Re: Apple - Slashing Features from Leopard Left, Right and Center!

In article <137a5cj1ed6m906@xxxxxxxxxxxxxxxxxx>,
"Dan Johnson" <danieljohnson@xxxxxxxxxxxx> wrote:

"ZnU" <znu@xxxxxxxxxxxx> wrote in message
news:znu-5598C8.17381616062007@xxxxxxxxxxxxxxxxx
In article <1378j1bfst1irfc@xxxxxxxxxxxxxxxxxx>,
"Dan Johnson" <danieljohnson@xxxxxxxxxxxx> wrote:

"ZnU" <znu@xxxxxxxxxxxx> wrote in message
news:znu-A9FDD6.15251516062007@xxxxxxxxxxxxxxxxx
In article <1378afanfe2v17d@xxxxxxxxxxxxxxxxxx>,
"Dan Johnson" <danieljohnson@xxxxxxxxxxxx> wrote:

I disagree.

Basically, they decided Longhorn would be an incremental release, based
on the XP codebase. Then they decided they'd go for broke, and
implement
most of what they had planned for Blackcomb, on the same foundations.
That wasn't working, so they ported the stuff that was working to the
Windows Server 2003 core.

If you really believe what you are trying to imply, why don't
you tell us what didn't make it?

We know about WinFS and NGSCB. What else?

Well, for starters:

I take it from the long quote that you have no answer yourself.

[snip- long quote]

But I think this about covers the gist of it:

"Everything in Longhorn was supposed to be written in C# and to be
managed code.

There was never *any* possibility that Longhorn would be
completely rewritten in any form, and even less that it would
be done in managed code. It is pure fantasy.

The fact that your 'unnammed developer source' thinks this is
even plausible makes me wonder if he is actually a developer
at all.

Several sources. And while the notion of rewriting the entire OS as
managed code does seem a little far-fetched and is perhaps best
explained has hyperbole, it was widely reported around the time that
significant user-level Windows components like Explorer would be
re-implemented on .NET.

[snip]

Microsoft constantly announces features without giving any clear use
case. Have you ever seen a demo showing exactly what difference WinFS
(assuming it ever actually shipped) would make to users, for instance?

WinFS is an API; it was for developers. They were supposed to use
it to store metadata about their documents; and they were supposed
to be able to store metadata against documents that were not their
own, as well.

Seems fairly clear.

Or, another example. When Microsoft originally announced .NET, nobody
knew what they were talking about.

Well, they made the 'NET brand so vague back then it confused
people; but if you bothered to look at the actual products, those
were much more clear.

Was it a new API for writing Windows apps,

The .NET Framework was this.

Yes.

Microsoft's answer to Flash,

They didn't have anything like this, and I did not notice
anyone thinking that they did back then.

Some people thought it was supposed to be a Java
killer, which it was. But that is not quite the same thing.

They did, though. There was a demo in the original announcement where a
component of some kind, embedded in a web page, dynamically generated
graphics client-side from XML.

a blanket term for a bunch of hosted services that would be integrated
with Windows,

They used the .NET brand for this also, leading to
confusion; but the services they offered were reasonably
clear (and not popular, as it happened).

Of course, that hasn't stopped them from trying again, with the "Windows
Live" brand that contains a couple of dozen tangentially related
services.

a shift toward thin-client computing?

They didn't mean this, and I did not notice anyone
thinking that they did back then.

The above-mentioned demo, all the noises about integrating web services
with Windows, and Microsoft's discussion of rentable software, lead to
fairly widespread speculation that Microsoft wanted to move in this
direction. You can find threads in this group in which people speculated
that the Xbox was a trojan horse, intended to sneak thin client hardware
into people's homes.

All of these were hinted at. Nobody had the slightest clue how
using a computer with ".NET" would actually work. Years later, it
turns out it's a new way to write Windows apps and also a
meaningless string that Microsoft attaches to the names of various
products.

They've cut back on the "meaningless string" part, and that's all to
the good.

Microsoft tries to seem visionary, but there's no substance there. They
have vast resources, and they've taken to implementing virtually every
technically interesting buzzword feature that anyone can think of. But
they seemingly have no idea what any of it is good for.

You can't relate to it, because you understand only end-user
products; But the core of MS business is developer tools.

Meh. I'm a developer as well, though I mostly write web stuff these
days. It's important to present use cases for developer technologies as
well. Apple often does.

