OS Wars
Brad Wardell
Bwardell@stardock.com
 
Our story so farÖ
It is a time of operating systems civil war as the empire of Microsoft looked to squash out the rebel operating systems still thriving on the edges of the computer marketís rimÖ
In 1992, Microsoft faced its most serious challenge yet Ė IBM OS/2. Up to this point, Microsoft had to only contend with DR DOS which failed to gain any sort of wide spread acceptance and, in the end, was only a DOS clone that had to be compatible with Microsoftís standards. The other primary threat came from Apple but MacOS was trapped in a totally different hardware dimension. Apple had been content to sit on its laurels and its market share was never going to be a serious threat.
But OS/2 was different because it was not only compatible with DOS and Windows, and arguably ran DOS and Windows better than Microsoftís offerings, it also advocated its own native application development. If successful, it meant that Windows would be at a severe disadvantage.
Luckily, a combination of poor marketing and management by IBM and ruthless and somewhat illegal competitive practices by Microsoft ensured that Windows would win the great 32bit battle. Its key to victory was ensuring that Windows came preinstalled with every single PC on the market. While OS/2 was a superior product to Windows 3.1 and Windows 95, Windows achieved the all important "Good enough" rating that discouraged people from actively looking for an alternative if their machine already came with Windows. If the OS you got free with your computer is "good enough", itís a tough job to get them to go out and pay $100 for something else.
That brings us to 1998Ö

To paraphrase Star Wars, "The tighter you make your grip there will be more operating systems that slip through your fingersÖ" While Windows 95 rules supreme, Windows NT has won the high end OS market, and Windows 98 is poised to succeed Windows 95 in a matter of months, there are numerous other operating systems now available that show some promiseÖ

BeOS

Figure 1 BeOS

 

BeOS got its start on the BeBox which from my perspective was an extremely promising route to success. Finally, a machine designed and built from the ground up, that would be affordable to bring the power and performance that many users have been looking for. A Super-PC that had multiple processors as standard equipment, it could have been like a modern day Amiga without the horrible marketing and limitations.

But Be ran into some problems. It looked for awhile that Apple might buy Be and integrate it into the Apple OS and many people, including myself, felt that this would be a great idea since System 8 is still behind the times technologically. But Apple pulled the plug on third party licensing of the Apple hardware which really took a lot of steam out of Beís Macintosh plans. Be therefore made the decision to come over to Intel which must have been quite a set back.

Most people take it for granted today but the PC, underneath the nice looking cases, is a complete mess from an operating system vendorís point of view. Hard core PC users get ticked off that their latest 3D card has crappy Windows drivers. Thatís only the tip of the iceberg. Try starting from scratch from a device driver standpoint. Imagine having to support all those wacko video cards, hard drive types, modems, sound cards, BIOSís, etc. What a complete nightmare it must be to try to have to support this. Itís taken Linux and OS/2 years to provide adequate support and they have had extremely strong device driver support from developers (some may disagree but trust me on this, it takes a huge amount of work to even remotely keep up).

This trend to take hardware support for granted works in BeOSís favor at first because people forget about the bad old days. But a few days with BeOS will quickly bring reminders of those days. Given the Intel version isnít even out except in developer release form, itís hardly fair to complain about this. Indeed, Iím not complaining, itís just a simple fact that if you take the BeOS route, be prepared to custom make your PC to fit it just as people did for NeXTStep for Intel before and OS/2 before that.

But the look of BeOS is quite striking and definitely appeals to a number of people. On the other hand, looks alone should not make you switch OSís. [Evil Plug Alert] Stardockís own Object Desktop for Windows (in beta) can radically change the look and feel of Windows. Hereís a screenshot of our BeOS theme (beta):

Figure 2 Windows with Object Desktop running BeOS theme

 

 

BeOS has a definite uphill battle to take though and I am not sure itís going to make it. I really hope it does but having learned the hardware from my OS/2 experiences, software is what makes OSís successful. And OS/2 had the added advantage of being demonstratably superior in a vast number of areas over Windows (user interface, multitasking, mulithreading, crash protection, 32bit, cheaper). It had all those things going for it and yet did not become mainstream. Today, all operating systems have decent user interfaces, multitasking, crash protection (or some at least), are 32bit, etc. So what will BeOS do to carve out its niche?

