How Vista Mistakes Changed Windows 7 Development

snydeq writes "For the past several months, Microsoft has engaged in an extended public mea culpa about Vista, holding a series of press interviews to explain how the company's Vista mistakes changed the development process of Windows 7. Chief among these changes was the determination to 'define a feature set early on' and only share that feature set with partners and customers when the company is confident they will be incorporated into the final OS. And to solve PC-compatibility issues, Microsoft has said all versions of Windows 7 will run even on low-cost netbooks. Moreover, Microsoft reiterated that the beta of Windows 7 that is now available is already feature-complete, although its final release to business customers isn't expected until November." As a data point for how well this has all worked out in practice, reader The other A.N.Other recommends a ZDNet article describing rough benchmarks for three versions of Windows 7 against Vista and XP. In particular, Win-7 build 7048 (64-bit) vs. Win-7 build 7000 (32-bit and 64-bit) vs. Vista SP1 vs. XP SP3 were tested on both high-end and low-end hardware. The conclusions: Windows 7 is, overall, faster than both Vista and XP. As Windows 7 progresses, it's getting faster (or at least the 64-bit editions are). On a higher-spec system, 64-bit is best. On a lower-spec system, 32-bit is best.

Comments

Will run on netbooks or drag?

[Anonymous Coward]

We have talked about W7 performance on netbooks which will only allow to run 3 apps. Perfect for an antivirus, a firewall, an antispyware, the WGA... oh crap!

Re:Will run on netbooks or drag?

[artor3]

My six year old laptop can run Windows 7 acceptably. It's not fast, but it's good enough to be usable for email, web-browsing, even YouTube videos. Therefore, I'd expect W7 to run fine on netbooks.

That said, there's the question of why you'd want it on a netbook. It's different enough from previous versions of the OS that your grandma would probably prefer to just use XP, like she has been for years. And if the user is willing to accept a change, why pay for W7 when you can use some form of Linux, custom tailored for netbooks?

The main draw of Windows is compatibility with all the apps out there. Netbooks aren't going to be running those apps, so why bother with Windows?

Re:Will run on netbooks or drag?

[jmorris42]

> That said, there's the question of why you'd want it on a netbook.

And that is their problem. Lets assume they really do make it faster than XP. (I know, but go with me here.)

You are looking at netbooks. Three options are lined up:

1. Linux. Cheapest on display, looks pretty but not Windows so it makes you a lottle nervous. (From POV of lifetime Windows user)

2. Windows XP. Only a few dollars more than Linux, familiar, safe choice. That's why it is smoking the Penguin now. Of course this is only because Microsoft is basically giving it away.

3. Windows 7. Folks say it actually runs a little faster than XP! Of course you pay even more than XP but you only get to have three apps open.... unless you pay a LOT more.

So hands up if you would pick option 3. Uh huh, and that's their problem. Cheap XP stopped the Linux threat but now XP is likely to kill Windows 7 just as dead on the netbook. And if they kill XP the odds are pretty good that the penguin will resume rampaging all over the netbook market. But if XP is kept available and security updates are kept going how the heck do they get the corporate desktops to do a full refresh? Because they WON'T believe Windows 7 will run so well they won't have to refresh most of their hardware. And in this economy that probably isn't in the budget, especially if staying put on XP is an option.

And all these careful plans are subject to being void if the ARM netbooks ever show up in force and live up to their prerelease publicity. Because then it is full Linux with OO.o, Firefox+Flash+plugins and repos with thousands of apps vs WinCE fighting it out in a segment where the prices will be falling into the $100-$200 range. Even if Microsoft 'wins' the hit to their revenue stream from competing with zero is going to start to hurt. Meanwhile those $400 x86 netbooks are falling to $300... at least if the cost of a Windows license stays cheap... but then it kinda has to since Linux isn't likely to have a price increase.

And it gets better. As more corporate IT peeps learn Microsoft is handing out XP licenses for darned near $0 but won't let them get it unless they pay extra on top of a full Vista Business license they just might start asking their Microsoft sales weasels questions that really have no good answers. Or run some Linux pilot projects and make sure word get back to Microsoft, since that seems to get their attention. More downward pressure on revenues.

Re:Will run on netbooks or drag?

