[Leaplist] Re: Microsoft Suggest Linux Hotmail Users Use Outlook Express (long)

[Bryan J. Smith]

Kevin Anderson wrote:
> You mean like IIS vs Apache?

Parallel, yes.

Microsoft has difficulty influencing the Apache Foundation because
 A) it's not a core Microsoft partner / IP licenser/licensee, and
 B) it does not require the Microsoft distribution channel

In fact, Apache has dominated the web space since day 1, at least with
Internet-centric firms.  It is also the basis for other, open and
proprietary frameworks.  Microsoft has failed to see similar happen with
IIS.

> Or .net vs PHP?

Huh?  Okay, maintain context and relevancy.  .NET !~ PHP.  If you think
so, then you should learn more about .NET.  ASP ~ PHP, some components
around ASP are ~ to some components around PHP, but not .NET as a whole.

> Or MSN vs Google?

Similar to IIS and Apache.  Google is not affected by the Microsoft
distribution channel.  About the only thing Google shares with Microsoft
is their own, viral approach to IP accumulation.

It's why Microsoft has sought Yahoo for a year now (and seems to be 

> Even IE vs Netscape 10 years ago where as IE vs Firefox now.

No, because Netscape _did_ require the Microsoft distribution channel.
Once Microsoft bundled MS IE and "rebated" PC OEMs from not distributing
Netscape, Netscape was toast.

> These are examples of competition.

But what type of competition?  If you compete with Microsoft in its
distribution channel, you better be ready for an ass kicking when they
want to deliver it to you.

> Partnerships are not competition.

Agreed.

> Are you paid by RH?

Let's ask that in a different question ...  ;)

Would I have grossed over a million dollars less, and half of that in
just the last 3 years, in just services income if I didn't deploy Red
Hat solutions?  Probably, probably would not even come close to anything
like it!

So if you don't think I'm grossly Red Hat biased, then you insult my
intelligence.  ;)  Now with that out of the way so there is absolutely
_no_ doubt ... ;)

> I get that they're a good company.

Define "good"?

> But no different than Postgres,

Yes different!  Red Hat is the tip of the GPL spear when it comes to any
Linux distributor, period.  They refuse to compromise on everything from
IP and indemnification to freely redistributable (sans the Red Hat(R)
trademark, something forced on them by several other companies** and Red
Hat tried, and failed, to solve legally with "trademark guidelines"
prior) and freely modifiable code.

**NOTE:  It should be noted that Red Hat is still the only, significant
Linux distributor to ever allow their namesake, registered trademark to
be freely redistributed.  SuSE vehemently defended their trademark
(although Novell has loosened the OpenSuSE name).  Fedora(TM) was Red
Hat's solution to issues that other companies forced.

Not even IBM has an IP approach like Red Hat.  Red Hat is one very good
entity for the industry.  I have very serious issues with Canonical and
Novell in how they conduct their business when it comes to IP.  And,
frankly, Spaceman's rhetoric is off-the-charts -- it's popular, but not
reality, but I'll leave that debate to GDK.  ;)

> Yes, Investors are greedy and stupid, but investors don't decide 
> anything of any value to me.

So far, I haven't seen Red Hat's long term investors compromise its
branding at all.  Fedora virtually remains unchanged from Red Hat Linux
prior in development and support -- sans the vendor certifications (of
which the vendors wanted, which SuSE proved with SLES that pre-dates
RHAS).  Only the day traders seem to care.

> I'd "valuate" *buntu MUCH higher than RH.  

Do you mean "Canonical"?  Or "Ubuntu"?
Do you mean "Red Hat"?  Or "Fedora"?

A lot of Ubuntu advocates like to do that, switch to RHEL when it suits
them, then Fedora when it suits them.  Let's talk Canonical to Red Hat,
Ubuntu to Fedora.  Let's not mix details.

What is Ubuntu's current model?  It's largely the "Spaceman endowment."
What does Ubuntu offer for 7+ years support?  It doesn't.

This is the common issue I see.  I also see a lot of repeat of Mandrake
here.  Once Mandrake (who didn't have something like the "Spaceman
endowment") had to start looking at its profit model, they alienated a
lot.

Ubuntu claims that it will keep updates going free for 3+ years.  And
then it often has issues with those updates, in comparison to how RHEL
and Novell approach things.  I've had regressions in the releases I've
tested.  And some upgrades have just broken bad (although Fedora did
have a major Python snafu that did also affect an update of RHEL I
believe too).

