Slashdot Mirror
META Predicts Linux Software From Microsoft in 2004
trandles writes "According to this story at NYTimes (FRYYY), META Group is reporting that Microsoft will begin selling Linux software in 2004. It also goes on to report that a META Group study comes to the same conclusion as the earlier (MS-funded) IDC study that Linux has a higher TCO than MS solutions for some applications." Remember, this is speculation on the part of META, and has to do with back-end software, not Office. (But if Microsoft wanted to, they could become the world's biggest producer of Linux software.)
All they need to do is create a free (as in beer) X-semi-compatible, but faster GUI. Then release Word for it.
Embrace, extend, control. After a while, everyone will write software for Microsoft X# or X++ or X-Windows(tm) or whatever they call it, and MS will call the shots.
Admittedly, if Microsoft thought that OpenOffice or any other office suite on Linux or other OSes represented serious competition to Microsoft Office, all they would have to do is port Office to Linux and they would own the office suite market, but at the expense of their OS monopoly. The only reason Office for Macintosh exists is to keep the DOJ, the FTC, and the courts off their back.
My journal has hot
Windows Solitaire was written by the well-noted eccentric Wes Cherry when he was an intern at MS in the 80s. He wrote it while goofing off one day, a manager spotted it, and said "we've got to put this thing in Windows!"
As more and more of Microsoft's software is built on top of .NET it will become increasingly easy to move that software to other operating systems.
Microsoft's not yet ready for the "join 'em" part of the "If you can't beat 'em..." argument - especially as today's Wall Street Journal has a very long, detailed article on Microsoft's efforts to lure national governments away from open source software, using carrots and sticks familiar to many /.ers. It's worth reading, and good to see the mainstream press like the WSJ taking an active interest on how Redmond deports itself.
... so here (for review purposes only) are highlights of the article - well worth the time:
It's a good piece, but it's subscription only
Microsoft Wages Campaign Against Free Software
By WILLIAM M. BULKELEY and REBECCA BUCKMAN
Staff Reporters of THE WALL STREET JOURNAL
Sometimes it seems as if Microsoft Corp. doesn't want government to save money -- at least not if it comes by using free software. Microsoft is waging a major lobbying and public-policy campaign to stop government agencies in the U.S. and abroad from embracing free, "open-source" software, especially the Linux operating system, which poses a growing threat to Microsoft's Windows.
In the past year it has argued with the Defense Department over the content of a report extolling free software. It has organized a world-wide lobby to oppose laws that mandate using open-source software. It has persuaded some congressmen to ask the new Office of Homeland Security not to fund research that uses certain open software.
But even Microsoft is having a tough time persuading governments from Washington to South Africa that getting software free is a bad thing -- especially when rivals like International Business Machines Corp. are telling them that open-source software works just fine.
Open-source software is software whose source code, or base layer of commands, usually can be copied freely and then modified, unlike most proprietary software, which is generally controlled by a profit-making company. It is championed by a far-flung community of programmers, researchers and companies who share their work over the Internet.
Open-source software has grown in recent years to become a full-fledged rival to Microsoft, used by companies, universities and others in their computer rooms. Many open-source programs are free, or nearly so.
The best known open-source software, Linux, increasingly is being embraced by computer companies including IBM, Dell Computer Corp. and Hewlett-Packard Co. as a way to sell more hardware and services. According to International Data Corp., a technology-research firm, sales of server computers that use Linux grew 6% in the most recent four quarters, while sales of Windows-based servers grew just 1% in revenue.
Microsoft says it isn't against the concept of open-source software. But it is working hard to prevent government researchers from adopting software covered by the general public license, or GPL, that governs reuse of much open-source software, including Linux. The GPL requires anyone who copies the software to freely share any improvements or additions they make to the code.
Because commercial companies often adapt programs written by government-funded university scientists, Microsoft argues that wider use of GPL-licensed software would stifle innovation. Commercial companies, it argues, would have no incentive to sell "free" software derived from the research. What's more, Microsoft worries that its own developers could inadvertently combine Linux or other GPL-licensed programs with Microsoft programs, which could potentially make the Microsoft programs subject to free-sharing as well.
"The GPL, in my view, is bad in all its dimensions," says Jim Allchin, the Microsoft group vice president who heads the powerful Windows group.
In some cases, Microsoft has leaned on government agencies directly. The U.S. Defense Information Systems Agency, an arm of the Defense Department, says that last spring it granted a Microsoft request for an exclusive advance look at a report by research firm Mitre Corp., Bedford, Mass., on Pentagon use of open-source software.
After Ira Rubinstein, a Microsoft lawyer, detailed Microsoft's objections, Dawn Meyerrick, chief technology officer at the agency, says she asked Mitre to make changes in the report. Among them, it dropped the conclusion that open-source software was more secure, and it added cautionary words about the GPL.
Open-software advocates also perceived Microsoft's influence in a letter from a group of congressmen to Richard Clarke, who heads cyberspace security for the newly created federal Office of Homeland Security. The initial letter urged the government to continue past practices by "explicitly rejecting licenses that would prevent or discourage commercial adoption" of software developed under federal contracts.
But as the letter was being circulated, Rep. Adam Smith, a Washington Democrat -- who receives the most donations of any representative from Microsoft's political action committee -- added a "Dear Colleague" letter to further explicate the original. That letter said that "licenses such as the General Public License (GPL) are problematic and threaten to undermine innovation and security," and suggested such open-source software shouldn't be developed by the government at all.
