Microsoft's Ancient History w/ Unix 403
NutscrapeSucks writes "The Register is running
a article which discusses Microsoft's experience running their own version of UNIX, called Xenix, as their standard desktop operating system. Before they got involved with OS/2 and later NT, Microsoft considered UNIX to be the PC operating system of the future. Talks about Bill Gates running vi, difficulties with AT&T, and other interesting tidbits."
There's a lot of stuff everyone knows, and a lot of stuff you probably didn't
know. Worth a read.
Windows NT == VMS (Score:5, Informative)
Actually, Windows NT was built very much like VMS, the operating system for the VAX built by DEC. David Cutler, one of the main architects for VMS, was hired by Microsoft to build Windows NT. The name Windows NT itself is one of those HAL like play on letters where each letter is the VMS letter plus 1. WNT VMS
a glorified email terminal (Score:4, Informative)
When I worked at Microsoft in the early 90s, the role of Xenix was pretty much relegated to a glorified email terminal. A few old-timer people on the teams I worked with used it, and few of those people did anything but read their email remotely on the Xenix email servers. I don't recall anyone actually running Xenix on any box within their own office.
At no time did I get the impression that a developer at Microsoft felt that Xenix/UNIX was the future of the desktop. It was big, it was bloated, it couldn't run on then-current PCs well, nevermind the smaller machines of the mid-80s.
Sure, maybe there were some hold-outs in groups I didn't interact with, and I was only there long past Xenix heyday, but Xenix had no chance at the desktop, really.
NHS (Score:1, Informative)
Re:Unix is the future. (Score:3, Informative)
Unfortunately, my memory fails as to whether this was still IBM PCDOS or MSDOS. I'm thinking by that time it was MSDOS.
Xenix was for email and little else (Score:5, Informative)
I joined Microsoft in 1988 and after working on QNX I was stunned at how primitive their development environment was. I would have been only too happy to develop on Unix
Instead, all development was done on PCs running DOS.
I don't know who the unnamed "former grunt" quoted in the article is but he's full of it.
Microsoft and UNIX (Score:5, Informative)
Re:Cut N Paste? (Score:1, Informative)
Re:Unix is the future. (Score:2, Informative)
DOS 1.0 was PCDOS unless you ran it on a DEC rainbow or a few other very rare boxes. MSDOS was later.
Of course the 1.0 and 2.0 syscals are still in win 2002 or whatever its called.
ATT's "failure" to properly manage UNIX (Score:5, Informative)
1. retain the telecommunications monopoly but refrain from any money-making ventures outside of the telecom area
2. become a real business, make money on anything you want, and open up competition in telecommunications.
ATT chose choice #1 -- retain the monopoly. This was for them a sure thing. They had always managed to retain the monopoly in the past and it provided a steady source of income. Computers were new, and internally were not percieved as a consumer item.
So at the time Bill was talking about ATT, the UNIX development/administration/lisencing was, by legal necesity, not a money-making area for ATT. UNIX was a tool to develop telecom products, the real business of ATT. Giving the technology away and managing the process "for the public good" was a means to demonstrate that it was not a money-making venture as well as a way to trumpet Bell Labs. It didn't recieve the best support from management, though, as they were focused on the money-making areas of the business.
On the other hand, the statement that ATT didn't know what they had, was that ever true! Once they did figure it out it was too late, they were legally barred from that market untl after deregulation (nothing is forever!) -- too late!
Re:Microsoft and UNIX (Score:5, Informative)
With NT, Cutler finally designed and helped implement an operating system explicitly meant to run on multiple hardware platforms, about 20 years after the second implementation of Unix in C made its debut.
Cutler's spent most of his life trying to snuff Unix, poor boy. The booming popularity of Linux in the server world running on those PCs must be incredibly frustrating to the NT hackers at MS who thought they were going to finally drive the last nails into the Unix coffin
Re:Unix is the future. (Score:2, Informative)
Comment removed (Score:2, Informative)
Re:Xenix? That's SCO's distro, isn't it? (Score:4, Informative)
>
>You might want to lay off that crack you
>mention...
Actually, he's not on any more crack than you.