Ubuntu also has sided with some serious indemnification issues that
Debian and Fedora do not.  It also has some ABI/API considerations that
RHEL and SLES do not.  And let's face it, some of the way Ubuntu has
been developed has been really questionable, including a compromise of
their server, which brings into question its lifecycle.  A lot of people
complaint about the "requirements" to be a Fedora maintainer, and it
wasn't long after a GDK post on Fedora marketing trying to defend those
practices, Canonical's Ubuntu workflow was compromised because they
don't implement things like GDK outlined.

These are considerations.  Considerations that drive actual payment of
subscriptions.  My point ...

People are willing to pay for Red Hat and Novell for reason, to avoid
"other costs."  It's not that Red Hat or Novell are "greedy," it's
because it costs _real_money_ to do what they do.  Heck, Novell
maintains so many backports and forks, whereas Red Hat stresses upstream
(over customers sometimes), that I don't see how Novell's model is
sustainable at all.

Even Sun has had to deal with that as their prior hardware (or hardware
+software solutions) profit margins have plummeted.  It just costs way
too much to compete with a commodity, PC platform like Linux, even when
you maintain it for 7+ years of ABI/API, when the vendors sides with the
upstream of developers (even if it employs many of them with that added
cost).  I have had to explain this over and over again with newer
clients, and most "get it" and "see it" rather quickly.

> Yes, that's partly because I dislike RH (based on history with
> RH 7-9, rather than "I hate RH cause it's cool to hate RH")

Which is?  While your composing your answer, read mine ...

Understand I wrote an "Open Letter to Red Hat" after Red Hat Linux 7's
release.  I got an answer from none other than Alan Cox himself.

Alan Cox pointed out that the popular /. stories were rhetorical BS.  I
already knew Red Hat was the official maintainer of GCC 3, and he
pointed out that Red Hat finally "forced" everyone to get to ANSI C++
compliance once and for all.  They also needed the IA-64 target too
(which was only in GCC 3).  This was not different than "forcing"
everyone to GLibC 2 prior in Red Hat Linux 5 as well, which I also
recognized.

And I also new about Red Hat's prior avoidance of GCC 2.8, moving to
EGCS 1.1.2 (aka GCC 2.91.66) from GCC 2.7, and their skipping GCC 2.95
as well.  GCC 2.8 and 2.95 introduced even more splinter LibC++ APIs
that would _die_ once GCC 3 came out, which would finally bring
standards and objects with much greater longevity.  There is a lot of
_orphaned_ code in GCC 2.8 and 2.95.  The C API has been solid since
Cygnus' (now Red Hat's) EGCS, and the C++ since ANSI C++ compliance with
GCC 3.

Sure enough, Red Hat was correct, as usual.  Because today, Red Hat
maintains the longest ABI/API lineage in LibC and LibC++ compatibility
because of its choices on EGCS and GCC 3.  I can run RHL 6 and 7
applications on Fedora 9 (which is in test).  No other distro can do
that.

Just like GLibC 2 before it, pushing everyone to GCC 3 finally got
codebased updated.  Other distros then benefited from the "trailblazing"
Red Hat did, including making the efforts to rewrite a lot of C++ code
in its distro to be ANSI C++ compliant and showing everyone, "here's how
you do it."  I was maintaining various GNU C code myself when the whole
GCC 2.8 release came out, and just went nuts in the changes from GCC
2.7, and thanked God for EGCS.  But that didn't solve the problems with
C++, which Cygnus wanted to.  That's why the FSF made them (hence Red
Hat shortly afterwards) the official developer and maintainer of GCC 3.

Anything else you've heard has been rhetorical BS from outsiders.  There
was an alleged insider/developer who came up with a story that was not
true at all.  But that didn't stop /. from proliferating in ignorance.

Red Hat has been anal about ABI/API for the longest time.  And it's
extremely the case in its RHEL products, because vendors want to certify
for 7+ years, and customers want that compatibility for 7+ years.
Customers who are willing to pay 3-4 figures for it per node per year.
That's when Red Hat was able to release RHAS 2.1 (later RHEL 2.1), and
quickly taking back a lot of marketshare from the earlier SLES 7 release
from SuSE.

