Microsoft Trying To Appeal to the Unix Crowd?

DigDuality writes "With the news that Windows 2008 (recently discussed on Slashdot) will have GUI-less installs and be fully scriptable, that they've opened up their communication protocols for non-commercial usage and are providing a patent covenant (Redhat Responds), and now finally an interesting rumor floating around that Microsoft will be taking on GNU directly. Has Microsoft totally switched gears in how it is approaching the Unix and FOSS sector for direct competition? According to an anonymous email leaked from a Microsoft employee, it seems Microsoft will be developing a framework that will be completely GNU compatible. Microsoft CEO, Steve Ballmer, said on Friday (23 February) that they are aiming to restore a Unix-like environment to its former proprietary glory, at the same time proving that Microsoft is committed to interoperability. Ballmer emphasized that Microsoft's new strategy is to provide users with a complete package, and this includes users who like Unix environments. According to the supposedly leaked email, UNG, which stands for UNG's not GNU, is set to be released late 2009."

MSFT used to be a UNIX vendor

[peter303]

In the late 1970s and early 1980s MicroSoft sold a version of PC-UNIX called Xenix (they didnt write it). Until the mid-1990s PCs were too-weak to effectively run UNIX, so it was not a popular product. In the early 1980s MicroSoft decided to concentrate on MS-DOS and other products, so it sold Xenix to a company which eventually became SCO.

Re:Makes some sense

[Constantine XVI]

IIRC, it's "Those who fail to understand Unix are doomed to reinvent it, poorly" Can't recall the source ATM

MS already has unattended scripted installs

[breagerey]

and has for a very long time.

Re:Wow

[diegocgteleline.es]

I don't agree. Microsoft IS trying to make Windows the best FOSS platform. The goal is not to be nice to FOSS, but to try to damage Linux. It's not me who says it, but Mary Jo Foley [zdnet.com] (who got it from a Microsoft), one of the most journalists experts in microsoft, if not the best. Quote:

"Microsoft is looking at open-source software (OSS) as just another flavor of independent software vendors (ISV) software. Microsoft's goal is to convince OSS vendors to port their software to Windows. But Microsoft doesn't want OSS software to just sit on top of Windows; the company wants this software to be tied into the Windows ecosystem by integrating with Active Directory, Microsoft Office, Expression designer tools, System Center systems-management wares and SQL Server database.

In cases where customers and software vendors want/need Linux to still be part of the picture for some reason, Microsoft will suggest they use Hyper-V, its forthcoming virtualization hypervisor, to run Linux and Linux-dependent applications.

Microsoft's OSS strategy makes a lot of sense for Microsoft. It's another way for Microsoft to try to make Linux obsolete, and not look as obviously ruthless doing so. And for OSS vendors who are selling a lot of their software on Windows -- Ramji repeated a couple of times that more than 50 percent of JBoss' business these days is from software running on Windows -- Microsoft's OSS push isn't a bad deal, either.

They have to by law

[esocid]

This is just a conference to make it look like this interoperability was all their idea. From this quote:

This covenant will use the exact same terms created in October for the protocols covered by the CFI decision. This means that open source developers will be able to use the documentation to develop implementations of these protocols without paying for a patent license. Companies that subsequently engage in commercial distribution of these protocol implementations will be able to obtain a patent license from Microsoft, as will enterprises that obtain these implementations from a distributor that does not have such a patent license. So that's how we're addressing the intellectual property rights

you can see that all they are doing is simply complying with the CFI's [weil.com] decision about how the

Commission found that Microsoft had abused its monopoly in the market for client PC operating systems by (i) refusing to supply its competitors in the market for work group server operating systems with "interoperability information," i.e., documentation allowing server products of Microsoft's competitors to freely interoperate with the Windows environment, and (ii) tying the Windows client PC operating system and Windows Media Player.

Re:MS is a business

[Shados]

Err? Nothing even remotely hint at that.

All it means is that Windows (which btw, already has a partial, optional, Unix-like stack btw!) is going to offer more open source tools, more command-line utilities, more GUI-less fonctions, more open protocols.

Thats it. And thats been in the process for like ever (The latest version of Exchange for example, is fully administrated from Windows Powershell. The GUI works Unix-style, with a front end calling the CLI commands.).

Nothing more, nothing less.

Re:Wow

[Anonymous Coward]

If only Microsoft would release a cute little action figure [microsoft.com] related to this, I would be interested. Until then, I'm not.