 

Linux

I have a dream. This dream involves being able to simply purchase a couple really really powerful PCís and then let me hook them up to the network and allow fairly wimpy PCís connect to them and run applications off of them where the application executes on the powerful PC but displays itself on the wimpy PC.

Iím in an office that is full of Pentium 75ís. Many of these PCís are just used for doing general office work (data entry, word processing, email, technical support, etc.). But now, weíre looking at having to upgrade these PCís as newer and newer versions of our applications seem to require ever more hardware to run decently. Who would have ever thought that an email package would bog down a Pentium? Stick with old software and get out of touch, go with new software and upgrade your machines.

But for me, Linux offers an alternative via X-Windows. Yes, X-Windows has been around for years and people still donít understand the promise it holds. Much of the JAVA fanaticism you see out there is misguided because these people basically just want to do what X-Windows has been doing for years. You know the typical conversation:

"You see, we have an internal Intranet and a couple servers running JAVA applications, then our thin clients, old PCís can connect to those servers and run the JAVA programs." JAVA has some great strengths which Iíll get into in a different article but what I think many corporations want to do is to do the same thing I want to do.

They want to have a couple really powerful PCís that have the bread and butter applications running on them and a bunch of "thin" clients (read: older PCís that weíre too cheap to upgrade) that can connect to them and run that software remotely. Iím writing this article on a Pentium 120, which was a pretty powerful machine once. But running Corel Photopaint on it today is extremely painful. Wouldnít it be great if I just had a P2-333 with 256 MB of RAM in the main lab that I could just hook up to and run Corel Photopaint on it but have it display on my machine? Iíd get the speed (or pretty darn close) of the fast machine while being able to continue using my Pentium 120.

Recently, a company in Great Britain released a program called VNC which allows people to remote control other PCís. This is nothing new of course but the kicker is that itís free! Itís a fine program but not quite the same as X-windows which lets you seamlessly run programs that are on other PCís. Anyone whoís used that program probably has the same hunger I do.

The problem with Linux is the same problem BeOS, OS/2, and other non-dominate operating systems have Ė applications. To the ultra-techie, the programs on Linux are fine, but for the majority of users, thereís no alternative to Page Maker or Corel Photopaint or Corel Draw (the current verison), or Office 97. Like it or not, people like these slick, bloated, feature-overkill applications and weíre not going to do with out them. Weíll never use 90% of the features but dog-gone it, we want them all there and we want the cute icons.

Linux is also a pain to install for casual users and adding hardware can be a pain. Itís getting a lot better but it still has a ways to go. Its user interface is also lacking. Itís not bad but for many people, user interface is the key. Anyone using OS/2 today would probably name its user interface as the #1 reason why theyíre sticking with it.

Hardware support is also problematic. Itís not bad, mind you but itís not 100%. When you buy something, odds are that the best case scenario driver wise is to scour the Internet for it. Smart Linux users will buy hardware with Linux in mind. But, Microsoft knows how to take advantage of this. You get your PC with Win95 or NT on it and that makes it a pretty tough case to convince someone to "upgrade" to Linux. Linux offers superior stability and vastly better performance, especially in multitasking. But to the average user, thatís not enough and itís a problem that will get worse over time. You buy NT, it comes with a web server. We got in trouble with some of our users because our website runs on NT. Why? Because we got a free copy of NT server and it was a 2 minute job to set up the server and that meant that we could allow someone non-technical to set it up leaving our precious developer/technical resources free to do other things. (The main reason though was because we had a PowerPC sitting around and it was either AIX, which weíd have to buy and learn or NT which we already had).

But I think Linuxís greatest strength is its ability to bring X-Windows to the masses. If there was a central marketing force for Linux, they could probably convince a lot of schools, a lot of companies, and a lot of government agencies to switch to Linux purely for the reasons Iíve mentioned. And this is happening to a degree already, just not as quickly or as wide spread as it could be otherwise.

