To view parent comment, click here.
To read all comments associated with this story, please click here.
The key problem is, if you stay with XP and abandon Vista, you have to add Vista capabilities to XP and maintain all the 1000s of configurations kept inside of XP to preserve backwards compatibility.
In essense, XP would become incredibly bloated. Eventually, Microsoft has to abandon its backwards compatibility. It's not without precident. It gave up DOS, in favour of the less close to the metal NT-DOS. This resulted in thousands of DOS applications failing to work, but they gave DOS enough time to survive. Microsoft can't do this now, but they can clean up Win32 to get rid of all the backwards compatibility kludges and build it on a lean kernel. Then the Windows XP and Windows Vista APIs can be built, WINE-like
on top of the Win32-clean API.
This approach has several benefits:
(1) The Vista libraries would only need to be
loaded if you have Vista apps.
(2) The XP libraries would only need to be
loaded if you have Vista apps.
(3) If you only have Win32-clean apps, your machine would be a lot faster and your kernel will be a lot smaller (idealfor the embedded market).
(4) Porting apps from XP and Vista would simply be a matter of replacing obsoleted APIs with new ones. Code analysis software can catch most of these cases.
I don't see the down-side, other than Microsoft possibly regaining its leadership (which is, Roman empire-like, slowly in decline).
X11 is a low level API, I'm not a Windows export but isn't this comparing Apple and Orange?
To compare 'identical purpose API' I think that it would be more fair to compare X11 to GDI and Win32 to Qt or GTK, no?




Member since:
2005-11-20
3) If Microsoft had a bit of vision and could swallow their pride, they could find a way out by going the Apple route...dump both XP and Vista and go for emulation -- none of this virtualization stuff that adds yet more bulk with little benefit -- virtualization should be the focus of the new cleaned up API. All future work is on a lean, cleaned up API over a stable kernel. They could go with a F/OSS kernel, but knowing Microsoft, buying QNX might be a better bet and would allow them to break into markets where they have no credibility.
I keep hearing this argument again and again, and I think it is flawed.
First, you assume that Microsoft problems are due to bad API. Well, more specifically, you seem to assume that XP had serious problems. Well, for business, not so much. XP was extremely successful product.
Now the for the API. Well, Win32 is not great, but if you compare it to e.g. what we have in X11, it is still a relatively good API.
Anyway, the most important problem is that this is API used by majority of world's software. Do you think that it makes a sense for business to rewrite everything? And if they do, would not be better for them to consider abandoning Microsoft altogether and go different path? (either OSX or FOSS).
Of course, everybody touts virtualization, but that will just make existing software second-class citizens.
The reality is that XP and Win32 API works quite well. Throwing everything away does not make sense.