Dave Winer blogs his reaction to the Allchin memo:
BTW, "Hard core" means "death march." It's the same trap that Apple fell into with Copland.
The devteam was always in death march mode, when one impossible ship
date was missed, they scheduled it for another impossible date. When
you ask a Microsoft person to say what Longhorn is supposed to do, you
get rambly hand-wavy words that mean nothing. A product with a purpose
has a two-sentence description that gets everyone so excited they can't
wait. Longhorn isn't designed to solve anyone's problems. I think they
all know it, but they can't say it out loud because they've all drunk
the Kool Aid on this.
Microsoft announced Friday afternoon that "Longhorn," the
next-generation Windows, was losing the feature code-named "WinFS" that
was to provide speedy searching and intelligent cross-linking of all
documents the machine had seen. In addition, Microsoft announced that
two of the other touted features, "Avalon," a new graphical sub-system,
and "Indigo," a communications sub-system, would be available for
Windows XP and Windows Server 2003 machines in the same timeframe as
the estimated release of "Longhorn," currently 2006 for the client and
2007 for the server. Sounds like "slipping is a feature."
The media has gone wild with speculation, of course. Mary Jo Foley must have put in overtime to post allthesearticles. A supposed memo from Jim Allchin is chilling in its NewSpeak terminology.
What does that mean for us? For most Windows developers, it should mean
two years of peaceful stabilization of the platform before the next
round of chaos, and a good long time to consider alternatives. If
Avalon changes everything on the graphical layer, is DotNet WinForms
dead? WinFS, which Microsoft insists never stood for "File System"
could be replaced by a half-dozen good third-party tools.
Microsoft needs to take a long, hard look at this, and realize that
"Operating System" no longer means "a way to tie people into all of our
software." They need to stop adding things to their operating system,
and start removing things until nothing is left but the core, and sell
add-on features like searchable databases and graphical UIs to
the people who need them as *products* and not as features. Sure, it's
easier to sell one SKU than many. The Operating System should provide
the facilities for the computer to operate, read, write and
communicate, and offer open and documented APIs through which the
appropriate tools can be added to provide features above the OS level.
Andrew MacNeill sees this as a good thing, "gutting" Longhorn, and Robert Scoble says
"Longhorn wasn't aimed at the sweet spot of the market anymore and our
customers were telling us to go in a different direction." It will be
interesting to hear what direction that is.