Figure 3 Linux with WindowMaker

 

Rhapsody/MacOS

If youíre not a Macintosh devotee, the Rhapsody/MacOS issue is pretty confusing. It essentially boils down to this: Steve Jobs talked Apple into buying NeXT instead of Be. So now Apple has this wonderful, if slightly long in the tooth OS called NeXT that has Unix underpinnings on top of the Mach micro-kernel. It also has a well established if not even more long in the tooth OS called MacOS.

From a casual user perspective, MacOS is the best OS there is (as long as it comes with the Macintosh hardware). With that said, MacOS is still far behind technologically. Iím no fan of Windows 95 but calling Win95 = MacOS 89 is not only a bit conceited but just plain wrong. MacOS 6 (from circa 1989) did not have memory protection, no real multitasking, no mulithreading, some pretty serious "init" issues, and was kind of ugly by todayís standards. MacOS 8 improves much on this but not nearly enough. Itís amazing how Apple just plain dropped the ball and allowed Microsoft to catch up on this.

For me, the strength of the MacOS has always been the vision of turning the PC into an appliance. People donít put a lot of thought into how their microwave oven works, they just know it works. Macintosh really was the computer for "the rest of us" (well, not me per se but for the general consumer). It is the computer that you buy, plug in, and it works like a simple appliance.

As time has gone on, Apple has moved further and further away from that. MacOS 8 is not nearly as simplistic as System 7 was and this is both good and bad. An Rhapsody isÖwell, good grief itís powerful but a far cry from the simplicity of the traditional MacOS.

Right now, it seems Appleís plan is to continue along with MacOS and slowly roll out Rhapsody as it matures. Itís hard to say how compatible with traditional Macintosh applications will be or how seamless that compatibility will be.

But Apple does have the third party support. For graphics designers, casual users, educators, and soon power users, the Macintosh/Rhapsody combo gives a pretty compelling story.

But over the long term, what will Apple be able to bring to the table that you canít already do with Windows? Gag I hate saying that but the Microsoft "rip off ideas, implement them half-assed" approach to things seems to be working pretty well for them.

And itís hard to really advocate MacOS or even Rhapsody from an objective perspective if youíre not already a Mac user. Linux holds out some unique promises, Be is at least different, but Rhapsody/MacOS are simply alternatives that are pretty comparable to Windows98 and NT. This is good but the question will then just come down to market share, can Apple, armed with an Intel version of their OS, get their OS preloaded on machines? This will be the key.

OS/2

Speaking of comparable alternatives, OS/2 is what happens if you fail to get pre-loads. OS/2ís installed base of end users grew rapidly during the time frame when OS/2 was clearly superior. In 1992, 1993, 1994, OS/2 was unquestionably a better OS. But when Windows 95 came out and when Windows NT improved to 4.0, the debate turned into an argument over technicalities. All the flame wars in the world over short-cuts versus shadows is not going to convince many people to give up all that Win32 software and jump over to OS/2.

Today, OS/2 is a viable alternative to Windows95/NT but it gets tougher each year to give a good reason to switch over to it when Windows 95 or NT came with your computer. It doesnít have the software support like MacOS has, it doesnít have X-Winodws built into it like Linux, and itís not "hot and new" like Be. So OS/2 continues to be used by people who were already using OS/2.

OS/2 does continue to thrive in corporate environments, largely because it can be run without a GUI and remain a 32bit, multitasking, multithreaded, protected operating system. IBM is trying to position OS/2 as the ultimate JAVA OS though no one is sure to what end. IBMís strategy for network computing is relatively sound from a certain perspective but end users trying to stretch that strategy to fit them is not. There is no compelling reason to write a JAVA game. There is no compelling reason to write a large scale JAVA application. Sure, I could get a JAVA mail reader or news reader but to what end? How many people canít find a better native alternative? Am I going to buy a JAVA office suite when my computer probably came with Microsoft Office? Even if it didnít come bundled with Office 97, would I go with a slower, feature-poorer office suite? With this in mind, what is IBM hoping to do with OS/2? Their plan is largely to create a plethora of JAVA applets that allows information to be handed around easier. Intranets running JAVA applications that allow clients to do their data entry easier. Or JAVA applets that allow remote users to easily access needed data. This is all well and good but why OS/2 then? Sure, the benchmarks show it runs JAVA faster but in reality does anyone notice?