Microsoft didn't sell Xenix directly - they licenced it to various companies, including SCO. They are the ones who actually sold Xenix for Intel processors, starting with SCO Xenix System V way back in 1983.
This is the eventually became the basis for SCO UNIX in 1989. When they released version 3.2v4.2 of this product, they also started selling it under the named Openserver and Opendesktop around 1991. Around 1995 they dropped the SCO Unix and SCO Opendesktop brandings, and sold exclusively under the name Openserver (in Host, Enterprise and Desktop configurations).
Unixware, far from being SCO's baby, was originally put out directly by AT&T's Unix System Labs back in 1992. They divested themselves rather quickly of this product by selling it to Novell in 1993, who themselves sold it to SCO in 1995.
Finally, in 2001, Caldera aquired the entire Unix division of SCO, to become the controller of the Xenix grandchild Openserver and the Unixware product lines.
Mind you, over the years SCO has stripped out virtually all of the Microsoft code in their product line, but had to go into litigation to remove some DOS compatibility code they by contract included in their software.
In any case, back to the point - SCO had 16 years of experience with Microsoft Xenix products before it even touched Unixware, and Unixware had 4 years of existence before SCO got their hands on it.
So both you and the person you responded to didn't have it quite right.
Matt
Re:Windows NT == VMS (Score:3, Informative)
NT actually started life as OS/2 3.0, not Windows.
Re:a glorified email terminal (Score:4, Informative)
Xenix ran fine on a 386DX-35 platform supporting 10 users off of that ONE computer using Wyse 75 terminals. It supported several businesses helping with Multiple tasks in that company using that ONE computer. Excalibur was the best Business accounting/inventory/Point of Sale software on the market at that time (1992) It ran faster than anything that microsoft offered it gave you more productivity than anything that Microsoft offere'd then and NOW from your equipment and coince it was really written by a group that were outside Microsoft at the beginning, bought by them and then re-sold (SantaCruz Operation) it was never tainted with the Microsoft Style. The Only thing that sucked about Xenix was that the Xwindows system was horrible and required specalized hardware, Compiling X11 on it solved that problem.
SCO Xenix was a awesome thing at the time, and I still have the origional disks and Manuals from that 386 version.
M$ used Xenix until 96-97 (Score:2, Informative)
If I recall correctly, the last Xenix server on the MS corporate backbone was removed in late 96- early 1997. Primarily, they were used as Internet gateways, running Sendmail. Also , they functioned as internal gateways between MSMail and Exchange while the company converted everyone over to having personal mailboxes on an Exchange server.
While we tried to get some improvements made to applications running on the Xenix boxes, rumour had it that no one could develop these apps, since the source code had been lost somewhere on campus. Also, this is why they couldn't sell the OS to another company.....c'est la vie
Re:Windows NT == VMS (Score:5, Informative)
IBM/HAL, Santa/Saten, its all part of a biiig plot...
what developers at Microsoft ran (Score:5, Informative)
There was a push to self-host on NT which I recall became feasible in early 1992. Eventually more and more of the group switched over to it. I think the rest of the company probably didn't switch until after the first version of NT shipped in July 1993.
But it is indeed true that the standard email terminal in 1990 was connected to a Xenix machine, and there was a card handed out "how to use vi to edit email" or somesuch.
Actually I used vi for writing code for a while...someone inside Microsoft had hacked up a version that supported multiple windows...but eventually I switched to slick like most of the NT team. BAsically the incremental search in slick was cooler than the . command in vi.
- adam
Some history notes on NT's development: (Score:5, Informative)
In there, you'll learn 'NT' was related to the first proc it was targeted to, the 860 of intel, codenamed 'N10', plus some juicy stuff about the development of NT3.1 and win2k, and some related notes to Unix and NT.
Re:Windows NT == VMS (Score:2, Informative)
all in all, not a bad idea: on paper
enjoy!
Re:Windows NT != VMS (Score:2, Informative)
Hint - it will be the most stable thing you've seen on PC (at that time)..
