Linked by Thom Holwerda on Fri 2nd Dec 2011 18:41 UTC
Windows Windows 8 will have both the new Metro-style applications and user interface and the traditional Windows 7 desktop for legacy applications, which kind of runs like an application. Since legacy applications have to be recompiled to run on ARM anyway, it's always been a bit unclear if the ARM version of Windows 8 would include the legacy desktop at all - even Microsoft itself confirmed it wasn't sure yet. Microsoft bloggers Mary-Jo Foley and Paul Thurrot have fresh rumours that Microsoft has now made the decision to remove the legacy desktop from the ARM version.
Thread beginning with comment 498769
To read all comments associated with this story, please click here.
Not necessarily bad but...
by Kalessin on Fri 2nd Dec 2011 22:46 UTC
Kalessin
Member since:
2007-01-18

Not maintaining legacy x86-centric APIs on ARM makes a lot of sense, given that the likelihood of an application which runs on x86 Windows being recompiled on ARM working properly is not all that great. Endian issues alone could be fatal, let alone what other assumptions were made based on x86. And fixing all of those APIs so that they work as they're supposed to on ARM? That's likely to be a nightmare, and the API functions' implementations are even more likely to be x86-centric than the code using them, since they're generally lower level. So, it makes a lot of sense to just say that legacy apps won't work on ARM - especially when most ARM devices are going to be tablets or smartphones rather than PCs where the legacy applications would actually be used.

The question is whether they'll be forcing ARM to be metro-only. If that is where they're going with this, then that is bad. I do not want to see metro take over the desktop, given how tablet-centric and stripped down it is, and the fact that it's being pushed for the desktop is a really questionable decision IMHO, much as metro may be fantastic for tablets. And while ARM may be primarily for mobile devices at this point, there's no reason why it couldn't be used in desktops, so restricting Windows on ARM such that it can't use the classic desktop instead of metro would be bad.

So, as far as the APIs themselves go, this is arguably a very solid decision, but if this means that they intend to force Windows on ARM to be metro-only, then this is a horrible decision.

Reply Score: 2

RE: Not necessarily bad but...
by Moochman on Sat 3rd Dec 2011 00:48 in reply to "Not necessarily bad but..."
Moochman Member since:
2005-07-06

I think the ideal outcome is that over the next few years, there are no more "pure laptops" anymore, just "laptablets" (laptops with touchscreens/tablets with keyboards) and tablets. If that's the case it shouldn't be any great loss to have Metro as the sole Windows UI on ARM.

Reply Parent Score: 2

Drumhellar Member since:
2005-07-12

I wouldn't worry about it. Win32 was designed to be fairly platform-agnostic (Remember; NT was on multiple architectures in it's early days). Endianness shouldn't be an issue, as ARM is little-endian by default (Though, capable of big-endian operation).

Plus, there is always .NET.

I think the main issue hinges on whether or not Microsoft expects manufacturers to attempt ARM desktops. If they don't expect any, there's no reason to support the legacy desktop, as touch-driven apps will be all users will want/experience.

Reply Parent Score: 2

lucas_maximus Member since:
2009-08-18

I think the main issue hinges on whether or not Microsoft expects manufacturers to attempt ARM desktops. If they don't expect any, there's no reason to support the legacy desktop, as touch-driven apps will be all users will want/experience.


This!

Completely agree. I personally would love a 12inch SoC Notebook running Windows and Visual Studio ... massive battery life and I could develop in the coffee shop (I have flexible working arrangements with my employer).

Reply Parent Score: 2

RE: Not necessarily bad but...
by jal_ on Mon 5th Dec 2011 09:24 in reply to "Not necessarily bad but..."
jal_ Member since:
2006-11-02

Endian issues alone could be fatal


Iirc, ARM has both big and little endian modes (selectable per process/thread). It'd be a bit stupid not to use the little endian mode for recompiled x86 applications.

Reply Parent Score: 2