boxed inside, possibly in the future only installed when a legacy app required it, and pruneable anytime. It made me think of the Win95+ transition ultimately leading to the MSDOS dependency being turned around and dropped.
Of course mistakes were made, and people cried foul about losing the desktop, so back to the desktop it went, with its legacy cruft.
Maybe indeed it can still move forward that way by using a more "Fusion" like approach where foreign legacy Windows appear on the desktop (like WSL2+GUI).
I guess the future of Windows continues to be the Win32 API.
[1] https://www.youtube.com/watch?v=ztrmrIlgbIc
The way forward is probably to partition off those compatibility layers somehow, and sell support for them as add-on services, with a price escalator balanced so as to encourage migration to modern APIs, while presenting no cliff-edge that would push customers to competitors.
I think the “new” UI is fixable, but the way to do this is to finish it and make it a coherent system, not to add another layer while keeping the old layers hanging around.
They would have spent their time better by reskinning win32.
That's clear to everyone except the author of the article. If Windows breaks backward compatibility, what's the reason to use it?
The whole idea didn't make any sense from the start. There are better devices and operating system options out there that don't run Windows applications. Why should I pick Win10X?
Windows HPC Server 2008 was decent, does something like that still exist?
While you can buy individual licenses from companies like CDW, some just get it via file-sharing.
The person who wrote this article seems to think that progress in OSs taking control away from the users. I'm not a big fan of Windows, but I give it a lot of respect for compatibility and allowing users to modify it to some extent.
This could be my command line bias coming through. But how does a graphical rework and new icons compare to real-world features? WSL1 and WSL2 are both impactful tools added to the Windows desktop. While reading this article would seem to have you believe that a graphical re-fresh is the most cutting edge feature that Microsoft has for Windows.
I am not a fan of the Windows Operating system, but this article seems to criticize and ignore everything I think it's doing right while ignoring everything that it's doing wrong. Like spying on users or not supporting better file systems.
How is Flatpack/Snappy approach helping me?
Reading about it, it seems ambitious but also kinda risky or unrealistic to pull of, much in the same way Windows 8 with the “modern” apps failed to garner developers and reboot the Windows developer-platform and ecosystem.
Maybe Microsoft concluded the same and decided to can it before investing too much in it?