> Reverse the situation: A bunch of critical systems which run Linux 2.4 are being compromised by cyber criminals via a kernel exploit. You're going to argue that it's Linus' fault for providing such a great kernel and not supporting it forever?
I think there's a simpler way to think about this. Windows XP still functions for these businesses, and many of them would gladly pay an ongoing subscription fee for support. Sure, Microsoft wants to expand, but I don't see why that has to happen exclusive of supporting their well-loved early 2000s release. I run software which hasn't changed substantially for more than 20 years on a daily basis, and I don't see why Microsoft has to pretend that that's not a use case.
> > Microsoft are the ones who worked so hard to make sure that software for Windows isn't compatible with not-Windows, creating all their own alternatives to POSIX, OpenGL, Open Firmware and everything else so that it's as difficult as possible for software compatible with Windows XP to be compatible with any Unix or Linux, leaving the user out in the cold if it also isn't compatible with Vista or later.
> And so does every other operating system on the planet. You cannot take a MacOS binary and run it on Linux. And until very recently you couldn't take a Linux binary and run it on Windows.
Open Firmware, POSIX, and OpenGL are explicitly not related in any way to binary compatibility. You can run unmodified OpenGL programs (with some abstraction) on Linux and OS X out of the box, and with some fiddling, on Windows as well.
It's clear that this is an argument that Microsoft has refused to support standard APIs, not standard ABIs.
How many apps are there really that run on the win32 API of Windows XP, but not on the one of Windows 7/8.1/10?
Not saying there aren't any, but nearly everything I've seen fail has been coupled to hardware drivers, where other platforms are just as bad, or other outdated APIs, e.g. ancient versions/insecure configurations of Java. This is a big problem in IT, but I don't think it's fair to blame Microsoft more than others for it. Maybe for making as much a mess of Vista as they did, but 7 was still available more than early enough for slow migration.
(And at least the NHS had an extended support contract for Windows XP, but ended that at some point, despite (as far as I know) it still being available from MS for other big customers)
I think there's a simpler way to think about this. Windows XP still functions for these businesses, and many of them would gladly pay an ongoing subscription fee for support. Sure, Microsoft wants to expand, but I don't see why that has to happen exclusive of supporting their well-loved early 2000s release. I run software which hasn't changed substantially for more than 20 years on a daily basis, and I don't see why Microsoft has to pretend that that's not a use case.
> > Microsoft are the ones who worked so hard to make sure that software for Windows isn't compatible with not-Windows, creating all their own alternatives to POSIX, OpenGL, Open Firmware and everything else so that it's as difficult as possible for software compatible with Windows XP to be compatible with any Unix or Linux, leaving the user out in the cold if it also isn't compatible with Vista or later.
> And so does every other operating system on the planet. You cannot take a MacOS binary and run it on Linux. And until very recently you couldn't take a Linux binary and run it on Windows.
Open Firmware, POSIX, and OpenGL are explicitly not related in any way to binary compatibility. You can run unmodified OpenGL programs (with some abstraction) on Linux and OS X out of the box, and with some fiddling, on Windows as well.
It's clear that this is an argument that Microsoft has refused to support standard APIs, not standard ABIs.