On Sep 15, 7:49 am, Roy Schestowitz <newsgro...@xxxxxxxxxxxxxxx>
wrote:
> Microsoft to 'inter-operate' with open source developers
>
> ,----[ Quote ]
> | "We want to compete with the products of open source," he said.
In the past, when Microsoft has said "we want to compete with X" that
has ultimately meant "We want to completely dominate the market and
completely eliminate X from the market".
Nothing in the article seems to make my think anything has changed
significantly.
Microsoft's quickest and easiest path into the Linux market, should it
flower, is the WINE library. PC Buyers who purchased XP Professional
were allowed to use Windows XP libraries and call them from WINE.
Some of the "bridge" technology was provided by CrossOver, who
licensed libraries from Microsoft.
Microsoft is losing Mind-Share. Developers are switching to Java, Web
based technologies, and Linux "compatibility libraries" such as
cygwin. Developers like the ability to write one version of an
application that runs on Linux, Mac, Windows XP, AND Vista. Microsoft
has dedicated itself to maintaining the support of Windows
exclusively. Even Mac support has been grudgingly given and even then
only under strict nondisclosure and only using Microsoft's proprietary
libraries.
Microsoft has not been kind to third party developers in recent
history either. The two big industries remaining with retail presence
- Security software and Game software, have also recently found
themselves being "locked up" or "locked out". Vista was designed to
destroy the security market. XBox was designed to "Lock in" the video
game vendors, forcing them to give up the highly portable OpenGL
standard and implement using the "Microsoft Only" DirectX standards.
It backfired.
> | To show its willingness to collaborate with
> | open source developers, dela Cruz
> | said the company is working with Novell Inc.
> | for technical cooperation in the
> | establishment of an open source interoperabi-lity laboratory.
> `----
One of the problems that Microsoft has had recently, is that there
have been a number of products, libraries, and tools that were
dirivative works based on BSD software, but the patches have only been
provided for GNU versions of the software, and only released under the
GNU licenses.
Microsoft needs these patches because they address known
vulnerabilities that can still be exploited.
> http://www.sunstar.com.ph/static/ceb/2007/09/15/bus/microsoft.to.inte...
> "Compete" -> "collaborate", "technical cooperation" (tax and exclusionarty).
> Yes, open source is liked at Microsoft, according to this article. Like
> OpenXML... or Silverlight/Moonlight... or OSI... Port 25... or shared shared
> source being called "open source" in the press.
> They pretend to be what they
> want to destroy or subvert.
Microsoft will try to "Embrace" Open Source, the same way a python
attempts to "embrace" a gazelle. The problem is that OSS is more like
a porcupine. Microsoft would try to "extend OSS" the way the python
extends it's jaws to swallow the Gazelle. But again, OSS has that
porcupine nature which makes it nearly impossible to swallow, without
killing the snake.
It would be interesting if Novell decided that instead of a $7 million
cash settlement (the net capitalization of SCO), they decided to
"settle" by declaring that ALL versions of ALL Unix source code was
now to be released under GPL. The proprietary licenses would still be
available, but getting patches and support that wasn't GPL'd would be
nearly impossible.
> Microsoft Linux? HAHAHA, er, wait a second...
> ,----[ Quote ]
> | What would happen if Microsoft were to come clean about
> | Vista, what a stinker it is and decided to base the next
> | version of Windows on a core of FreeBSD?
Probably not FreeBSD but a BSD variant.
Microsoft has realized that they need to do major rework on the kernel
because it is too slow and inefficient compared to Linux and OS/X.
They have had a project, I think it's code-named "Synergy", which is
essentially an attempt to create a port of the BSD kernel in C#.
According to one video prepared by the development team, it will be
available around 2017. If Vista were to bomb the way ME did,
Microsoft might give the project more funding and attempt to
accelerate that kernel.
It steals, I mean borrows, many of the concepts introduced by Linux,
including the use of memory mapping rather than direct mapped memory
allocation, the mapping of "big buffers" and "small buffers" to reduce
the amount of garbage collection, the allocation of most free memory
to system functions rather than to huge application library buffers.
The use of memory paging to pass messages between processes.
This would result in the ability to create smaller and more efficient
simple programs which were connected using streams or message queues
(most likely MSMQ instead of SysV IPC Message queues).
The use of these smaller loosely coupled components would facilitate
faster and simpler applications and workload could be spread across
the cores of multi-core processors.
