Linked by on Tue 27th May 2008 15:00 UTC
Windows So far, Microsoft has been very tight-lipped about Windows 7, carefully trying to prevent another Longhorn PR disaster where the company promised the heavens and more for Longhorn, but in the end ditched Longhorn to make way for Vista. Chris Flores (Windows Client Communications Team) as well as Steven Sinofsky, has broken the silence a little bit to talk about Windows 7. In addition, it is believed Windows 7 will make its first official debut at the D6 All Things Digital conference today, during a keynote held by Steve Ballmer and Bill Gates.
Thread beginning with comment 315801
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[2]: Another Missed Opportunity
by phoehne on Tue 27th May 2008 17:43 UTC in reply to "RE: Another Missed Opportunity"
phoehne
Member since:
2006-08-26

Okay, you're clueless. What's not working with the Windows kernel is that it does require massive amounts of memory to run the basic O/S. In fact, a recent posting (can't find it right now) about an Intel engineer that benchmarked various tasks under different versions of Windows and Office found they were slower on Vista (on modern hardware) than on Windows 2000 on old hardware. That's right. You're upgrading to Vista to actually run slower on brand-spanking new hardware.

Windows suffers from another problem, which is 20+ years of cruft. Last time I was coding on Windows there were 3 different C API's for memory management. Only one was current but all three were supported so code going back to Windows 3.1 had a shot of running. It's time Microsoft dumped some of those API's the way Apple did with the Carbon migration prior to OS X. Why dump those old API's? Because they lock in old, out-dated concepts and practices, which can interfere with new kernel development. Even if they're just wrappers around the new API's, they should go the way of the do-do.

As recently as early releases of XP there are way too many reports of user programs causing kernel crashes. (And no, I'm not talking about buggy device drivers). These are instances of user programs passing bad parameters back to the Win32 functions causing Windows to BSOD. That should happen 0 times.

I've written software on Windows at various stages in my career since Windows NT 4 so that people can run businesses and make money. I've had the dubious pleasure of writing OCX and Active X controls, COM+ components, all the way through .NET GUI's. What was modern in the Windows NT 4 sense (circa 1997) is now no longer modern. It needs a facelift. It needs a cleanup. It needs to live well in a reasonable environment. It needs that because otherwise it becomes increasingly a dead weight instead of a useful tool.

Windows is not the best kernel available by any stretch of the imagination. For one thing, it lacks real time capabilities (which Windows CE ironically provides). The thread scheduler is counter intuitive. The filesystem is slow (especially over network shares). The part I especially love is the random waiting while network browsing returns information about computers, printers and networks. Windows isn't evil. It does need work. It would be nice to see Microsoft stop dicking around with Yahoo or Zune and focus on it's core competency, Operating Systems before it begins to look like it's core incompetency.

Edited 2008-05-27 17:50 UTC

Reply Parent Score: 14

Hae-Yu Member since:
2006-01-12

The "cruft" guarantees Windows the largest market share. Because so many applications and drivers run on Windows, people use it. Because so many people use it, applications and drivers will almost always target Windows.

Not just on the desktop but in support of things like cash registers, auto diagnostics, programming LMRs (radios), medical systems, ... These systems evolve slower than the desktop and there are far more of these than desktops. Their equipment investment is in decades. You just can't kick all those important users in the nuts and say "upgrade every business process end-to-end every few years to support our new OS." These customers just won't upgrade. MS doesn't get revenue if users don't upgrade. Therefore the "cruft."

RToS - Windows isn't running your brakes. RToSes have a different application and philosophy than GP OSes like Windows or Linux. That's why Windows CE and RTLinux are real-time and not the main desktop or server OSes.

Reply Parent Score: 4

phoehne Member since:
2006-08-26

Actually real time behaviors are becoming important for things outside the traditional RTOS. For example, it's can be used in simulations and games to make for a more responsive, more realistic experience. In the financial community soft real time behaviors are becoming part of trading systems. In addition, as embedded systems just become more like little (SoC) versions of desktop systems, a stripped down desktop version of the O/S is nice to use because it simplifies development. I just pointed out real time scheduling as one thing the standard Windows kernel lacks that's available in a number of other kernels. For example, both Linux and Solaris come with real time schedulers (not just a special RTLinux).

But I don't see your connection between the OS that's in your cash register and your desktop OS and why you'd have to upgrade your cash register or your business processes if you upgrade your desktop. For example, say your cash register runs Windows CE, or your service station pump head runs Windows XP embedded, it has nothing to do with your desktop OS. What's more, if you do build a medical imager with Windows XP embedded, you probably will not ever upgrade it to Windows Vista, anyway. You don't need to keep the entire kernel the same to provide for backwards compatibility. As many people have pointed out, you can run the legacy Windows version in a virtual machine. That gives you the freedom to build a solid, next generation kernel while satisfying your user needs to run old code. Their current strategy ties them to old code while constraining their ability to modernize their key product.

There are two ways of looking at the present Windows situation. The first is the odd/even phenomena. Windows 3.1 - successful. Windows 3.5.1 Not so. Windows NT 4 - successful. Windows 2000 not so. Windows XP successful. Windows Vista not so. That may mean they get all their ducks in a row and Windows 7 continues the pattern by being successful. The pessimists (of which I am a part) look and see real issues with where Windows is and where it's headed. Microsoft could turn into the GM of operating systems, producing generations of mediocre products, slowly loosing just a little bit of market share every year. That kind of slow grind to mediocrity will be very expensive for end users.

Edited 2008-05-27 20:26 UTC

Reply Parent Score: 3

red_devel Member since:
2006-03-30

This isn't a missed opportunity. The opportunity to "pull an OSX" was before they released Vista. But they did, so now it would make no (business) sense to start from scratch and break compatibility.

"We're very clear that drivers and software that work on Windows Vista are going to work really well on Windows 7; in fact, they'll work the same. We're going to not introduce additional compatibilities, particularly in the driver model."

Doesn't anyone else see why this is a great business move for them?? Vista adoption is slow, drivers have been lacking, people are still developing for XP, etc...So, you say, "Everything that will work on Vista will work perfectly on Windows 7", and now companies see a better reason to develop applications, and more importantly drivers, for Windows Vista. Kill the, "we'll just hold out for Windows 7" mentality, and in the process, give people a big motivation to adopt Vista.

And lets face it folks, no matter what they say, Windows 7 will NOT be out in early 2010 or whatever pipe dream they are pushing now.

Reply Parent Score: 4

rockwell Member since:
2005-09-13

@phoehne: I'm a long-time Windows user and still think it's the best desktop OS (i actually like Vista) ... but you make some great points.

If UltraEdit ran natively in Linux ... then I'd seriously consider making the swtich.

Edited 2008-05-29 20:41 UTC

Reply Parent Score: 2