A lot of flannel there (Score:2, Informative)
SCO's move from Xenix to Unix coincided with their less developer-friendly, more grab-the-cash mentality (adding RAM to your box? That's an additional license fee, please.) as Doug Micheals took over from his dad (Larry), and played a large role in SCO's decline and eventual purchase by Caldera.
I'll always have fond memories of my years with Xenix, though. Even though my video card has more RAM than any of my Xenix boxes ever had -- hell my Palm IIIxe as as much.
Re:Unix is the future. (Score:2, Informative)
Re:Windows NT == VMS (Score:5, Informative)
What killed VMS was not DEC, but Unix - mostly Sun. Their stuff was 10x as fast at 1/10 the price, so people bought Sun instead. DEC was never really able to adapt from the closed proprietary business model to the open commodity business model. Even with Alpha, DEC never got more than 1% of the Unix market.
Re:Windows NT == VMS (Score:2, Informative)
Re:Unix is the future. (Score:3, Informative)
But none of the 'innovations' you cite came out of UNIX. The closest one would be Kerberos, but even that was conceived from day one as being independent of the O/S. MIT has developed enough O/S to know that there is more than one.
UNIX was not an O/S with lots of innovative features, the main innovation was the idea that most of the O/S could be written in a high level language. Most of the advances in UNIX consisted of removing unnecessary junk from Multics or ITS.
UNIX was not the first O/S with symbolic links, it was however the first where the feature was widely used. There is even a way to create symlinks in VMS, although you have to go through an API to do it.
Re:Brighten up everyone!!! (Score:3, Informative)
Apple also had Unix on the desktop in years past (Score:3, Informative)
A/UX had a nice GUI (it was from Apple after all!) which was very similar to the Macintosh GUI of the time (System 7). It had all the greatness of UNIX, including pre-emptive multitasking and protected memory, and it was even able to run most Macintosh applications without modification. Yes, you could bring up a terminal window and much around with a command line if you so pleased, but like today's Mac OS X, you never needed to. Sadly Apple only marketted it to corporate and higher-education users, so it never caught on and was forgotten.
Re:Two interesting points (Score:2, Informative)
NT on Alpha, MIPS, x86, PPC, & Itanium (Score:3, Informative)
PPC made it to NT 4.0, Service Pack 2. The last to go was the Alpha in 9/99 which was suprising to many as it was in line to be the first 64bit Win NT implemention. There are still sites like AlphaNT [alphant.com] out there providing resources for the last holdouts.
Re:Windows NT == VMS (Score:2, Informative)
Th VMS file system was awesome, it worked transparently across a cluster in the early eighties and it supported ISAM out of the box. If you paid extra, you even got transaction journalling, on straight files. Anyway, Andy G stayed with Digital so you only got part of the time working on NT.
Another point is that Cutler is a legendary coder, he added paged support to 11M over a weekend, but I guess that he wasn't much of a team worker.
I agree about the market/sales droids - they were considered legendary in their awfulness even within the company.
Re:ms agreed to never again produce a version of u (Score:2, Informative)
This is true. M$ sold Xenix to SCO SCO Unix is Xenix or was Xenix. But part of the condition of the sale was the M$ could NEVER produce a version of Unix again. That is why NT was touted as an OS like Unix but better. .
Re:Brighten up everyone!!! (Score:3, Informative)
He speaks before he knows... People like this give software engineers bad names. It is so obvious he hasn't looked at the Unix API... The wait() call is a central part of Unixes, since day one. Signals, semaphores, mutexes, they are used abundantly in Unixes. Whoever posted this should be tarred and feathered..
I program both Windows and Unix, and have written OS Wrappers which allow me to port my applications between OS's. Everything you can do in Windows OS's can be done in Unix. Threads, Processes, Semaphores, Mutexes, Spin-Locks, Signals, memory maps, pipes, timers, etc. To make assertions that Windows uses WaitForObject, etc. is a ridiculous one. I would use semaphores or mutexes to co-ordinate two threads.