Linux already has all of these features and functions, and has been
using most of them since inception. When Linus built Linux, he wanted
to play with the 80386 Memory Management Unit. He was intrigued with
the possibility of using memory management instead of manually copying
buffers or using DMA. Using memory mapping, you can move large
buffers of 2K or larger in just a few instructions.
Microsoft reviewed the source code of Linux and found that one flaw
was the use of spin-locks to coordinate processes. IBM proposed the
use of message queues instead of spin-locks, which it has been doing
in it's mainframes since the late 1960s. Today, Microsoft jealously
restricts the publication of benchmarks. Does anybody, even a
WinTroll have a recent benchmark between Windows XP or Windows 2003
and A Linux 2.6 kernel based distribution?
A few "unnofficial" and "confidential" benchmarks indicate that Linux
is almost 50% faster than Windows on several "standard" benchmarks.
Unfortunately, these are unofficial and confidential, and I didn't run
them myself, so I couldn't vouch for their accuracy anyway.
Microsoft has published one benchmark designed to exploit allocation
differences between NTFS and Ext3. NTFS allocates by cluster. By
formatting the disk in large clusters, and using cluster sized
objects, the performance of NTFS can be made to look much faster than
Ext3 which uses index nodes that allocate by i-node, but work in
allocations of about 1kbyte or 1/2 kb at a time. Using random-access
writes to force fragmentation of the indexes, (a very atypical use of
Linux), Microsoft was able to craft a nonstandard benchmark that made
Windows appear faster on this specific benchmark.
My guess is that you won't be seeing any benchmarks between Linux and
Vista either. Vista is very memory hungry, and several applications
written in C# are having garbage collection issues.
> | Why not have a branded version of SLED?
> | See the previous discussion about the GPL.
Microsoft could use Linux, even with GPL. They would just
have to provide an interface to the LGPL libraries such as GLIBC.
The WINE library is one example of such a library.
Keep in mind that Microsoft has licensed certain Microsoft code that
allows many Windows applications to run on Linux. In many cases,
Microsoft makes more money on the compatibility licenses than they
make on eth OEM licenses. In most cases, the Microsoft libraries
themselves can only be used on a machine licensed for Windows XP, such
as the OEM license of a DELL or HP PC. This may be one of the reasons
that even when a PC is designed to run Linux and isn't designed to run
Windows that well (OpenGL card, Centrino or Athalon processor, Intel
or Cisco WiFi cards...), are still purchased with Windows XP even if
the user ultimately intends to run Linux.
Microsoft attempted to change the Vista license and forbid the use of
Microsoft libraries with anything other than Vista. The result has
been that many corporate and individual customers have said "No
thanks, I'd rather have a Mac". This has forced Microsoft to allow
OEMs to offer XP licenses even though the licenses were originally
sold as Vista licenses. The Customers want XP software and XP license
terms.
> | Whereas the BSD licenses are basically "Give credit and don't
> | pretend you wrote what you didn't" and they can use BSD code in any
> | commercial product they want to.
BSD was really the first OSS license. It was originally used to
prevent corporate users from trying to sue UCB and to keep UCB from
being a co-defendent in lawsuits. These earliest licenses date back
to the late 1970s, and are almost as old as the Copyright License
itself. Remember, prior to 1977, the copyright was subject to "fair
use" which could be determined by Judges. Judges could decide that
taping a full-length picture off of HBO and watching it several times
with friends who brought beer and food, was "fair use", while a music
publisher could sue a church for allowing a church choir to sing a
song that they had learned by rote.
Prior to 1977, software publishers relied on EITHER copyright, or
nondisclosure agreements. Many military contracts included software
which was classified, not because they were afraid that the USSR would
get it, but because vendors didn't want their competition to get the
code. Code not protected in one of these two ways became public
domain.
Prior to 1977, software publishers tried a number of ways to make
their works copyrightable, including publishing hex code listings,
assembler listings, and even the use of various bar codes and paper
tape images.
Microsoft owes it's success to their profound understanding of the new
provisions of the revisions passed in 1976, that took effect in 1977.
Because Bill understood the license provisions of the new law, he was
able to create practical licenses that were very effective in court.
> http://opseast.wordpress.com/2007/09/13/microsoft-linux-hahaha-er-wai...
> It's a recurring theme now.
|
|
