Microsoft is dead: 1 year + 8 people to develop Vista shutdown (and it still sucks)

[Undertoad]

This blog entry from an ex-Windows developer describes how it took an entire year to develop the shutdown feature for Windows Vista.

I repeat, let it sink in... a YEAR to develop SHUTDOWN. SHUTDOWN. Where you tell your system to stop doing things. You might consider that in the old days, it was an "off" button and took seconds to operate...

You might also consider that orderly shutdown was, at one time, supposed to be the reason why Unix/Linux was incompatible with everyday computer users.

(This Joel on Software entry describes how and why Vista shutdown sucks, and how it really should operate.)

The tale might be tedious for people who haven't been developers, but most people will recognize the basic situation. The people who want and need to do real work are utterly preventing from doing so by A) layers of bureaucracy, and B) a technological tangle of code and developers, where things have become too complicated to manage successfully.

First he addresses the layers of bureaucracy:

So that nets us an estimate -- to pull a number out of the air -- of 24 people involved in this feature. Also each team was separated by 6 layers of management from the leads, so let's add them in too, giving us 24 + (6 * 3) + 1 (the shared manager) 43 total people with a voice in this feature. Twenty-four of them were connected sorta closely to the code, and of those twenty four there were exactly zero with final say in how the feature worked. Somewhere in those other 19 was somebody who did have final say but who that was I have no idea since when I left the team -- after a year -- there was still no decision about exactly how this feature would work.

Then, the software development tangle, which results in a delay of months upon months to integrate new code:

...there are far too many developers to access one central repository. So Windows has a tree of repositories: developers check in to the nodes, and periodically the changes in the nodes are integrated up one level in the hierarchy. At a different periodicity, changes are integrated down the tree from the root to the nodes. In Windows, the node I was working on was 4 levels removed from the root. The periodicity of integration decayed exponentially and unpredictably as you approached the root so it ended up that it took between 1 and 3 months for my code to get to the root node, and some multiple of that for it to reach the other nodes. It should be noted too that the only common ancestor that my team, the shell team, and the kernel team shared was the root.

So in addition to the above problems with decision-making, each team had no idea what the other team was actually doing until it had been done for weeks.

Any developer who reads this knows... the end is nigh. The Windows practice of throwing every last little thing into the OS (to maintain its dominance) has now led to its practical demise.

They're gonna have to start over again. Future versions of Windows for desktop computers will have to be revolutionary to compete, and will be incompatible with past Windows in the same way that Win 95 was incompatible with Win NT -- there will have to be a wholesale rewrite and and entirely new approach. If they're smart, the new approach will be modular and standards-based. If not, buy Apple stock and switch yourself to the free alternative.

[richlevy]

Originally Posted by Undertoad

If not, buy Apple stock and switch yourself to the free alternative.

Which is why Microsoft claims Linux violates it's patents without actually naming any.

[tw]

Originally Posted by Undertoad

This blog entry from an ex-Windows developer describes how it took an entire year to develop the shutdown feature for Windows Vista.

What it really took is a year to remove the head of OS development. Those layers of bureacracy are classic symptoms of top management that has little clue what the workers are doing / stuggling with.

Microsoft finally transferred the head of Office to fix Vista. But that is long after the above shutdown fiasco was history.

I have some concerns about Microsoft. Steve Balmer is an MBA. He can protect the status quo. But moving into new innovation, well, Vista took how long before he realized he had a problem?

[mbpark]

TW,

If you remember, they forked the original Windows development from Windows XP. It was only after they got (rightfully) raked over the coals that they switched to a Windows Server 2003 (SP1 I believe) code base to restart Vista development from.

This was also during/after the XP SP2 fiasco, where they were really under the gun to fix major problems in Windows XP that were overlooked. The big problem here was that people were getting infected with viruses and spyware at an epidemic pace, and there were so many major vulnerabilities in Windows XP that they were starting to look really bad, and Linux looked like it was going to get an even bigger hold. Some may argue that it did with the major uptick in Ubuntu installations.

They had to get SP2 out or else lose a very large amount of major accounts to other OS vendors. To do so, they had to re-task the Longhorn team to get this fix out, which when installed, literally was a new operating system. The patch was half the size of the original OS install, and should have been released instead of XP in the first place.

If they had not released SP2, several major customers, including several divisions of the US Government (DOD and their associated military branches), were more than prepared to call IBM Global Services and drop Citrix/Red Hat Enterprise solutions in place.

The problem wasn't that it took 8 years. The problem is that they were going down a path which was leading to failure (Longhorn), and they had to restart development in 2004 due to multiple reasons. They started full-on Longhorn development in the 2000-2001 time frame, right after XP RTM. If they had done what they should have, they would have hit a 2004 time frame. However, they let the problems fester with XP and Longhorn, which really screwed up a ton of other projects.