And what does charging all that money do?  It funds a crapload of GPL.
>From the kernel (which, outside of Novell, you'll find Red Hat virtually
alone among largely hardware contributors and still at the top) to core
GCC, GLibC and other key, components.  Red Hat has _always_ been the
major catalyst on a lot of things.  And that includes NTPL and, more
recently, MAC/RBAC in SELinux.

I remember having a "discussion" with someone on AppArmor and he kept
saying it was the "right way."  Today Red Hat has the highest, general
EAL level 4 certification with RBAC, higher than Sun on Solaris, and far
more general.  Where is AppArmor?  They aren't employed by even Novell
anymore.  ;)

> People complain about CIOs only reading the headlines, but there are
> countless, and alleged "real" Linux advocates, who are part of that same
> problem.
http://www.theregister.co.uk/2008/01/30/french_open/
http://www.desktoplinux.com/news/NS4254330887.html
> There's plenty of others, if you try one of the more popular Linux
> "applications", Google, you'll be lead to them.

Google is _not_ an "open source" company.
Understand that first and foremost!  ;)

Google is also a "fly by the seat of their pants" company with serious
retention and lifecycle issues, as well as standards compliance that
takes years.  That's because they are a web company.  They have to be
dynamic.

They are _not_ looking for 7+ years of ABI/API compatibility.  RHEL is
not a good candidate for them in most of their own infrastructure (with
exceptions, of course, or the appliances they sell -- there could be
some argument there, depending).  Heck, Fedora or any "packages" distro
probably isn't either.  They probably want to maintain their own.

If anything, Gentoo would be far better, because it's a source-build
system.  Gentoo removes the dependency, integration and regression
checking, allowing major endeavors to do their own, as they see fit, for
their products.

> Red Hat is not Linux.  Let that sink in.

Red Hat and Novell *ARE* the *ONLY* Linux vendors providing 7+ years of
ABI/API compatibility, subscriptions and SLAs, virtually the only *2*
vendors offering the same as Microsoft and Sun in that regard.

So let that sink in, that was _my_ point.

I.e., when it comes to rolling out a platform for 7+ years of duration,
CIO's think in terms of Red Hat and Novell like they do Microsoft and
Sun.  That's who I'm talking about in terms of "cost."

If you want to do your own backporting to maintain strict ABI/API
compatibility for 7+ years, and the related integration and regression
testing, that is also a "cost."  That cost is probably doable for
Google.  It might even be doable for a large government.  It's not for
most SMBs, and not for most enterprises who are not technology focused.

Google is technology focused, very leading edge.  It's in their business
model to do this.  Defense and financial are not, they have other
products and services to deliver.  The same could be said for many other
types of clients.

It's all about what you need, and what you are willing to pay for.  For
many clients, RHEL and SLES do _not_ make sense.  For many other
clients, it does.

> Red Hat might be price comparable with Windows, sort of.
> Look at *buntu.

Show me the Ubuntu release with 7+ years of support, including stringent
backports of fixes to ensure ABI/API compatibility?  That's what Red Hat
and Novell do in their products.

Before I say anything else, please do _not_ ass-u-me:  

- I haven't been a Debian maintainer, and a fan of APT-DPKG (or even
APT-RPM, or even SmartPM from the same people behind APT-RPM)

- That I don't run or support Ubuntu, Ubuntu LTS and/or am not
investigating Ubuntu Server, and general like Ubuntu (although the
Canonical-Spaceman rhetoric gets old, especially when its not true)

- Canoncial makes any money from giving away Ubuntu to Dell, which is
not sustainable, and requires the continued "Spaceman endowment"

- Canoncial really opens some bad indemnification considerations,
especially with Ubuntu, and it's the one area where Red Hat's stance is
virtually a "no good deed goes unpunished."  Novell actually licenses
things, although the OpenSuSE project has similar issues at times.

> Ultimately though, you've reitterated (or missed) my point.  IN THE
> US, cost is no object.  In other countries, it is.

Apparently you haven't been keeping up with Red Hat and Novell's sales
then.  You missed _my_ point.

You are re-iterating your thoughts.  I'm re-iterating my _exposure_ to
the market where Red Hat and Novell is used _internationally_ and why.

Red Hat's sales in China and several other markets are staggering, and
it's only increasing.  Today, over 50% of Red Hat's subscriptions are
_outside_ the US.  _Outside_.

