[Leaplist] [Meta/OT] Debian Lenny might be pushed back again.

[Bryan J. Smith]

[ I'm going to tag this Meta/OT, I started it, my apologies ]

On Tue, 10/14/08, Hank Lambert <hank at hanklambert.com> wrote:
> I completely agree that is what happens, and it's a
> great argument in comparing Microsoft to Apple, but doesn't
> Linux have the same inter-dependence to overcome

I would Darwin (Apple's codebase of BSD) doesn't either.
Maybe some Cocoa, etc..., but I can attest 0% to that.

I know a lot of NT, as a result of throwing a lot of crap
into the core of the OS during Chicago (DOS 7/Win 4 aka
Win9x/Me) does.  Kinda sad.

> (this isn't a rhetorical question)? It seems that Linux
> has even more to overcome because some distros cover a
> wider variety of platforms, far more than Microsoft.

Well, that's a little bit different and another issue.
Although it does add to their portability issues.

Microsoft's Win32 API has absolutely and always been tied
to little endian.  And pretty much the death of non-x86
NT resulted from most of Win32 developments assuming no
data alignment rules.  This has also been their portability
nightmare for Win64, period.

In that regard, as I understand it, Apple does have some
difficulty getting to x86-64 above the Darwin platform too.
The Darwin platform may be x86-64 portable, but some components
still seem to have issues.  Although that could be more of an
issue with their unified model, as the PPC end of the game was
never designed for Power 4+ (64-bit).

But beyond that still, even some legacy open source codebases
have their issues too.  E.g., OpenOffice.org still has some
portability issues to x86-64.  But those based more on legacy
community developments that did take GNU64 into account do
not, like Mozilla/XUL, GTK+/GNOME, Qt/KDE, etc...

But that's still unrelated to the inter-dependence issues.
GNU/Darwin and GNU/Linux don't have them.  NT/Win does.

> However, I believe that Microsoft's biggest problem is
> that their release cycle is set by marketing, not
> engineering, and that is the point I was not
> doing a good job making in my first e-mail ;)

Actually, it's far more simple than that.  They promise
something, then never ship it.  NT 6.0 "Longhorn" was more
of the NT 4.0 "Cairo" issue -- vapor, vapor and more vapor.
Cario technologies, then Longhorn technologies, then WinFX
technologies, etc... never happened at all.

Even Windows 7 is still being based on NT 6.1, not a new
NT 7 release (NT 6.1 = Vista SP1 / Server 2008).

And after about 2001-2003, they lost all of their major,
core architects anyway.  So they've gone down the tubes
from there on core developments.  Microsoft produces little
new innovation, at least at the core level.  Even AD and
Exchange are little changed from the '90s.

> When Debian isn't ready, it doesn't get released. When
> MS isn't ready, it is still released.

Again, too simplified.  Dealing with patch management
is an outstanding example of the difference.  ;)

> I do acknowledge, however that Debian is at the other
> end of the scale in relation to stability, and not the
> typical distro or a fair distro to choose in comparing
> to MS.

I'd argue inter-dependency and lack of standards is their
biggest issue.

True, any distribution that lacks unit, integration and
regression testing can have its own sets of issues.  Distros
can and will vary there.

But Microsoft cannot accomplish regression testing, but
only regression mitigation, often as a result of integration
issues they can never solve.

In other words ...

Standards-based software, in general (even if the code
is proprietary), can reach a respectable level of unit,
integration and regression testing, especially the more
"piecemeal" and "broken down" the components are.  Open
source helps even more, as long as there is community
development (and it is not stangant/abandonware), to
ensure interfaces are open, documented and interact as
designed.

Even proprietary-based standards can also have the same,
depending on the dedication of the company to those
interfaces.  Several vendors have maintained outstanding
records of portability and data interoperability as a
result.  Corel's long-standing history of products, even
non-Windows, Adobe as well, are great examples.  On the
OS and protocol side of things, Apple too has developed
some excellent, but standard (and proprietary standard)
centric software sets that do many things correctly
(including some solid IETF-based implementations).

But Microsoft isn't even proprietary-standard.  They
purposely abandon their designs, over and over.  It used
to be bad in the '90s, when they did it on purpose, even
when they had the talent to not.  But now it's gone
pathetic, as they've lost virtually all of their core
architects.  That, combined with the inter-dependency
non-sense of things getting slapped on and in both core
subsystems as well as core libraries built by Visual
Studio, really killed everything.

I realized it the first time I build a service with
Visual Studio and it required Internet Explorer libraries,
which had nothing to do with anything it did.  That's where
the inter-dependency non-sense comes from, unlike most other
OSes entirely.  And they have no control over it.  They can
never complete integration/regression testing, so it becomes
more of a matter of mitigation as some point, just so they
do ship.  Because if they waited, it would never happen.  ;)


-- 
Bryan J Smith          Professional, Technical Annoyance
b.j.smith at ieee.org    http://www.linkedin.com/in/bjsmith
--------------------------------------------------------
I don't have a "favorite Linux distro."  I use, develop
and support community efforts, often built around Linux.
Technology and solutions are my focus, not dragging in
assumptions, marketing and other concepts which dominate
non-community developed software, which I left long ago.


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.