[SLUG] RE: OT: M$ deals the final blow

From: Bryan J. Smith (b.j.smith@ieee.org)
Date: Fri Sep 10 2004 - 17:25:55 EDT


From: slug@nks.net [mailto:slug@nks.net] On Behalf Of Pete Theisen
> Hi Ken!
> What will you bet we see a Linux distro from the Great Satan in the next
> 5 years? Yes, they will charge for it as do a number of others, but dammit,
> your printer, sound card and modem will work.

That would require GPL incompatible extensions to the kernel, and open
Microsoft up to a lawsuit by all copyright holders on the kernel.

Understand that Linux developers ship 100x as many drivers as Microsoft
itself does. Microsoft merely includes drivers from OEMs that release
them. It has a formal signature process now, but the OEM still does
"all the work." OEMs can released binary-only drivers in NT.

The problem with drivers is that they must be GPL. As such, the source
code must be published. Publishing the source code of software-only
drivers is largely impossible. Not because of the OEMs, but because the
OEMs typically license from a 3rd party. Since that source code is the
"crown jewels" of that 3rd party, they are not about to see it
disclosed.

Ironically, with all these OEMs licensing the same 2-4 3rd party source
code, if they'd only switch to the full Linux model, they wouldn't have
to license. Thus far, only HP and a few others have signed up -- not
even IBM ironically, except where it sells their big items, because even
IBM outsources much of its system development.

The only way "around" the GPL is to have the end-user build the driver.
If the end-user builds the driver, and only uses its internally, that
abides by the terms of the GPL. So an end-user can built a non-GPL
driver -- which is how some of these approaches work (e.g., nVidia,
Intel, etc...).

On Fri, 2004-09-10 at 16:16, Ken Elliott wrote:
> Under SuSE, they already do...<grin>

Just remember, any Microsoft port of MS Office to Linux will have as
many document porting issues as MS Office for MacOS does as well. The
problem isn't so much sending to Mac, but sending back. Most MS App for
Windows software developers don't concern themselves with data
alignment, whereas MacOSX developers have to.

> I used to think of Novell and IBM and the greedy, controlling players.
> Microsoft (at that time) was open and friendly. They gave their user
> interface to OSF for use on Unix, and NT Server 3.5 was for _unlimited_
> users. Sadly, that has all changed as the more profit-minded guys took
> over.

Or as I said earlier, "NT became DOS' bitch" about 10 years ago. The
same has yet again happened with Longhorn. ".NET has become [bastard]
Win32's bitch."

> I have no doubt you'll see a new and improved Linux from Microsoft,
> featuring .NET, and Windows-look-alike GUI, ability to run MS apps (at least
> .NET) and the ability to connect to NTLM shares. All for the low low price
> of .... About the same as Windows....

Not unless Microsoft just repackages what people already do. Mono works
quite well right now. They'd just leverage Novell's work. Heck, they'd
probably license it just like they re-licensed Java.

Microsoft can't remove binary and API compatibility. They also don't
have the manpower/capability (yes, this is _very_true_) to re-invent
everything atop of GNU/POSIX. They could only use what others have
created.

So I consider the effort impossible.

> I suspect the .NET framework and NTLM access

The two are distinctly different.

CIFS/SMB (NTLM is only considered the technical term for authentication)
is still very much "Chicago bastardized" Win32.

The .NET framework has been reduced in Longhorn to what Microsoft is now
calling WinFX. Namely, the Avalon desktop and Indigo Internet Services
over Win32, the GDI and IIS-IE respectively.

Avalon will bring about the same requirements of Win32 that caused NT
4.0 to be "cracked." Indigo will suffer the same fate of a Java sandbox
atop of Win32 -- it just can't be done securely.

Win32 is here to stay. Microsoft cannot even attempt to change. If
they tried to ship a "pure .NET" OS like they tried to with NT for Win32
prior, it would only suffer the same fate and cost to develop without
return.

> will become the new MS-Tax that can be applied to any OS, not just
> Windows.

The only way Microsoft will try to tax people is with patents. They
have been hijacking patents in their licensing agreements. That's
probably the scariest issue right now.

The overwhelming majority of Microsoft's Win32 codebase only runs on
Win32 -- IA-32 (x86). Even the XP 64-bit Edition is completely 32-bit,
except it puts the x86-64 processor into "Long" mode. Nearly all the
libraries and subsystems are 32-bit, and use Win32-on-Win64 (WoW) at a
performance hit. It's is technically infeasible for them to develop an
NT release with full dual-32 and 64-bit libraries like Linux.

That exercise is left to programs who want to ship their own 64-bit
support. Otherwise, even if they are 64-bit, they utilize almost
entirely 32-bit libraries.

Microsoft Win32, by its very nature, is not 64-bit clean -- let alone it
is data-alignment ignorant and stuck at x86. That is a continual
problem. Visual Studio through version 6, still accounting for some 90%
of the vertical app development on Windows, uses DOS Int20-3Fh, Win16
and direct Win32 calls that bypass the NT model.

Apple maintains a full port of MacOS X in-house to x86 and x86-64 to
avoid any future porting issues. Sun has also done the same with
Solaris, and will likely be thanking itself very soon. GNU by its very
nature is very portable -- and anyone who writes GNU software learns who
it is important to do so very quickly once someone ports it to non-x86.

> And it seems to make good business sense to me - any MS.NET app will
> run on any OS with the proper MS.NET framework installed.

The problem is that Microsoft has basically "given up" on .NET other
than a basic application technology. It still requires massive amounts
of Win32 underneath. Even supposed "pure .NET apps" still have Win32
VM/scheduler tie-ins and x86 requirements -- so much so that
WINE/WINELIB is used in Mono to emulate these portions.

> Why not let all the open-source guys develop their next OS for them?

That's exactly what Novell is doing. After 10 years of screwing around
with UNIX, they finally got it.

I'm sure Ray Noorda fired off an e-mail, "I told you so 10 years ago."

There's nothing wrong with value-added atop of Linux. Most commercial
Linux distributors do this. Leverage commodity, building a specialized
solution atop. It's a very smart business model.

The only major commercial Linux distributor I know of that does not is
Red Hat. Red Hat not only releases all source code of anything it
writes, even original or bought out products, but it almost always does
it GPL. With the development revenues from their former Cygnus
division, maintainer of the GNU Compiler Collection (GCC) which they
offer dual-GPL/commercial licenses and support, and newer revenues from
their re-structured commercial offerings, they don't have to value-add
in their view.

-- 
     Linux Enthusiasts call me anti-Linux.
   Windows Enthusisats call me anti-Microsoft.
 They both must be correct because I have over a
decade of experience with both in mission critical
environments, resulting in a bigotry dedicated to
 mitigating risk and focusing on technologies ...
           not products or vendors
--------------------------------------------------
Bryan J. Smith, E.I.         b.j.smith at ieee.org

----------------------------------------------------------------------- This list is provided as an unmoderated internet service by Networked Knowledge Systems (NKS). Views and opinions expressed in messages posted are those of the author and do not necessarily reflect the official policy or position of NKS or any of its employees.



This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 17:17:51 EDT