[snip]
I'm not talking about speccing everything up-front, I'm talking about
letting the kernel guys nail things down and then creating a UI for the
sleep/hibernate/shutdown features it actually has. Given that creating
such a UI should take a week at most, this doesn't seem like a major
problem.

You are mistaken. You cannot finish the implementation
of hibernation, hybrid-sleep, and so forth without testing it;
and you cannot test it until it is hooked up end to end. That
means a UI, and it includes end-user testing.

OK, but a trivial UI is the work of five minutes. Test with that. Don't
have an 8 member team working on the final UI when they don't even know
what they're designing a UI for.

Anyway, if everything in Vista had to be serialized out the
way you describe, we'd still be waiting for it. :D

Obviously *everything* doesn't have to be serialized. But there are some
cases where it's nuts to do anything else.

You need to co-ordinate with other teams to deal with
this, and you need to evolve the spec and the software
together based on what you learn by doing this.

"Co-ordinate" tends to mean "have lots of meetings with",
of course.

Except that, if you read the post, you'll notice meetings weren't
between teams. They were within the team.

Where the organizational boundaries are is not the point. It is
the people that need to co-ordinate, and declaring them
'all one team!' doesn't change anything.

Reading that post, it doesn't sound like they were the people who needed
to coordinate. That is, it doesn't sound like the team that was meeting
regularly had members of all the various departments. It sounds like it
was contained within one department, and information from other
departments showed up at unexpected times.

Inter-team coordination sucked
because there were so many levels of hierarchy and because code checked
in by one team took weeks or months to show up in repositories used by
others.

This is because MS is so big, and the source code for Windows is
so large.

One manages large projects by breaking them down into separate units.
Where one draws the boundaries between such units plays a large role in
how effectively this works. Microsoft seems to have drawn them in the
wrong places here. I suspect, from the outward signs, that such issues
are systemic.

And another reason; MS is now trying to maintain a very high bar of
implementation quality, for security reasons. As I understand it, the
big thing they did in the Longhorn Reset was to stop allowing code
to reach the central repository until after it had undergone extensive
testing and review. This, of course, means it takes a long time to make
it from one team to another.

There is, I believe, supposed to be a tree-like structure of repository
brances, so that related teams can share their code before all that
has happened. It may be that this does not work perfectly, though.

[snip]
This all sounds fairly meaningless, actually, just org-chart stuff.

Except that the way this guy describes things, what would happen is his
team would settle on a design, and then an external team, the shell team
would veto it. That's not meaningless.

I mean the *names* of the teams are meaningless org-chart stuff. That
veto would be there whether you called them one team of five teams.

That external veto wouldn't be there if the people who were doing the
vetoing were the people on the team.

[snip]

--
"That's George Washington, the first president, of course. The interesting thing
about him is that I read three--three or four books about him last year. Isn't
that interesting?"
- George W. Bush to reporter Kai Diekmann, May 5, 2006
.

Relevant Pages

  • Re: Building VC++ 8.0 code on a machine without Visual Studio 2005
    ... developer has an individual VS 2005 Professional Edition with MSDN ... "Microsoft Windows Software Development Kit for Windows Vista? ... Windows SDK is definitely a scenario we do support and in fact the VC ...
    (microsoft.public.dotnet.languages.vc)
  • Re: windows mobile 2003 barcode scanning
    ... First, I'm a developer who supports bar codes in my applications, and ... I rely on keyboard wedges provided by the reader vendors. ... barcode scanners running Windows Mobile for a client. ... that the platform is an issue. ...
    (microsoft.public.pocketpc)
  • Re: pygame and python 2.5
    ... I realize you're a Windows user, and a Windows user with an AOL email ... developing a math library based on GMPY to use ... No, as I said elsewhere, I'm not a software developer, ...
    (comp.lang.python)
  • Re: Think longhorn will run only at super-computers? Take a look at th
    ... > Think the new windows longhorn will be something completely different from ... > serves me well) and a better Windows explorer. ... Haven't you figured out how MickeyMouse operates yet? ... To keep their partners happy they'll add something to force computer users ...
    (microsoft.public.windowsxp.general)
  • Re: App for Windows 98?
    ... can get just the Developer Edition for Office XP, ... Windows XP, but I'm getting requests to produce a version that will ... a2000 format (note that the default format for a2003 is in fact ... You can use sagekey to build a install here...but sagkey changes ...
    (microsoft.public.access.gettingstarted)