Ironically, IBM has largely given up on the parts of OS/2 that really could have made OS/2 grow in more market segments Ė object oriented technology. They could have continued forward with DSOM and integrated OS/2 truly into the Internet. Not in some half baked way like Microsoftís "active desktop" but made the workplace shell truly internet/intranet aware. They could have also made it so that OS/2 could remotely execute programs ala Linux. Workspace on Demand has some of this but itís not enough and IBM continues to market it to only certain segments. The problem with OS/2 at IBM is that IBM isnít sure what exactly to do with it.

If youíre already an OS/2 user, thereís not a lot of reason to move away from it at this point. OS/2 continues to get pretty good hardware support, application support is still there from Lotus which just released a new version of Smartsuite for it. Where OS/2 does lack is in grass root development support. One developer I talked to said "OS/2 users are takers. They cry gimme gimme gimme but never do anything to give back." Upon reflection, it is true of certain parts of the community. Many OS/2 users only try out freeware software but they themselves arenít willing to give a stab at trying to contribute anything back to the "community" (i.e. lots of freeware users but few freeware developers). This is in stark contrast to the Linux and BeOS communities that are jumping over themselves to write software for free for the OS. No where is the problem more visible in the VNC porting. As mentioned earlier, VNC is a free remote control package that runs on several OSís. The source code is freely available and already thereís versions for Linux, Mac (I think), Windows 95/NT and even a BeOS port on the way. Whereís the OS/2 version?

Thatís not to say that thereís nothing going on in terms of grass roots development. The Win32OS2 project is succeeding beyond anyoneís expectations. It is a program in which you simply run its converter on a Win32 executable or DLL and it converts it to an OS/2 EXE or DLL without even a significant performance loss. Truly amazing. When our company came out with A Map Creation utility for our game Entrepreneur, we only had a Win32 version. But an OS/2 user managed to use the Win32 converter on it and turned it into an OS/2 converter. Thatís amazing. Naturally thereís been no mention in the press since the press, which now lamely complains about Microsoftís dominance in the OS market, still shuns any discussion of OS/2.

Figure 4 OS/2 running Object Desktop Professional

Despite OS/2ís problems, I canít help but wonder exactly why someone would use say BeOS over OS/2? It has to be a perception thing. In fact, I asked a group of BeOS users this very question and they said that the difference is that they feel like IBM doesnít care about OS/2 and that itís stagnating while BeOS is up and coming.

IBM is certainly not the master of perception but in reality, OS/2 continues to get updates on a regular basis. There have been 6 significant updates to OS/2 just in the past year. While many BeOS users may not agree, I just canít see why they donít just use OS/2. Itís already a powerful multitasking mulithreaded OS with a great user interface, pretty decent software application support, and good device driver support. I think in the end, OS/2ís biggest problem is perception and that is what is holding it down.

Windows 95

One canít mention perception affecting reality without talking about Windows 95. Microsoft is the king of making a good first impression. On the surface, Windows 95 has plug and play, its short cuts allow for an almost object oriented working of the system, it has a registry for keeping good track of your programs, and it is just plain slick.

I donít want to make it sound like Windows 95 is junk -- itís not. As an engineer, I just hate seeing inferior technology win. Windows 95 isnít bad but itís a far cry from being the best choice out there strictly from a quality stand point. As you run Windows, you tend to discovery that the registry gets into trouble and the system becomes less and less stable. Programs can hang the system causing you to have to reboot and this happens quite occasionally.

But overall, Windows 95 works and does the job adequately and the key, preloads, has ensured that it and Windows 98 will be the dominate OS. Indeed, the OS Wars could be considered over and all that are left are bandits in the hill sniping at Windows 95 and NT.