Re:MS is a business

[AKAImBatman]

it would mean that Microsoft is pretty much scrapping it's entire codebase for Windows and replacing it with a Unix or Unix-like architecture.

Says who? The NT kernel was designed to be able to project different "personalities", much in the same way that Mac OS X does. The POSIX system necessary has been available in Windows for just shy of forever [wikipedia.org] in an effort to win government contracts and companies that require POSIX as a checkbox on their requisition forms.

Of course, their support hasn't been very good, but that has more to do with an unwillingness on Microsoft's part rather than any real technical reason. Typically Microsoft implements sub-standard support, then claims that their support is top notch. A few examples of this are the David Korn debacle:

http://slashdot.org/article.pl?sid=01/02/06/2030205 [slashdot.org]

Just as bad was the Kerberos debacle where Microsoft extended Kerberos for Windows [schneier.com] such that Unix machines could subscribe to a Windows domain, but a Windows machine could not subscribe to a Unix domain. I called a rep on it in one of their presentations on Win2K, and he assured me that I was mistaken.

Re:MS is a business

[MightyMartian]

Windows NT has had a POSIX layer since the beginning. At any point Microsoft could have extended this and ported over GNU tools if they had wanted. The whole thing smells of bullshit, and Powershell is not bash. It may have its advantages, but you sure can't bring over a library of thousands of shell scripts.

Cygwin is a solution, but of course, that has nothing to do with Microsoft.

Re:this has to be fake

[betterunixthanunix]

Competing with GNU does not mean that they are releasing open source software. What it means is that they would be release tools that are compatible with GNU, so that companies that are running GNU right now would have an easy time switching to Microsoft. It should be interesting to see Microsoft pull this off without violating the GPL.

Also, this idea reeks of embrace/extend/extinguish.

Re:Wow

[Llanfairpwllgwyngyll]

...and I remember when some "microsoft partners" were told (very hush hush) that a GUI-less version of windows2000 would be released.

We're still waiting.

It's not like a Unix system, where a GUI is built on top of a CLI. Windows is GUI by design from the start. It's a whole different kettle of meat.

Re:MS is a business

[TheRaven64]

Unfortunately, POSIX allows a large number of functions to return ENOTIMPLEMENTED, instead of actually working. The original POSIX subsystem did this everywhere it was permitted. It also couldn't be used in conjunction with the Win32 subsystem, so you could only use it for command-line apps. Cygwin filled the gap a bit by providing an implementation of the POSIX APIs that wrapped Win32 calls. More recently, Services For UNIX have provided an updated POSIX subsystem and userland that is more-or-less useable.

Re:Wow

[gnu-user]

Colinux is exactly that: http://www.colinux.org/ [colinux.org]

Its a linux distro that runs on top of the NT kernel

Runs pretty fast, for what it's worth

Re:Wow

[malevolentjelly]

When I can recompile the Windows kernel to my liking, then we'll talk about how Windows will be a better FOSS platform.

Of course, you don't need to because it's not monolithic. The benefits of recompiling the linux kernel stem from the fact that everything in linux-land is jammed into the kernel.

You can simply load different drivers in pseudo-userland and run a separate set of services to completely rework your windows system. As far as enterprises and business customers are concerned, there's little to no benefit for them to be able to compile their own kernel unless it is completely monolithic- it's just a waste of time and a leak of talent for microsoft.

FOSS is mostly GNU userland, not the linux kernel. If you don't believe me, try BSD or OpenSolaris.

Re:Wow

[incripshin]

I think Cygwin's full of too many hacks to be a good starting point. For instance, Windows programs have no ability to fork, and yet cygwin has a fork() implementation. Personally, I don't want GNU compatibility but POSIX compatibility. There are POSIX makefiles and there are GNU makefiles. The difference is that POSIX makefiles run everywhere, while GNU makefiles don't. Just the same, I try never use GNU-specific language features in gcc (I use -std=c89 or -std=c99 with -pedantic). GNU hinders interoperability, themselves. It would be good if a Microsoft-developed make (there is nmake, but I don't know how it works at all) had a POSIX mode and a GNU+POSIX mode, in the same way that GCC allows by use of -std=XXX -pedantic flags to disable GNU extensions.

Also, Microsoft's library model is positively nutty. Static libraries are stored as a big .lib file, while shared libraries are stored as a small .lib file together with a .dll file. Unix has .a and .so files, respectively. Inter-operable makefiles need simpler compilation systems than having three kinds of library files.

