__/ [ BearItAll ] on Friday 30 March 2007 14:39 \__
> Roy Schestowitz wrote:
>
>> Windows Vista VPN: A Step Backward
>>
>> ,----[ Quote ]
>> | Unless you are running a full set of domain services, the process
>> | is a little bit like divining the future amongst chicken entrails:
>> | messy and unpleasant with a heavy dose of guesswork. It typically
>> | involves manual manipulation of firewall ports, manual mapping of
>> | hosts on the VPN side, and a lot of shrewd guessing.
>> `----
>>
>> http://saunderslog.com/2006/10/13/windows-vista-vpn-a-step-backward/
>>
>> And they say Linux is complicated and Windows gets simpler at each
>> iteration (release)...
>
>
> It's getting harder to believe that they was a controlled development
> process involved with MS Vista isn't it. I certainly can't believe that no
> one over the 5 years actually sat down and asked the question 'Will this
> work?'.
>
> There is a huge list now of things that do not work. Amongst them some
> individual ideas that might have actually been very good ideas. But you
> can't help feeling that despite the beta versions, Vista with all the
> modules etc from the various teams was never tested as a complete unit and
> that many of the individual areas, well the development teams must have
> tested them on XP but never got to test on Vista. How can it have been
> tested when they is so much that doesn't work, and so many other parts that
> work badly.
This was said some time in late 2005 or 2006 (by the press, IIRC). It turned
out that there is poor communication between the teams and lacking proper
protocols (which are needed where there is choice, e.g. desktop
environments, bootloader) there's poor integration. The teams work in
isolation and then just 'glue' the pertinent bits. In an amusing statement,
this was compared to the lego-type approach, which is wrongly attributed to
Linux (a myth).
> Either the development teams were lieing to the board about their part
> being ready and working, or those teams were not allowed to test them on a
> Vista.
Vista was created in about 6 months before feature freeze. Longhorn was
scraped in 2005, which led to confusion and a rush. It's hard to say how bad
Longhorn was. It couldn't be built properly (monsterbuild) and the teams
could not get it to run, according to Allchin. It was probably chaotic if
the /better/ choice was to drop 3-4 years of development down the chute,
rather than work on fixing it.
> And that is what I think has probably happened. I know from the past that
> MS has many small teams, some even being teams of one, they do not share
> but are given a goal for the function class or procedure, but not given
> information on the other modules that it forms a part of. That stops
> programmers leaving and giving away the secrets of the next release. So
> each of those gets to the end of their little part of the whole project and
> tick the box to say their part works. It hasn't been tested in the end
> product, because they don't have access to it.
Also bear in mind that Microsoft /buys/ technology (software), rather than
develop it. Even a senior Microsoft exec admitted this a fortnight ago. This
leads to inconsistencies. There's no seamlessness.
When they try to develop something in-house, then look at Live OneCare and
see what happens...
> But somewhere along the line MS have to trust someone, a core team, who
> pull all of the pieces together and test the entire unit. That clearly
> doesn't happen because Vista is having such a hard time in every
> department. This puts the onus on that core team, if they exist they
> haven't done their job, if they don't exist they should.
There was at least one project manager shuffle in the Windows team. I don't
know how Allchin fits into all of this, but he retired as soon as Vista was
out the door...
There are lots of links I could add here, which support everything above. I'm
in a bit of a hurry now, but I'll gladly support each point if you want.
--
~~ Best wishes
Roy S. Schestowitz | GPL'd Reversi: http://othellomaster.com
http://Schestowitz.com | Free as in Free Beer ¦ PGP-Key: 0x74572E8E
Cpu(s): 22.9% user, 4.9% system, 0.8% nice, 71.5% idle
http://iuron.com - semantic engine to gather information
|
|
