Linked by snydeq on Tue 4th Jun 2013 01:46 UTC
Windows First looks at Windows 'Blue' have revealed an upgrade composed of cosmetic fixes, suggesting that Microsoft may be blowing its chance to turn the tide on Windows 8 blow back, and make good on its promise to truly 'rethink' Windows 8 with the release of Windows Blue. As a result, InfoWorld has issued an open letter to Microsoft to consider Windows 'Red' -- what InfoWorld is calling a 'serious plan' to fix the flaws of Windows 8, one that could rescue Microsoft's currently flagging promise to deliver a modern computing experience on both PCs and tablets.
Thread beginning with comment 563606
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[3]: I don't think so
by dpJudas on Tue 4th Jun 2013 14:50 UTC in reply to "RE[2]: I don't think so"
dpJudas
Member since:
2009-12-10

They certainly didn't rewrite the framework in one go. WinRT is in its infancy compared to Win32. They have a long way to go.


Correct, but the reason I do not consider WinRT a (full) rewrite is because the most important earlier Win32 APIs are still available. This means porting can be done gradually where newer code can use the new better API, while older code can coexist with it.

That WinRT isn't supported in any form (i.e. not even the System namespace in C++/CX) on Vista and Windows 7 somewhat ruins the possibility to take advantage of their new APIs until Windows 8+ becomes the dominant version of Windows.

I agree that it is a problem for very large apps, but I don't think Microsoft expects those to be ported over particularly


That unfortuantely becomes a problem for Modern because apps from the "hwnd" world does not properly coexist with Modern desktop apps. The poor user experience switching between them is probably Windows 8's biggest problem.

Actually, Windows 8.1 allows a larger variety of snap states (especially on large, high resolution monitors where I think you can snap 4 or more apps at once). There is definitely some better multitasking support there.


What I do not see them address is the key problem: that if a user has 50% win32 apps, and 50% modern apps, then the user experience will be very poor. And it will stay this way for a decade unless they find a proper way to address it. Carbon to Cocoa took this long with slackers like Adobe. ;)

Reply Parent Score: 2

RE[4]: I don't think so
by Nelson on Tue 4th Jun 2013 15:41 in reply to "RE[3]: I don't think so"
Nelson Member since:
2005-11-29


Correct, but the reason I do not consider WinRT a (full) rewrite is because the most important earlier Win32 APIs are still available. This means porting can be done gradually where newer code can use the new better API, while older code can coexist with it.


I'm not particularly excited about the existing Win32 support in WinRT. At best it has some nice things, but at worse it's just a temporary shim until they write proper WinRT APIs.

If you have a reasonably complicated code base, the chances of you running into a roadblock with a restricted API is pretty high, especially given that most people used and abused Win32 for ways outside of what Microsoft thought.

Do any IPC and you're SOL on WinRT, any kind of dynamic execution is forbidden, threading stuff is a non starter, etc. Its pretty annoying to be honest.


That WinRT isn't supported in any form (i.e. not even the System namespace in C++/CX) on Vista and Windows 7 somewhat ruins the possibility to take advantage of their new APIs until Windows 8+ becomes the dominant version of Windows.


My problem with the back porting efforts are the architectural compromises that must be made. I still have nightmares about the absolute BS we had to deal with because Microsoft got pressured into backporting WPF to Windows XP. Layered Window performance fell off of a cliff, a bunch of things jumped into software rendering, ugh.

I just don't know if anyone would bother if it was just the non-UI bits backported. Its like, why should I use that vs the Win32 counterpart?

WinRT is also much more than the API/ABI its the entire catalog apparatus which includes with it the sandboxed RuntimeBroker and all of the changes to Windows that it brought.

I saw a Channel9 video about the architectural changes they had to make to deep subsystems to enable the app container model and it made my head spin.

Sometimes when you're on a tight release schedule its very hard to justify back porting something for limited impact, at best they'd port some of the Desktop capable WinRT APIs, which really isn't a lot.


That unfortuantely becomes a problem for Modern because apps from the "hwnd" world does not properly coexist with Modern desktop apps. The poor user experience switching between them is probably Windows 8's biggest problem.


Right, they have a bunch of work to do. You can see where the trend is going though, and its a multi-year thing they have.

Big suites like Adobe, IE, Office, etc. will take a while purely because the code bases are so incredibly massive and in some cases, WinRT is too immature.

However there isn't that big of a gap before large applications can be brought into the fold -- look at Visual Studio. Its a mixed mode C++/CLI app with a WPF front end. Thats not too much of a far shot from WinRT.

I can see maybe in the next major client release the WinRT platform maturing to the point where medium to large scale apps are possible.

In the meantime it will be a problem for people living in both worlds, but for people who purchase something like a Surface RT, a lot of their needs are suited by Metro apps which are getting increasingly better.

For the rest of us there's the Desktop, and with Blue its been made easier to live in both worlds.



What I do not see them address is the key problem: that if a user has 50% win32 apps, and 50% modern apps, then the user experience will be very poor. And it will stay this way for a decade unless they find a proper way to address it. Carbon to Cocoa took this long with slackers like Adobe. ;)


They can do a number of things to ease the transition for sure, some of which they're not currently doing. I think given feedback they'll continue to make more changes into post-Blue releases.

I'm just glad we're getting these changes yearly instead of every three years.

Reply Parent Score: 3

RE[5]: I don't think so
by dpJudas on Tue 4th Jun 2013 19:38 in reply to "RE[4]: I don't think so"
dpJudas Member since:
2009-12-10

If you have a reasonably complicated code base, the chances of you running into a roadblock with a restricted API is pretty high, especially given that most people used and abused Win32 for ways outside of what Microsoft thought.

Yes, even simple things like a LoadLibrary call is blocked. But once again I feel that it falls back on Microsoft, since if the porting work gets too extensive most companies decides to simply not do it until absolutely forced to. Microsoft cannot get a nice new consistent user interface if all the classic productivity apps keep their old UI.

My problem with the back porting efforts are the architectural compromises that must be made. I still have nightmares about the absolute BS we had to deal with because Microsoft got pressured into backporting WPF to Windows XP. Layered Window performance fell off of a cliff, a bunch of things jumped into software rendering, ugh.

With C++/CX the problem is that I cannot even instantiate a simple System.String. This part of their framework doesn't really require anything unique to Windows 8 and would demand no special compromises.

I just don't know if anyone would bother if it was just the non-UI bits backported. Its like, why should I use that vs the Win32 counterpart?

Mostly to support calling into Windows 8 APIs when they are available, and fallback to earlier code when they are not. Then slowly over the years retire old code as Windows versions get too old.

Now you can only do such a thing with defines and build configurations and different executables for desktop Windows 7 and Windows 8.

However there isn't that big of a gap before large applications can be brought into the fold -- look at Visual Studio. Its a mixed mode C++/CLI app with a WPF front end. Thats not too much of a far shot from WinRT.

Visual Studio still relies (although increasingly less so) on old Win32/hwnd parts that they are slowly rewriting as they revisit them for new updates anyway. This won't work for Windows 8 as you cannot have any hwnd parts in a Modern app.

I can see maybe in the next major client release the WinRT platform maturing to the point where medium to large scale apps are possible.

I agree, they could fix these problems if they are truly aware of them and feel they are important. Maybe they are even doing it as we speak. Personally I think they will continue to see little interest in porting applications to WinRT until they are resolved though.

Reply Parent Score: 2