Re:MSFT used to be a UNIX vendor

[Anonymous Coward]

Um, you're a little confused on the facts there.

Around the mid eighties, Xenix was the most widely installed unix, due primarily to the cheapness of the hardware on which it ran. To say it wasn't popular just isn't true.

Also, MS never sold Xenix directly to customers, quoth Wikipedia:

"Microsoft did not sell Xenix directly to end users; instead, they licensed it to software OEMs such as Intel, Tandy, Altos and SCO, who then ported it to their own proprietary computer architectures. Microsoft Xenix originally ran on the PDP-11; the first port was for the Zilog Z8001 16-bit processor. Altos shipped a version for their Intel 8086 based computers early in 1982, Tandy Corporation shipped TRS-XENIX for their 68000-based systems in January 1983, and SCO released their port to the IBM PC in September 1983."

Re:Wow

[EvilRyry]

You might not have heard this, but Linux does have support for these things called kernel modules. Also BSD and OpenSolaris don't use the GNU userland for the most part.

Re:MSFT used to be a UNIX vendor

[yorugua]

In the late 1970s and early 1980s MicroSoft sold a version of PC-UNIX called Xenix (they didnt write it).

Great old times..!! I remember I had a 80266 machine back then, 10 MHz (way faster than the original IBM PC-AT, but you could always press CTRL-ALT-minus to set it back to normal speed in case of incompatibilities).

Until the mid-1990s PCs were too-weak to effectively run UNIX

Well, on my 80266 10MHz/640kb RAM I used to do the college work (Turbo Pascal, Turbo C, documentation) on PC-DOS. When I "discovered" Xenix-286, the same machine could run 4 virtual terminals on the console, I was able to edit, compile, run/test on three different terminals. If I made a mistake on C, I'd get a coredump, but the machine kept running. Also, I was able to enable my modem, so a classmate could also work on what I was doing.

Great times, 80266 machine, 640 KB ram, 40 MB Hard drive.

Then I met a lot of people that were using SCO Xenix/UNIX on 80386 class machines, doing all kind of things from running a BBS with 20+modems, or running the billing system of local companies from multiple RS-232 terminals in the late 80's, early 90's.

Re:I think its great news!

[Anonymous Coward]

Apache doesn't have a copylefted license. Its just like you people, I bet you almost posted GNU-Apache didn't you? Why don't you lay off taking credit for other peoples work, thats the whole point, work has value, and you guys just never get that part.

Re:Wow

[mikael]

Microsoft did try this before when they first brought out Windows NT. They provided a very minimal shell environment along with some unix emulation commands (make, ls, df, du, vi) as well as being able to get OpenGL drivers ported over. The idea was to provide these commands to get the applications ported over, and then to silently withdraw the suppport once the applications were ported.

There are still emulation libraries by Cygwin [cygwin.com] and MKS [mkssoftware.com]

Shell scripts are Microsofts weakness. Microsoft held off from including Monad [blogspot.com] into Vista for security fears. This was in a previous Slashdot discussion [slashdot.org]

Re:non-commercial uses

[MrNemesis]

Really, this comment should be +5 insightful by now. By having a caveat that basically says "if you're a corporation making money off Linux (hello Redhat!) and your distro happens to contain a nifty utility that some dude made based on stuff in our patents, we'll sue the hell out of you unless you pay us this protect... er, technology licensing fee", this is just yet another version of embrace, extend, extinguish.

When commercial distros (and community distros used commercially, like Debian) can't implement the tools needed to interoperate with Microsoft on a solid legal basis, distros will fail and Linux will once again be relgated to a) businesses outside of software patent control and b) yo' mommas basement. Like the Samba team say, stay away if you value using FOSS in the commercial world without being held to account by MS. It's a poisoned chalice.

Disclaimer: I'm in the EU and, since most of the patents don't apply over here (yet) we're hopefully in the clear for now. Until MS "lobbies" WIPO to "encourage" the EU to adopt a US software patent policy, or something similar. Yes, I wear my tinfoil hat shiny side out.

Re:Wow

[nxtw]

MS also has existing software available for making Windows UNIX compatible, "UNIX Services for Windows", if memory serves. It's not a long distance from that to GNU compatibility.

With Cygwin already around, and UNIX being open and readily able to be integrated into Windows, it would be a smart way to envelop potential UNIX users. Personally I'd like a Microsoft supported Cygwin, which isn't as buggy and doesn't feel as detached from Windows.

As of Windows 2003 R2 and later, it's now called Subsystem for UNIX-Based Applications.

