6 ms·

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. Ther

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.

Santosh83 9y ago
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.
- contextfree 9y ago
  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.
  - garganzol 9y ago
    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).
    - pjc50 9y ago
      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.
      - garganzol 9y ago
        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.
garganzol 9y ago
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.
quincunx 9y ago
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.
- watmough 9y ago
  Yes, look at the bits used by Microsoft Office ... Win32. 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.
  - DougN7 9y ago
    Office 2016 really sucks compares to 2013. It feels like a browser app. Is it UWP?
- gt_ 9y ago
  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.
vim_wannabe 9y ago
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.
- WorldMaker 9y ago
  You can open multiple instances of Calculator. Paint 3D running single-instance only is a design choice.
  - megaman22 9y ago
    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.
wila 9y ago
You mean they want to pull a nokia? The legacy is their crown jewels.
pjc50 9y ago
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".
- pjmlp 9y ago
  Easy, unless one has a salary of a first world country, Macs are out of question. 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.
jxdxbx 9y ago
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.
- Santosh83 9y ago
  > 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?
robomartin 9y ago
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.
usrusr 9y ago
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.

Santosh83 9y ago

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.

contextfree 9y ago
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.
- garganzol 9y ago
  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).
  - pjc50 9y ago
    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.
    - garganzol 9y ago
      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.

garganzol 9y ago

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.

quincunx 9y ago

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.

watmough 9y ago
Yes, look at the bits used by Microsoft Office ... Win32. 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.
- DougN7 9y ago
  Office 2016 really sucks compares to 2013. It feels like a browser app. Is it UWP?
gt_ 9y ago
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.

vim_wannabe 9y ago

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.

WorldMaker 9y ago
You can open multiple instances of Calculator. Paint 3D running single-instance only is a design choice.
- megaman22 9y ago
  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.

wila 9y ago

You mean they want to pull a nokia? The legacy is their crown jewels.

pjc50 9y ago

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".

pjmlp 9y ago
Easy, unless one has a salary of a first world country, Macs are out of question. 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.

jxdxbx 9y ago

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.

Santosh83 9y ago
> 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?

robomartin 9y ago

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.

usrusr 9y ago

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.