MS Upgrades To Be Smaller And More Frequent

duplicantk8 writes "Following the numerous delays to the Vista launch, MS is planning to have more frequent and smaller incremental upgrades, according to the Financial Times." From the article: "Those delays are set to end late next year with the simultaneous launch of new versions of Windows and the Office suite of PC applications in the company's most significant new product cycle since Windows 95. The new versions of the company's key PC software are likely to rekindle higher growth after a period that saw its growth rate slip below 10 per cent for the first time last year, according to Wall Street analysts. Mr Ballmer's comments are the most public sign yet of the dent to Microsoft's confidence in its core development process that resulted from the Vista delays."

Here we go again

[FatRatBastard]

Those delays are set to end late next year with the simultaneous launch of new versions of Windows and the Office suite of PC applications in the company's most significant new product cycle since Windows 95.

This phrase gets dusted off for every OS release MS makes. Heard it for 98, ME, 2000, XP, 2003... and will continue to hear it for every other bloody version MS flogs.

Re:Beleaguered

[sg3000]

I can't find the original article, so here's the only reference I could find:

During his keynote speech at MACWORLD Expo Boston ... Dr. Amelio announced a fundamental shift in the way that Apple delivers new operating-system functionality.

Dr. Amelio stated in his keynote speech that Apple is changing its strategy to deliver new functionality through incremental releases rather than large monolithic releases. Moving forward, Apple intends to follow the industry model of shipping software releases in incremental segments. ... The motivation for this change is that Apple believes that its current model of monolithic system-software releases isn't working, and that it doesn't allow Apple to get software advancements out to customers and developers soon enough.

I found a similar statement in a Boston Globe article from August 8, 1996:

As far as Apple's new operating system, known as Copland, Amelio wouldn't give a release date, saying instead the company would begin selling components of the new operating system as they become available. Such

piecemeal advances in the operating system are part of a broader shift by Apple away from big, monolithic upgrades. "Copland is going to appear, but it's going to appear over a series of releases," said Amelio.

Who would have thought that about a decade later, it would seem like Microsoft was having the same problems:

Microsoft has overhauled its core software development practices to avoid any repetition of the delays that have bedevilled the next planned version of Windows, according to Steve Ballmer, the company's chief executive.

The changes, along with plans to release more frequent, less ambitious versions of the widely used software, mark a significant shift in Microsoft's approach following one of the most troubled new product cycles in its 30-year history.

"We attempted something that was beyond the planning and conceptualisation of the system," Mr Ballmer said of Windows Vista, the much-delayed version of the software that is now planned for late next year.

"The product cycle has been longer than it should have been," he told the FT.

Of course, what fixed Apple was not doing incremental releases. They had to do a step-function switch to Mac OS X.

Re:Smaller changes?

[amliebsch]

Windows has had translucency since Win2000. The big deal for me is the deprecation of the Win32 platform, the first-class status of managed code, the deprecation of GDI and the introduction of the new DirectX compositing system, and some very significant changes to the security model. The Windows UI that they create is more of an afterthought as far as I am concerned. I'm much more interested what I can do with it as a developer, and I was blown away by the Sparkle demo yesterday.

Re:Just last night . . .

[caluml]

From what I understand, Linux doesn't lock the files like Windows. You can overwrite a file that's already open, and all new opens of that file will use the new contents. I've certainly never seen an error like: "cp: Error: Unable to copy file - destination file locked" or similar.

Re:Just last night . . .

[QuestorTapes]

> The catch is that if you need to patch a critical system file, it's orders of magnitude
> more simple to just replace it upon reboot (since nothing's running). Otherwise you need to
> close down any applications and services that are using that file. Some system files are used by
> the GUI interface itself, at which point you're crossing your fingers and hoping it pops back to
> reality during the patch process.

Yes. But a lot of that is due to the fact that MS never really structured the system files properly. If they had done so, this would not be the problem it is.

> It's probably technically possible to do certain patches without rebooting

Very possible.

> but you'd have to have a savvy enough user to shut down and bring back dependent services.

Not really. If the installer is properly designed using MS Installer, it should fall back to copy-on-reboot if anything is in use, and alert the user to reboot. It's only the install programs that make assumptions that are a real problem. Instead of falling back to copy-on-reboot, they choke and die with a cryptic error message.

Re:too ambitious?

[pammon]

Spotlight is not powered by mySQL.

Re:Just last night . . .

[mchawi]

Just write a utility to do it for you, or download one of the numerous ones that do this for you. IE:
http://www.dr-hoiby.com/WhoLockMe/index.php [dr-hoiby.com]

I don't really see the big deal...

Re:too ambitious?

[sqlrob]

You're right about Cairo, but that's NT 4 (1996), not Win 2k.

Re:Just last night . . .

[Spoing]

From what I understand, Linux doesn't lock the files like Windows. You can overwrite a file that's already open, and all new opens of that file will use the new contents. I've certainly never seen an error like: "cp: Error: Unable to copy file - destination file locked" or similar.

Inodes are a feature of all file systems under UNIX and unix-like systems including Linux. When you access a file, it's 'locked' in that it will not vanish on the process that opens it...yet, each process has a different inode.

Because of that, you can have one program that moves a file, another that deletes the 'same' file, and yet another that is currently editing the file. Each has a different inode. The result is that you can update a program, for example, and not have to exit it...but still fire up the new version!

Here are a few notes on this nifty feature;

http://www-1g.cs.luc.edu/~van/cs219/lect0/

http://www.unix.org.ua/orelly/networking/puis/ch 05_01.htm

Re:Just last night . . .

[Spoing]

Slight clarification: The inodes are per-file per access.

If you have an app that loads 3 libraries, it has 3 different and unique inodes.

If another program loads the same libraries, that program has 3 different and unique inodes...for a total of 6 inodes between the two programs.

Re:Just last night . . .

[dodobh]

You are confusing filehandles and inodes. Inodes are unique on the filesystem. Filehandles are per program.