[Anonymous Coward]

You are looking at netbooks. Three options are lined up:

1. Linux. Cheapest on display, looks pretty but not Windows so it makes you a lottle nervous. (From POV of lifetime Windows user)

2. Windows XP. Only a few dollars more than Linux, familiar, safe choice. That's why it is smoking the Penguin now. Of course this is only because Microsoft is basically giving it away.

3. Windows 7. Folks say it actually runs a little faster than XP! Of course you pay even more than XP but you only get to have three apps open.... unless you pay a LOT more.

So hands up if you would pick option 3. Uh huh, and that's their problem.

They are paying OEMs to put Windows XP home on netbooks. Savvy people are buying these, wiping the disk, and putting Ubuntu on them. A full, unconstrained version of Ubuntu. Exactly what Microsoft cannot compete with and doesn't even want to try.

Savvy people such as the French gendarmerie:

http://www.osor.eu/news/fr-gendarmerie-saves-millions-with-open-desktop-and-web-applications

I find it amusing to think of Microsoft subsidising the hardware of my ex-XP Home-now-Ubuntu netbook.

The really amusing thing is going to be watching Microsoft try to figure out how to get Windows 7 installed on future netbooks in place of XP Home ... and yet still make a profit.

Same price as current XP Home ... no profit.

Reasonable price for Windows 7 ... no Windows 7.

Re:Will run on netbooks or drag?

[whoever57]

I just love this stement in the article:

The decision in 2004 to move to open source, was raised by one of the Gendarmerie's accountants. "Microsoft was forcing us to buy new software licences. This annoyed our accountant, who tried OpenOffice." According to Guimard the proprietary software maker then started lobbying the Gendarmerie, which is how the general manager found out about the experiments. "When he saw OpenOffice worked just as well and was available for free, it was he that decided it should be installed on all 90.000 desktops."

Talk about firing both barrels of a 12 gauge footgun!

Captain Obvious descends

[girlintraining]

"How Vista Mistakes Changed Windows 7 Development"

You got it wrong: Vista was the mistake that caused Windows 7 development.

Re:Let's see it against Ubuntu 9.04

[morgan_greywolf]

Well, instead of throwing a chair at you, I've decided to take your challenge! I had Netcraft test our Microsoft Office benchmark suite with Office 2007 running under Wine on Ubuntu 9.04 32-bit and under the latest 64-bit build of Windows 7.

Unsurprisingly, Windows 7 wins by a longshot! Ha! *throws chair* I'm gonna fscking KILL Mark Shuttleworth! Muahahahaha!

-- Steve Ballmer

Re:Let's see it against Ubuntu 9.04

[Jurily]

Can't wait for those results!

Let's pitch those against my Gentoo. Next month, when I'm done with the compiling.

Whitewashing

[SuperKendall]

Microsoft is basically doing a Vista service pack with Windows 7, but they have put out a TON on press on sites like Digg and Slashdot to change the mental landscape around Windows 7 with consumers and the core technical crowd. At this point I'm pretty skeptical of every pro Windows 7 article and poster, though of course by now you'd expect Vista to have been improved.

OK, so we fucked up. It's good now, really!

[Trip6]

MS deigns to send the message that they care about the customer and the community. It would have been nice if they did that the last time. Sorry, I'm already on OSX.

Re:Will Windows 7 support the devices I already ha

[BradleyUffner]

"Vista wouldn't support the perfectly good Epson Perfection 1200U scanner that I bought some years ago, and for which Epson chose not to release Vista drivers. Likewise for other devices."

No. Epson choose to not support your scanner any more. It's not Microsoft's fault that a 3rd party decided not to fully support your hardware with drivers for the latest OS. Vista would support it perfectly fine if Epson would write drivers for it, but they are banking on you choosing to buy a newer model scanner.

Don't blame Microsoft for Epson's greed.

I don't want Windows on my netbook

[Gothmolly]

I have a Dell Mini 9, and it does just fine with Dellbuntu 8.04. Even the 512MB RAM is fine - the screen size and form factor does not lend to massive multi-app multi-desktop kind of work. It's an über PDA, that I can put Postgres on if I need it.

Re:release date

[Bored Grammar Nazi]

What are you? Another Microsoft marketing/misinformation drone? Or have you just been brainwashed?

You're an IT administrator for a bank. You support about 35 mission-critical applications that go to a mainframe. [...] All those applications were written for Windows 95.

And the mainframe is running what? Windows For Mainframes Edition? I don't think so.

Now, Microsoft is a safe bet because you know those applications were written decades ago and will still work.