Please, stop pulling assumptions out ... well, you know.  ;)

> That's why OLPC choose Linux.  The US isn't their market. 

The OLPC choose _Fedora_.  Wow!  You couldn't make my point better!

And need we add the Creative Commons?  And the countless other projects
now based on Fedora, or even the Fedora-CentOS developments?

BTW, Red Hat does _not_ "hate" CentOS.  Just understand there is some
real, legal history on why the trademarks aren't included.  Red Hat
tried, several times, to clarify "trademark guidelines" so the
"community" could use Red Hat(R) and the "bad guys" couldn't.

Unfortunately, at some point, when you're threatened in a way that any
company can destroy your trademark rights, you have to do something.

You want to compare Ubuntu to RHEL.  How about all of the Fedora that is
developed thanx to Red Hat?  How about all the contributions that _all_
distros benefit from in Fedora?  Other distros?

This is what I don't like about "distro pissing."  You basically argue
about what differs 1%.  If you really want to "get a 'tude," research
all of the GPL that Red Hat and Fedora e-mail addresses pump out.  It
will shock you. 

> Even if MS wants to compete with *buntu,

Again, what Ubuntu product is supported with complete ABI/API
compatibility for 7+ years?  End of story.

> and give away Windows, for free, to 1 billion citizens who can't
> afford it, how long can they continue that proposal?

The question is how long can the "Spaceman Endowment" do it for Ubuntu?
How long before Canonical runs into the same processes that SuSE did,
then Red Hat, then Mandrake, etc...???  And how does Ubuntu defend its
trademarks?  And what happens if it doesn't?

> Where I can run Linux 2007 (pick your distro) on the same 15 year old
> computer that initially came with Windows 3.11, the same can't be said
> for Vista.

Frankly, I'd like to see you run it on that same 15 year old computer,
with any software that is usable on the Internet today for a desktop.

The best you can do is RHEL 2.1 (now that SLES 7 is retired), which is
still supported.  But that only goes back to its RHL 7 lineage, circa
2000-2002.

> So in order for MS to play in the same league, they need to provide
> both hardware and software to the end user.

They sort of have that.  It's called the distribution channel.  It's why
most hardware only has Windows drivers that work for a few years of
Windows.  It's called the 2-3 year OEM/superstore upgrade model.

And in response to:  
> BS> Also, read up on South Korea.  ;)
 
> Pirating might be true.  I'll grant you that.

Again, I repeat ...
  "Read up on South Korea.  ;)"  

I'm not talking Piracy (which is actually worse in other countries in
the region).

> But look at the path to upgrade from it.  Is it easier to go to Red
> Flag, for free, and migrate the data to Open Office?

Is "Red Flag" really "free of cost"?  What does it provide?  Why would
Red Hat have booming sales in China.

[ READ THE FILINGS NOTE:  Also remember that Red Hat, unlike virtually
all other software firms, does not count "sales" until they are in
effect.  I.e,. Red Hat has 2008, 2009 or even 2010 subscriptions that
are not yet reported in their current FY figures.  Red Hat's public
financials are always an interesting read, and its model -- based more
on traditional manufacturing fiscal approaches -- is why its earnings
are on-the-mark, or updated to reflect new estimates, that are pretty
"on-the-money" in its history since switching to that model. ]

> or to upgrade to Vista and Office 2007.  Even in the US, where money
> isn't such an obstical, look at the adoption rate for Vista.  It's an
> ugly product, sure, but it's also a ton of useless training and
> learning for no real advantage.

This has typically been the case with any new, major Microsoft NT
release, such as NT 5.0 (Windows 2000) as well.  NT 6.0 is just
following that trend.

> Add to that the cost of hardware upgrades that I already mentioned,
> and MS's increasing piracy controls, and this is no longer true. 

Huh?  You're still on piracy?

I will stipulate that 90% of US consumers are stupidly programmed to
upgrade every 2-3 years.  But so are a _lot_ of other consumers all over
the world, including the increasing number in China.

But that still has _nothing_ to do with piracy.  In fact, I have no idea
where you got "piracy" from my mentioning South Korea.  South Korea has
another problem.

Again, please read up.

> Microsoft is about cost effectiveness.  If there is no cost effective
> market for their product, it written or sold, regardless of the demand,
> period.