SFU/SUA applications are not Win32 applications; they operate on the POSIX layer. The apps are still Windows PE formatted binaries. Libraries are also PE and do not have a .dll extension.

The Unix environment is more Unix-like than Cygwin. Executables have no file extension; their names are all lowercase and appear that way in the Task Manager. SUA is aware of NT ACLs and permissions and appears to work with ACLs. It's possible to suspend and kill processes like any unix system.

SUA borrows a lot of stuff from BSD and includes some GNU code. Much of the userland is based on BSD; the SUA FTP application supports HTTP downloads as well, like NetBSD's IIRC. SUA applications can be compiled with the Microsoft Visual C++ compiler or the included gcc (version 3.3). SUA provides a /proc filesystem.

The userland is not as "complete" as a GNU system; commands like top and killall are missing, but ps and kill are functional.

Ports of some GNU software are available here [interix.com].

I haven't found much of a need for SFU/SUA, mainly because I typically have some sort of Linux system accessible and because PowerShell makes it possible to do many of the same things. But it doesn't feel too different than any other Unix

Here's the output of a few commands on my Windows Vista box:

bash-3.00$ uname -a
Interix bobspc 6.0 10.0.6000.0 x86 Intel_x86_Family6_Model15_Stepping10
bash-3.00$ gcc -v
Reading specs from /opt/gcc.3.3/lib/gcc-lib/i586-pc-interix3/3.3/specs
Configured with: : (reconfigured) : (reconfigured) /dev/fs/D/gnu2.intel/egcs.s
ource//configure --verbose --prefix=/opt/gcc.3.3 --disable-shared --with-stabs -
-enable-nls --with-local-prefix=/opt/gcc.3.3 --with-gnu-as --with-gnu-ld --enabl
e-targets=i586-pc-interix3 --enable-threads=posix
Thread model: posix
gcc version 3.3
bash-3.00$ pwd
/dev/fs/C/Users/bob

Re:Wow

[EvanED]

MS has kept it up to date [microsoft.com] too. (I think there was some time when it was neglected, but it was then restored a bit ago.)

They've got several programs (though not that many) including Bash, SSH, and GCC.

The integration with the rest of Windows isn't great.

Re:Wow

[Java Pimp]

Of course, you don't need to because it's not monolithic.

Windows is absolutely monolithic. Even though you can dynamically extend it with drivers/kernel modules, it is still monolithic. As is Linux. When the module is inserted, it essentially becomes part of the operating system.

You are likely thinking of a Microkernel [wikipedia.org] architecture which separates services into completely independent components. However, the difference being if one component goes down, it does not take the entire system with it.

Re:Wow

[Anonymous Coward]

> The architecture of dos/win is derived from CP/M

Yeah, because you see, Windows NT is built on DOS, right?

Would it kill you people to just not speak when you don't know what you're talking about?

Re:Wow

[nuzak]

CoLinux runs User Mode Linux on NT. It's really quite a far cry from "running on top of the NT kernel", and more like a paravirtualized guest.

Re:I think its great news!

[Anonymous Coward]

Apache doesn't have a copylefted license.

Nor does perl.

Re:Wow

[SL Baur]

Proprietary emacsen have died on the vine. The Gosling split (which was really the driving force behind RMS inventing the GPL) - Unipress Emacs never did well. When DEC tried to reinvent Emacs with EDIT/TPU they did a passable job (it was a fantastic editing environment for VMS), but it was so tied to the system that it could never make itself free.

I am certainly not wedded to the command line in and of itself (though zsh is a tool I cannot live without), but I wouldn't be tempted even if they duplicate the Unix open architecture of interchangeable parts. I'm a Linux developer and user because I got extremely pissed off by having my system (the AT&T PC7300 aka the Unix PC) end-of-lifed on me and I never want that to happen again. Never. Those of you who love Microsoft Windows XP, take note. Maybe instead of complaining to Microsoft, you should join up with the ReactOS guys and keep the environment you love so much.

It worked for us ...

NT and forking

[Myria]

Win32 does not have a way to fork a process, but NT does. Passing a NULL image handle to NtCreateProcess() is similar to calling fork(), cloning the memory space as a new process. The NT kernel supports a lot of system calls that are not exposed through Win32, and it's a shame. The NT API is much more elegant and self-consistent than the Win32 wrapper, yet it's the officially undocumented one.

NT is almost a superset of the features of Linux. There are only a few concepts that don't exist in NT, like signals.