I disagree. I only use Windows at work, but it is my understanding that it is very difficult to make older Windows applications run in newer versions of Windows, especially applications that were written for Windows 95/98.

But you go with Apple, or Linux and what do you get? Every five years, maybe ten if you're lucky, you have to rebuild and redesign everything to make it work with the latest and greatest.

That's assuming that you keep updating Linux or Mac OS to the latest and greatest. But you don't have to. In your mainframe "example" it is assumed that the system images running the applications are not being updated. And then you complain that Linux/Apple apps may break if you update the OS? Come on.

You might want to change your desktop background to this one.

Re:release date

[jmorris42]

> Microsoft delivers what businesses want: Reliability. Long. Term.

You clearly don't know what the heck you are babbling on about. You were on target with the mainframe, that is reliability over the long term.

Windows? You think going back to Windows 95 is long term? Bah. Windows 95 wasn't even close to usable until OSR2 and that was practically Win98 and as I recall didn't ship until '97. So a puny dozen years.

> Now, Microsoft is a safe bet because you know those applications were written decades ago and
> will still work. They're horrible, out of date, and make your butt itch just thinking about
> them, but they work, and it's cheaper to keep them going than to invest in an all-new infrastructure.

Small midsize shops are the ones who fell into this trap, usually called Visual Basic. Crappy little apps written by long forgotten consultants. And nobody had enough sense to demand the source code so now changes aren't possible. I have about as much sympathy for these fools as the Masters of the Universe on Wall Street currently reaping their reward for being dumb. You base your business on stuff you can't repair, realize the problem and don't make fixing it a goal. Then someday when it does go foom they will be shocked! shocked! and probably be lining up at the nearest public teat looking for a bailout like the banks.

Oh, and see above about 'decades ago'. Now there ARE some industrial process controls still running DOS that can get over two decades old... barely. Go really get DECADES you have to look at mainframes and COBOL.

> But you go with Apple, or Linux and what do you get? Every five years, maybe ten
> if you're lucky, you have to rebuild and redesign everything to make it work with
> the latest and greatest.

I won't argue about Apple, which is probably why it has had and has no future in the Enterprise outside of the occasional graphics arts dept full of Macheads nobody wants to piss off. Linux/Unix on the other hand.... Do you realize how old UNIX is? Even the POSIX standards predate Win32 and UNIX had a rich history already.... which was sorta the reason for POSIX in the first place but that is another tale for another day. Write to the specs and any end user app will probably be ok for the foreseeable future. Yea if you want to run an old 90s app today you will probably need to scrounge up the Motif libs but they are still available on supported Enterprise distributions. Sure it will LOOK like an old Motif app but then you want it to be the same, ya know, reliable. You could also get even older UNIX applications going but good grief, before Motif X programs were primitive, Gilligans's Island primitive, ugly things.

Re:release date

[vadim_t]

Win95 isn't supported as of 2001. So it's equivalent to old releases of Linux in that regard.

I worked on projects that still run on kernel 2.2 (this is from 1999-2000) as of today. So I can tell you how that works from personal experience. Hardware support is complicated, valgrind doesn't work (which makes debugging C apps a bit of a pain), some things like LVM and RAID are much inferior to their current state, but other than that, it's a perfectly functional system, and most software that's not tightly linked to kernel functionality (like valgrind) works perfectly fine on it.

Nothing stops you from using the latest version of firefox, vim and gcc on 2.2 if you so wish. Try to install IE7 on Win95 though.

I've seen ancient Windows boxes used in the same way, and in my experience it's a lot more unpleasant. At least you can coax Linux to work in unplanned situations, but good luck on getting anything modern installed on a Win95 box. The installer will probably refuse to even try.

You have exactly the same tradeoffs with both systems: Keep it running, even after the vendor pulls support, or keep upgrading. Keep it running for long enough, and eventually you will have to catch up with lots of things at once.

Re:release date

[DoofusOfDeath]

Microsoft delivers what businesses want: Reliability. Long. Term.

I'm sorry, and I honestly am not trying to troll here, but are you fucking kidding me???

Long term? Linux supports pretty much the whole POSIX API and, for graphics, X11. Those were mature before Steve Ballmer threw his first chair. Many serious, graphical programs written 20 years ago for Unix still build and run no problem on Linux. And it's a pretty damn good bet that it I write clean Linux code today, it will build in 2019 version of Linux or its successor. Tried running a Win16 program lately? Or tried lately accessing a web page written in their proprietary dialects of HTML from back in the browser war days? Good luck being able to use those web applications with the browsers that are available in 20 years.