Their "demand" comes from control of the distribution channel.

> Where you can add "Florida slang" to your KDE desktop, if you 
> so choose, you can't do that to Microsoft, and they never will,
> because the market is too small to ever justify it.

Er, poor example.  Try to use something real-world, like Florida labor
laws or US federal tax codes as examples of locale, which apply
differently around the world and different, local regions of every
nation.  Defense and communication security standards are yet another.

> I don't think that really matters,

Of course it matters!  You didn't think it mattered because you couldn't
think of a "real world" example.

> but I mention it to point out that Microsoft IS nation centric.
> They will accommodate nations that provide income to their business.
> And increasingly that is US North America.  Even the rest of the
> Americas are walking away, albiet slowly.

Again, you have shown no statistics other than assumption and a pointer
to a few companies (and some questionable sites at that, which makes me
think you used Google).  I don't know where you come from.  I'm not
going to assume.  But please, I don't really care for "advocates" who
feel like the have to "prove" adoption, or think Microsoft "dominates"
the core infrastructures in the US only because they have money
(especially when they do _not_ "dominate" the "core infrastructures" in
the US in the first place -- a poor assumption).

> Agreed.

Okay, at least we have that.

> I'd point out that Gnome too, is mostly paid corporate developers,
> whereas KDE is mostly volunteer.

Because virtually the entire former X/Open and CDE consortium of
partners have moved to embrace GNOME and its object model.  Naturally,
that lead to more commercial involvement.

But that included Sun's work on accessibility that hit GNOME 2.0 and won
the Hellen Keller award.  And Sun has contributed significant amounts of
GPL in the past.  Sun has also been more "open systems" than most other
commerical UNIX vendors as well.

And they are finally starting to "get it" on GPL in general too.
 
> Does this mean RH is wrong to use Gnome?

Core contributors in Red Hat/Fedora long ago came out and said their
unintentional preference for GNOME is a combination of the licensing
issue history and the resulting staff.  They always stated their lack of
attention to several KDE aspects wasn't intentional, but they fully
admitted there was a clear bias as a result.  I don't know how much more
honest they could be.

I know a lot of rhetoric creators like to play the GNOME v. KDE games.
Frankly, every GNOME and KDE developer I've personally or professionally
known don't understand why people even do that.

> Win is total domination.

Is that a joke like Linus' quip "Global domination"?
Or are you actually serious?  Then that would be a good joke!

> Really?  Open Colaboration?  Like what?  I'd like to start a Ms Office 
> on Linux project.  Who do I call at MS to start that collaboration?

Actually, MS Office can_not_ be ported to Linux any easier than it
hasn't been for MacOS throughout its entire history.  You have the
x86-based ignorance and data alignment issues.  This is why x64 versions
of Windows still ship with a crapload of Win32 libraries with no Win64
equivalents (left to software developers to include for themselves).
Portability is a joke in 99% of Microsoft's own applications.

> How about Instant Messaging Protocols?  (MS isn't alone here,
> admittedly).

Actually, most are standardizing now.

> MAPI standards?

What is "MAPI"?  It's a prime example of a piss poor protocol design
that has been hacked and over-hacked.  Even Microsoft dropped it,
although it still provides it for 3rd party compatibility, largely
because it sold licenses to those 3rd parties it must continue to honor,
legally.

Microsoft still does _not_ have a _true_ server collaboration stack
either.  That's my #1 issue with MS Exchange.  Always has been.  Always
will be.

> You know as well as I do that there is no such thing as collaboration
> with MS, and you even alluded to it earlier when you talked about
> partnering with them.

Agreed.

> Open collaboration == we win.
> No collaboration == MS wins.

You're still making it about Microsoft v. Linux (or whatever you define
is the "community" and how).  I'm telling you it's far more gray.

And it's not about "winning" at all.  It's about "costs" and "risks" and
other things.

> Does that more or less answer your question with your own statements?

Nope.  Because you didn't recognize the fact that integration and
regression testing is still a "cost" that open source has to deal with.
That is just a law of the software engineering model.

Heck, NASA proved it to be the case with COTS on the Mars Polar Lander.
They slashed the budget 90%, including QA.  What they didn't realize is
that it doesn't matter what you use and what savings you get, you still
have to put the same amount of integration and regression testing in.