 

Windows NT

Windows 95 may have dealt the deciding blows to the competitive OSís but itís Windows NT that benefits. After its disappointing start with Windows NT 3.1, it continued to steadily improve. NT 3.5 was quite decent if you could ignore the user interface and NT 4.0 has become the mainstream choice of power users. People who dislike Microsoft will often slam NT for problems that do exist but are not really problems that affect most people. NTFS has problems with fragmenting, for instance. But how much does this affect the average user? Hardly at all. NTFS also has the ability to seamlessly compress directories which is pretty nice. Microsoft, always looking to eliminate competition has also ensured that it comes bundled with pretty much everything. The microwave oven and stereo that NT 4.0 came with, while now in my kitchen and living room respectively, are arguably meant to be part of the OS and those who disagree are just sorry losers who obviously just envy Microsoft.

But NT disappoints me and hereís why: Microsoft has succeeded in ripping off good ideas but they continually implement them half-assed and now, most professional PC users are stuck with a Microsoft solution and cool features like what I mentioned in Linux or in OS/2 are just not in NT. Microsoft just implements that easy to see features. They clone much of OS/2ís desktop for their GUI (sorry Mac users, it was OS/2 that Win95ís UI was largely based on, not MacOS Ė of course, OS/2ís desktop was derived much from the Mac but there are a lot of key differences). But Microsoft only implements it half-way and so much of the elegance of OS/2ís workplace shell is lost in NT.

Just as NT tries to claim to be "multiiser". Without an X-Windows type system integrated into NT, NT is not multiuser in the traditional sense.

NTís multitasking also leaves much to be desired. Processor intensive apps, particularly Win16 ones, can really bog down the systemís responsiveness unless you manually change the processís priority (and how many people know how to do that?). On OS/2, I used to do a lot of rendering in the background and never even noticed the impact of it. On NT, the system gets quite sluggish and it still takes it just as long for it to finish the rendering.

But NT is very solid and rarely crashes in its normal use. Given that Windows 98 is going to be the last of the "regular" Windows line, for most people, Windows NT is the future of the PC OS.

Figure 5 Windows NT with Object Desktop - Rhapsody/OS/2 theme

 

Conclusions

I read a recent article in one of the trade magazines where the author felt that now was the time to create an alternative OS to whatís out there. That Windows 95 and NT had too many flaws and was ripe for some good competition. The new OS would just need to avoid the "mistakes" IBM, Be, and Apple have made such as lack of apps, bad driver support, and an unclear marketing message.

I couldnít help but think "Well DUH!" Like those companies werenít aware of those problems. As if some new company could just wave a magic wand and not make these "mistakes". Kind of ironic for a columnist in a weekly trade magazine to complain about the lack of choice after they spent years deriding non-Windows OSís for pretty suspect reasons. The lack of for sight by many in the media has lead to an industry where the magazines are now getting thinner and thinner (how thick are the weekly trade magazines these days for instance?). Those who have spent much of their time advocating AmigaDOS or Macs or OS/2 or Linux or BeOS can at least see the irony that those same magazines that spent just as much effort trying to ensure that it is a Windows world sewed their own destruction (or at least their own downsizing).

So the war has pretty much ended, the remnants of the rebel forces have taken to the hills and prepare for the future when the dominate OS is ripe for attack again. And while Windows 95 and NT are far from being the ideal OSís weíd like them to be, theyíre not too horrible and the alternative OSís are so much better now than they used to be that you really donít lose out by trying out Linux or OS/2 or BeOS or MacOS (if you can afford it).

 

Resources:

VNC, a free remote control software product.

http://www.orl.co.uk/vnc

Object Desktop, a desktop environment that extends the features of an OS.

http://www.stardock.com

WindowMaker, an enhanced shell used by Linux users.

http://windowmaker.org/

BeOS, a new alternative operating system with some unique design concepts.

http://www.be.com

Win32-OS/2, a program that converts Win32 programs to OS/2 programs.

http://www.os2ss.com/win32-os2/