Reliability? Windows servers have historically needed a period reboot, just because. The DoD recently disallowed USB thumbdrives on any of their computers. Hint: it wasn't because of the Linux computers. And what would you rather hook up to the open internet for 24 hours after installing the operating system: Windows XP, or Linux?

Or maybe you're referring to their steadfast trustworthiness as a company. Surely we can trust their products because as a company they're so wise, right? Like their decision to encourage web page designers to include ActiveX controls on the web pages? Or how many apps broke when Vista was rolled out?

I must concede, though, that Linux might just not be ready for mission critical deployments.

But you go with Apple, or Linux and what do you get? Every five years, maybe ten if you're lucky, you have to rebuild and redesign everything to make it work with the latest and greatest.

I can't speak about Apple stuff, but for Linux, who cares if the people shipping a distribution needed to re-compile 50% of the apps when preparing a release, because of some library ABI change? When you have the source code to the apps, and someone else (the distro maintainers) recompile everything for you anyway, it. just. doesn't. matter.

Re:release date

[Saint_Waldo]

Okay, here's an "artificial barrier": You're an IT administrator for a bank. You support about 35 mission-critical applications that go to a mainframe. Why keep the mainframe? Because it's the only thing that's gone through the laborous process of being documented, audited, and certified for use. Those certifications could run into the tens of millions of dollars, plus another fifty million to retool your existing infrastructure, minimum. All those applications were written for Windows 95.

Now, Microsoft is a safe bet because you know those applications were written decades ago and will still work. They're horrible, out of date, and make your butt itch just thinking about them, but they work, and it's cheaper to keep them going than to invest in an all-new infrastructure. But you go with Apple, or Linux and what do you get? Every five years, maybe ten if you're lucky, you have to rebuild and redesign everything to make it work with the latest and greatest.

Microsoft delivers what businesses want: Reliability. Long. Term.

And that costs money, time, effort, and yes... it's a MUCH higher standard to reach for.

Whoever thought this was insightful, isn't.

Your use of "Mainframe" could have client apps written in anything. In fact, you fail to point out what the mainframe is running. If, as you claim in your hypothetical, the mainframe system is the part that's documented, you can always write a conforming client on just about anything, yes, windows included but linux and MacOs as well.

As a real-world proof, I've assisted building a web application that interfaces with a legacy PIC database and replaced proprietary desktop apps with a thin net client. After our work, what OS is required by the millions of users? We don't care, any browser made after 1998 could run the app, on any OS that runs the browser.

If you fail to see this, you deserve to pay Redmond every dime you already obviously do.

Re:release date

[Tr3vin]

http://www.virtualbox.org/register

Re:Vista SP2

[NeverVotedBush]

I bet if you do a uname -a at the C> prompt, it will say something along the lines of:

Windows 7 Desktop 2.6.27.19-3.2-default #1 SMP 2009-02-25 15:40:44

Re:release date

[Technomancer]

Take any linux binary compiled 10 years ago and run it today on a shipping kernel. Oh wait... you can't.

Sure I can! This is probably the oldest binary app that I have and coincidentally it was compiled more than 10 years ago.

root@damage:/usr/local/games/quake#ls -al quake.x11
-rwxr-xr-x 1 root root 427892 Feb 10 1999 quake.x11

root@damage:/usr/local/games/quake#uname -a
Linux damage 2.6.26.8 #1 SMP PREEMPT Mon Dec 22 02:52:09 PST 2008 x86_64 Dual Core AMD Opteron(tm) Processor 285 AuthenticAMD GNU/Linux

root@damage:/usr/local/games/quake#date
Tue Mar 10 22:28:41 PDT 2009

root@damage:/usr/local/games/quake#./quake.x11

Added packfile ./id1/pak0.pak (339 files)
Added packfile ./id1/pak1.pak (85 files)
PackFile: ./id1/pak1.pak : gfx/pop.lmp
Playing registered version.
PackFile: ./id1/pak0.pak : gfx.wad
Console initialized.
UDP Initialized
Exe: 14:08:23 Jan 25 1999
  8.0 megabyte heap ....

and so on

Re:Vista SP2

[Dreadneck]

I propose calling it ReVista

Re:Vista SP2

[Slorv]

No, Vista ME seems more proper