That echoed Microsoft's position. A Microsoft spokesman acknowledges that Rep. Smith met with its chief technology officer, Craig Mundie, before the letter was sent, but only for "informational" purposes. Mr. Smith's press secretary says that the "dear colleague" letter was meant to clarify the original because "we believe in innovation."
Open-source fans believe Microsoft is bringing its political power to bear because it sees a market threat to its desktop-software monopoly. But in some cases, Microsoft's appeals have fallen on deaf ears. Last year, according to people familiar with the situation, Microsoft objected "vigorously" when the super-secret National Security Agency developed a secure version of Linux and then posted it on the NSA Web site for anyone to download. But NSA didn't back down and the software is still available.
In the developing world, where free software like Linux may have its greatest appeal, Linux advocates say they have "noticed that Microsoft has made a substantial portion of their quote 'gifts' to developing nations that have indicated a strong preference for open-source software," says Mark Webbink, general counsel of Red Hat Inc., a Raleigh, N.C., company that sells versions of Linux.
In India, where at least one state government endorsed Linux recently, Microsoft Chairman Bill Gates last month announced a $400 million gift of donated software and business-development aid.
In South Africa, a Microsoft offer to provide software for 32,000 schools came just days after that country's National Advisory Council on Innovation called for the government to adopt open-source software to build local programming skills and avoid sending hard currency to the U.S. to pay for Windows. Nhlanhla Mabaso, a government chief information officer, says that while the free software from Microsoft is tempting, "Personally, I believe this is not good for South Africa."
Bradford Smith, Microsoft's general counsel, says any donations "are made to meet a social need" and not to counter Linux.
Microsoft concedes that its opposition to open-source software has sometimes backfired, and it says it intends to move the battle to more straightforward commercial issues.
* * *
"Inferior/superior" are such loaded words. Let's try to be a bit more objective. The design of NT is more "advanced" (incorporates more recent design concepts) than the Linux kernel (which holds to older and more "traditional" systems software designs). It is more "sophisticated" than the Linux kernel (again, a microkernel design where the "monolithic" part of the kernel handles only processes and interprocess communications) and thus permits a more modular OS that can be transformed, expanded, changed, with theoretically much greater speed and much smaller chance of unwanted side effects from those changes.
All of these advantages are theoretical. They also carry some disadvantages, notably in performance.
People have argued for years that traditional monolithic kernels must reach the point where network effects in the code make every change too expensive to make (too much chance of side-effects). So far, this hasn't happened with the Linux kernel. I'm only an amateur kernel hacker (working on reverse engineering an old framebuffer video card so I can make a kernel driver for it) and I don't claim to know much of the Linux kernel code (apart from the framebuffer), but it sure looks to me like the Linux kernel has managed to acheive a similar level of code independence by using good structure programming practice.
As for the superior "security" of the microkernel model, this comes from that same separation of service processes. Compromise a microkernel service and you cannot (in theory) leverage that into a compromise of other services. At the service level, this is true. A Mach or NT microkernel has this feature. The problem is that the Windows kernel isn't the Windows OS. The Windows OS is the gigantic flabby shared APIs written as DLLs. These are the things that are attacked and compromised. These sit on top of all of the sophisticated kernel plumbing and they provide a path to blithely leap between unprivlidged and privledged user space in the stuff that matters: network, file, and directory services. You don't need to compromise any kernel service to own a Windows box because it the is the Win32 APIs that have the holes, not the kernel services (Yes, I know I'm making rather broad generalizations, but the point is still true). Much of "Windows" privledge/authentication/authorization is in "user land" code. Microsoft emphasizes the sophistication of their underlying microkernel architecture. And I agree. The trouble is they have carried over the top much of the cruft from the design of the win16 and pre-NT win32 APIs in the name of backward compatibility and this has carried forward fundmental design weaknesses from those systems.
To be fair to Microsoft, these weakenesses weren't particularly problematic when they were introduced. At the time, each machine was an isolated, single-user device. Little or no networking was done with them. Also to be fair, Microsoft really didn't have a choice but to be backward compatible. They never would have got any users for NT if it didn't run all or nearly all existing software. I'll go further: they never would have got ISVs to switch to win32 if they hadn't done Windows95, marketed it like the second coming, and required ISVs to use win32 if they wanted the "Works with Windows95" logo. The much-maligned Windows 95 was the only reason every major piece of Windows software came to run well on NT.
So feel free to bandy the words "inferior" and "superior" but I defy you to provide and objective criterion by which you may fairly apply those words to the two kernels. "More advanced" v. "Less advanced", yes. "Sophisticated" v. "Simple", yes. I don't buy "superior" v. "inferior," unless you believe that newer necessarily means "better," which, obviously I do not accept.
I also do not buy the statement that these weaknesses "are a thing of the past." They have done a great job of cleaning up many of the holes, but the DLL hosted APIs are still a bridge that just circumvents the good kernel design. They have plugged thousands of holes, but the system design is still subject to them, just as the Linux kernel is.
I do agree that 2000 and XP are vastly more stable than any previous versions of Windows, but this is a product of API cleanup, not the inherent "superiority" of the NT kernel. The "NT" kernel has had these "superior" features from day one and it conferred no magic "superiority" or stability on early versions of NT.