Regressions are why vendors backport.  Because while Google may be a
dynamic company that changes software every minute, banks want to run 7+
years.  Heck, some are asking for 10 years of ABI/API compatibility.

> From an investor perspective, that might be true, but you've asked
> several times to look at things from a technical perspective, so why
> change here?

Because you said Microsoft was dying.

> I don't care about their stock market performance,

Huh?  I don't think you even remotely got my point.

> and I think it's completely (and obviously) different from their action
> as a technology company.

Like IBM?  Like countless other firms?

Wow!  You really missed the point there.

> If they didn't care about their proprietary tech, they could open
> source everything they own (like RH or Postgres) and not lose.

Huh?  You mean Microsoft actually has code that adheres to its own
proprietary standards that can be easily implemented if they'd "just
open up"?

I think you _missed_ my prior point that Microsoft doesn't even have
"proprietary standards."  Heck, Samba's a great, self-documenting
example of that reality.

> But they can't and won't.  Because they'd immediately become
> irrelevant, and nobody would do business with them anymore.

No, because they can't even understand their own software internally.

That's the biggest problem with NT 6.0.  They've had the greatest brain
drain during any time of the NT lineage during NT 5 -> 6.

But even before that, their own application developers (heavily
outsourced and/or contracted), don't even follow their own APIs.  That's
why you get different formats sometimes from even a new build of Office.

It's also why the MS Office for Mac team curses them.  The Mac team can
import fine, as they build a good import module.  But when they save,
they can't predict the often exact same binary writes that MS Office for
Windows assumes.  It's the same reason with OpenOffice.org/StarOffice as
well.

RTF is a joke (don't get me started).  It's why the DOC (Word) import
module is used when reading RTF in every program as well.  Microsoft
doesn't even follow its own RTF specs, not just rev'ing every time, but
it has embedded DOC tags in its as well.

There's a reason why OOXML came about.  It's called standardization and
portability for just Microsoft itself.  Unfortunately, the OOXML spec
was developed independently of MS Office 12 (2007), and that's why they
don't match either.  I'd really hate to be on the Mac port.  ;)

> They will have to, and I agree they will.  The point is, they will
> delay that as long as possible, because they generally do not
> provide end users with anything close to a compelling reason to
> stick when them.

But they control the distribution channel, and that is powerful.  It's
why many hardware vendors do not release specs as well.  Only select,
large companies can afford to, and even then it can be limited.

They like the forced upgrade model that Microsoft partnerships bring.

> This is why IP should have FAR less legal protection than it
> currently has.  But yes, you're right, they could.

I disagree.  I think IP shouldn't be granted for simple things.  Too
many companies brand themselves as "IP holders" these days.  Big problem
with that is that countries like the Chinese just ignore all of them --
the simple with the major.  It only benefits lawyers in the US.

If there is one area where I agree the US is "the money," its lawyers.

> I bet those companies see a win vs lose perspective.

No, they see a risk and profitability perspective.  This "win" is part
of the advocacy problem.  It doesn't sell CIOs on it.  You scream
Microsoft, but there are other proprietary software than Microsoft (of
which, I've repeatedly pointed out that Microsoft isn't even
proprietary).

A lot of proprietary software has value.  In fact, I think the labeling
of Microsoft as proprietary is an insult to proprietary in general.  And
I say this as a huge proponent of GPL, and a major, long-standing
"apologist" on Red Hat's dedication to GPL.

> IBM makes money on hardware.  REALLY GOOD hardware.

Yes, although services as well.  Although there was a time they did not.
And, frankly, on the more PC end, I'd argue HP can do better.
Especially since HP is still far more of a real engineering firm than
IBM in that regard (even before the Levono sale).

> The most recent server I bought was an x460 for a hige ESX cluster.
> With the capability to support 32 cores, and hundreds of gigs of
> ram, this was a bigger box than Windows could even use.  There was
> no competition for it.  There might be now, but there wasn't then.

Intel only supports 64GiB or, in the latest revisions, 256GiB.
AMD supports 1TiB, or in the latest revisions, 256TiB.

> IBM "sells" several OSes, such as the AS400, the S390, etc.
> Each has it's own place.  They make much of their own hardware.