Personally I find the Unix OS much more straight forward and easier to design for. Microsoft keeps on making programming more and more esoteric, more difficult to understand. I use COM all over the place, and have started to port COM to Linux. It is nice, but it is not anything new, it is basically dynamic libraries with a known exported interface which exports class factories. I write low-level, often device drivers, or interfaces to video capture devices for DVD burning software. I use DirectShow which is a layer on COM. I find COM beneficial for some things and think Linux needs a similar framework.
greed kills new platforms (Score:2, Informative)
Most likely, what killed Xenix was AT&T's licensing fees--it's hard to see how such a premium priced system could have caught on with normal users, in particular at a time where people expected their OS to just "come" with the computer for free and nobody else was offering something as nice and expensive as UNIX. Many other great software systems have been killed by the desire of their creators to milk the market early and often with "breakthrough platforms". Smalltalk80, CommonLisp, and NeXTStep all priced themselves out of the market. Gates has the right business idea: make it cheap and simple and tighten the screws once you have a monopoly and people can't jump ship anymore.
I find this particular report rather dubious, however:
"Anyway, when I worked there back in the mid 80's every poor sod in the company from Bill down to the mail clerks had a Xenix terminal on their desk and used it daily for email at least. Meaning every poor sod had to master vi before they could request vacation time (and everyone wonders why Microsoft is such a hateful cramped little place.)
As opposed to what? The text editor that came with DOS? It seems to me they should have been so lucky as to get "vi". If Microsoft didn't like it, they could have developed whatever you think they did was so much better for DOS. And where did "vi" even come from? Xenix derives from 7th Edition UNIX, and I don't remember "vi" coming standard with 7th edition UNIX. If Xenix had "vi", someone must have decided that it was a good idea to backport it.
Re:Windows NT == VMS (Score:3, Informative)
The price performance was never quite that extreme, SPARC was about double to tripple the price performance of the equivalent VAX workstation when it first appeared.
The thing that killed VMS was not UNIX, it was RISC. People moved to Sun in spite of UNIX, and for that matter in spite of Sun's quality control. In those days, interms of reliability SunOS was to VMS what Windows 3.1 was to AIX.
Incidentally, DEC was a very early member of the UNIX club. The first virtual memory UNIX was developed on a VAX. It is a pity that Thomson et. al. were so determined to learn as little as possible from the design of VMS.
In the very early years Sun attempted to license VMS. DEC refused, claiming that it could not be ported because of the dependency of VMS on a couple of fairly specialized processor instructions, like remove from head interlocked and the security ring instructions.
The reason DEC was so far behind Sun in the first place was that their bean counters axed the PRISM project that was meant to built the successor to VAX and VMS. Dave Cutler left DEC for MSFT and vowed to force DEC to buy the O/S they could have had for free - whats more he did exactly that. When the Alpha chip appeared much later than SPARC it was named AXP or Almost Exactly Prism as insiders call it.
WNT is not VMS but it has most of the best features of VMS and is the type of thing you might build if you were designing a sucessor to VMS but did not need to have backwards compatibility.
There are a bunch of late VMS features that WNT is noticably lacking, in particular the transactional file system. Hopefully we may see some of that appear in OFS. What is disappointing about WNT is that many of the interesting O/S features are sumberged in low level APIs. It is possible to do VMS tricks like ASTs but you have to really know the layout of the O/S.
Unfortunately there is no guide that compares with the Digital VMS architecture manual.
Re:Cut N Paste? (Score:4, Informative)
That's just off the top of my head. Things beside these I can usually find in New Riders' book Vi IMproved -- Vim
Good luck. I use VIM almost exclusively for my editing needs; over the last ten years it has been my constant companion through thick and thin. I wouldn't work without it.
Re:ATT's "failure" to properly manage UNIX (Score:1, Informative)
+ They introduced their own line of workstations running UNIX. This would be like Microsoft coming out with a PC line today.
+ They stedily raised the price of a UNIX licence from near-0 to high enough that UNIX was effectivly excluded from the low-end (PC) markets.
+ They forbid their licencees from referring to their products as "UNIX", thus causing much branding confusion and ye olde *nix jokes.
+ They got into bed with Sun and started the "Unix Wars" of the early 90s.
True none of that directly affected MSoft, except allowing them to sell into all sorts of markets abandoned or scared off by the UNIX masters.