It takes a much longer time to restart development after a failed attempt than it does to do it right in the first place. Microsoft had a somewhat decent code base with Windows 2000 that the big customers liked. They made a bunch of kernel modifications, added a theming engine, and added more home-friendly advancements (UPnP (one of the first major XP bugs), Windows Movie Maker, System Restore, etc.) to it to create Windows XP.

Unfortunately, Windows 2000 was deployed mainly in large enterprises or in places where there were very competent IT people. When they loosed it upon the world with multiple features enabled that weren't in the 9x series (such as DCOM as LocalSystem, UPnP, File and Printer Sharing, and a host of other ports) by default, there were major problems.

The next thing you know, every malware maker saw the Windows NT variants for the easy targets they were. Both Windows 2000 and Windows XP got hit with a major array of spyware and viruses that caused havoc. The hardware and software vendors saw their support call costs skyrocket because of this, and so did the customers (especially certain very large enterprises).

This outcry got to the point where Steve Ballmer had to re-task the Longhorn team, and every other available developer, to create what we know as Windows XP Service Pack 2. This OS update literally was a new OS release, and was tested as such by the big customers. Many programs had to be updated to work properly with it, including Office XP, Symantec AV, and a whole host of third-party applications. This fix also extended to Windows Server 2003 through Windows Server 2003 SP1.

Since Longhorn was built upon the Windows XP codebase, they decided it would be better to carry the security enhancements through to Windows Vista through building on the Win2K3 SP1 code base. Vista development only really started in late 2004, including reportedly scrapping the complete code base, and starting anew. This always takes longer, no matter what you hear.

If you want to really know where MS is going, look at the Xbox 360 development. They bought Connectix to get Virtual PC for the PowerPC Mac, and restarted the PowerPC port of Windows NT, porting it to a simpler version of the G5 (a three-core variant of it) that happens to run all code under a "Hypervisor" that uses one of the cores. They also started up their own chip design division (I am not kidding here), and are seriously working on hardware-based DRM. The Xbox 360 is their first consumer release of it.

They also happen to have a research OS called Singularity that is designed for security, and is a ground-up new creation that happens to run Managed Code (the .NET code) really well. I would think that some of those concepts in Singularity made it to the Xbox 360. Some of them have already made it to the .NET framework, including the ability to lock out managed code that hasn't been cryptographically signed with a digital certificate from an approved code-signing authority (aka Verisign, Entrust, or one of the other big boys).

I really think that Microsoft is looking to run future versions of Windows in their own managed virtual machine, much like how Parallels Desktop 3.0 does on the Mac, where Windows code can run in "Coherence" mode, and looks like a Mac app, but is running under an isolated copy of Windows.

I also think they're going to push DRM on the hardware so you can't install other OSes, and you're locked into Windows. Typical MBA thinking here. I also think they're also going to try to get into the PC market with the next version of the Xbox, and offer a seriously-locked-down PC that runs "approved" apps really well, thereby cutting off the air supply to the rest of the market. I would not also be surprised to know that the PowerPC port of Office 2009 has started, since they've already had Office, SQL Server, and other apps running on IA64 (Itanium), Alpha, MIPS, PowerPC, x86-64, x86, and reportedly SPARC processors.

You're right. They've lost the way of innovating to make customers happy. They're looking at ways of removing your ability to do things and calling it innovation. Vista is just a first stage, and so is the Xbox 360 (and the Zune).

[tw]

Originally Posted by mbpark

You're right. They've lost the way of innovating to make customers happy. They're looking at ways of removing your ability to do things and calling it innovation. Vista is just a first stage, and so is the Xbox 360 (and the Zune).

Somewhere in the middle of Vista development, Microsoft had a 'change of heart'. They began listening to the big security companies. That meeting is said to have gone on all night and into the morning. Microsoft made some fundamental changes in the middle of Vista development that granted security software deeper access into the kernel.

Don't know if that meeting was the result of new OS Development Group management or just coincided with that change. But Vista is MS finally addressing their security problems rather than patching and kludging solutions.

In that same light, could Vista appear on an iPhone clone? Not really. I believe OS Group management had so little grasp that size restrictions make it unable to accomplish what Apple can do with OSX.

Long ago, I believed breaking Microsoft into application software and OS companies would have been a good thing. I believe by not doing so, then bad management was able to survive in the OS Development Group. This original blog would simply be a symptom of that MS management problem.

It took a Vista fiasco to recognize the symptoms. But I do suspect the real problem lies with Steve Ballmer. He may be a ruthless deal maker, but he is not and will not attract people necessary to make innovation thrive.

It is rather interesting how Xbox 360 got developed completely independent of MS legacy products - complete with their own chip designs. I wonder how that happened - how Xbox got so much independence.