Yes, I know.  And I've still supported HP VMS as late as last year.  We
can talk about that, but all that's non-commodity and has little to do
with Linux.  Sans the fact that, for the longest time, IBM used to
"fight" HP Linux/x86-64 solutions when it "intruded" on IBM
Monterey/Power.  So IBM is our "partner," not necessarily our friend.

> They sell end user software for desktops and servers of many sizes
> and flavors.

Unfortunately they have long outsourced a lot of it, just like their
hardware designs too (again, even pre-Levono).

> Windows has dropped support for hardware down to the x86 or the
> extended x86 line.  I don't believe they even support a true 64 bit
> processor any longer.

The _never_ supported NT in 64-bit.  It ran 32-bit on Alphas.

It's only the IA-64 that they ship a true Win64 stack.  Even the x64 is
still heavily Win32.

> Certainly not RISC.

"RISC" is abstract.

> Their ventures into the embedded realm are half hearted at best,
> and consistently lose to Linux or other OSes when sold against
> another. 

Microsoft targets user-interface embedded.  They get their ass reamed in
traditional UNIX areas where Wind River VxWorks and Linux dominate.

[ Again, be careful, I went back to embedded for much of my last 2 of 3
years, including working at 1 of the 2 industry leaders last year.  ;) ]

> Look at the Supercomputers. 

Define "supercomputer"?

There are differences between simple computational clusters,
psuedo-shared memory systems using InfiniBand (typically via HTX or
PCIe) and then true, shared memory solutions.  There are also
interconnect considerations, especially where Intel loses to AMD, and
massively, at those larger, more inter-CPU and I/O bound aspects.

> Look at the older computers.

Doesn't matter if they don't run software that will work on the modern
Internet desktop.

> Look at the smaller or more portable computers. 

Actually, Microsoft runs well there.  Anything with an user interface.

> IBM plays in all of these places.  This is diversification.

IBM actually doesn't.  That's a misconception.  IBM partners with
providers of those solutions.  I think you're listening to a bit too
much marketing.  ;)

Now IBM as a foundry, that's different.  IBM's Power (please be a dear
and don't call it "PowerPC" ;), dominates performance embedded there.
They are the foundry of so many solutions.

But that's different.

> MS needs a desktop.  They do not play anywhere else.  That's not
> diversification.  Even Cisco is better diversified.

No, Cisco and Microsoft are core partners who complement each other with
their diversifications.  IBM also still makes a lot of money providing
Cisco and Microsoft solutions as a partner.

> Suse is Novell's last real kick before death.

Not according to some of their sales in many areas.

Although SuSE does fork and maintain way too many patches that aren't
upstream.  Red Hat has always had the right attitude on that.  And where
Red Hat strongly differs with mainstream, they just throw as many
developers as they can in the mainstream and there is often a
compromise.

> Yes, open sourcing Yast was a nice move, but by the time it happened,
> there were other, and better tools.

Sigh.  YaST was just a _small_ portion.

I don't think people realize how "protective" SuSE was with all sorts of
things.  E.g., there's was a reason why there wasn't a CentOS equivalent
for SLES during that time.  Johnny can tell you about that himself.  ;)

> Who cares about the investors?

Because they help fund a crapload of the GPL you use in your Ubuntu
system.  ;)

Again, you compare Ubuntu to RHEL when it suits you, and ignore Fedora.

> Again, why does wall street matter at all if you want a technical 
> discussion

Because the development of GNU/Linux is inter-twined with the ability to
employ developers.  Yes, there's a lot of "scratch an itch" type
development in the open source world.

But if you trail back to why Red Hat was a "troublemaker" on GLibC 2,
ANSI LibC++, NPTL and, now, SELinux, you'll eventually discover why.
It's more to do with ABI/API longevity, longer term interoperability,
and missing components in a pure, GPL GNU/Linux stack.

It's also why they bought Cygnus.
It's also why they bought Sistina.
It's also why they bought iPlanet.
It's also why they bought JBoss.
It's also why they bought Metamatrix.

You think investment has nothing to do with Linux technology?

Again, I'd like to know what you didn't like about RHL 7-9.  It's very
likely you're relying on that "trailblazing" done by Red Hat right now.
Same deal with those who complained about RHL 5 too.


-- 
Bryan J  Smith              Professional, Technical Annoyance
mailto:b.j.smith at ieee.org  http://www.linkedin.com/in/bjsmith
-------------------------------------------------------------
           Fission Power:  An Inconvenient Solution