The Technofile Web site has moved.

Technofile is now located at
Please update your links, bookmarks and Favorites.  

Why Linux matters: An admiring essay
technofile  by al fasoldt

Columns and commentaries in a life-long dance with technology
Simple gray rule

Why Linux matters: An admiring essay

July 18, 1999

Note: This is a very long article, so be forewarned. Also, I've found that many people who read this have no idea that I'm an expert on Windows and or that I know what I'm talking about when I criticize Windows or when I talk about the way PCs work. That's because they're reading this via an external link that bypasses my site. If you're one of those readers, check out the Technofile site using the link on this page. Please.
By Al Fasoldt

Copyright © 1999, Al Fasoldt

You'll see familiar surroundings if you switch from Windows to Linux. If you install Caldera OpenLinux 2.2 or the latest Red Hat Linux, you'll find most of the familiar things Windows is known for -- windows you drag around, a right mouse click that pops up a menu, even a "Start Menu" for launching programs.

But all versions of Linux are drastically different from Microsoft Windows in hundreds of ways. Even the stuff that seems similar is different, sometimes in small ways and sometimes in very big ways. That's because Linux was not created using any of the old-fashioned computer code that Windows is full of. Linux was a new project, engineered to do things the best way possible. And in some cases -- in a lot of cases, probably -- the best way possible is not the way Windows does things.

So switching from Windows to Linux can be an eye-opener. My first impressions were full of exclamations -- "Ooh!" and "Wow!" were just about all I said for the first few hours -- and my experiences in the weeks since the day I installed Linux have done nothing to change those initial impressions.

Linux is outstanding in dozens of ways that Windows is not. Linux is powerful and stable and forgiving. Linux can't possibly run low on "resources" -- the single biggest Dumb Thing in Windows -- and it doesn't use DOS in any way at all. It is perfect for families because everyone who uses a Linux PC can have separate passwords and totally separate mail and Web-browser settings. (And each user will have a unique desktop and a separate, private storage area, too.) Windows 95 and Windows 98 can't do that, even if you go to all the trouble of setting up Profiles for each user. (Basically, Windows 95 and 98 are just not multi-user operating systems and never will be. Since they are based on DOS, which in some ways isn't even an operating system, they can never be able to handle more than one user without the danger of messing things up badly.)

And there is much more, of course. Let's take a look at the similarities and the differences, one by one. First, I need to explain where Linux came from. This will take a while, so get a cup of coffee or some hot tea and settle down.

Linus Torvalds created Linux in 1991 as a freely available clone of Unix, the operating system used on many of the computers that formed the Internet. Programmers liked Unix because it could do just about anything. (You might think of the early versions of Unix as workshops for computer programmers. These operating systems had all the programming tools built in.)

But Unix (which is supposed to be spelled as "UNIX" if you really care, but I don't) was under the control of the people who owned it, and that meant it cost a lot of money to get a copy of Unix. But far more troubling than that was the difficulty of fixing problems in Unix and getting improvements into Unix (two different things, if you think about it).

Torvalds wanted an operating system that anybody could use. This meant it had to be free or as close to free as possible. And he wanted to make sure anyone could fix and improve the operating system.

Go back and read that last sentence again. It tells you more about Linux -- and more about the big pile of doodoo that Microsoft has found itself stuck in -- than anything you're likely to read for the rest of this year. It expresses the central issue in the battle between Microsoft and the Open Source community. Microsoft, understandably, wants to sell its software and wants to control it. The Open Source people want to give their software away and don't want ANYBODY to control it.

Control, in this way of looking at things, doesn't mean "direction" or "guidance." The Open Source folks want a lot of guidance and need a lot of direction. (So does my cat, but that's another story.) Control in this sense means preventing something, sort of like being a busybody. It means telling people they can't drive faster than 65 mph or making people stand in line for hours to get their Social Security checks or preventing PC users from getting fixes for important programs. That's control.

These two principles the idea of making something freely available and preventing anyone from controlling it -- made Linux what it is today. Linus Torvalds created an operating system that was available to anyone, at no charge. All you have to do is download it. And he made it available under an unusual kind of license: You're free to change what Torvalds created -- you can even charge money for your changed version of what Torvalds made -- as long as you make the code available so that somebody else can change YOUR version.

Does that make sense? The first time I encountered this philosophy, I thought it was nuts. That was about 10 years ago. Reality can make believers out of doubters every time. What Microsoft has done with its attitude of control and conquer is clear: It's turned the strange notion of free (and freely available) software into a very real and amazingly effective approach.

I said a minute ago that under the Open Source idea, you can even charge money for your changed version of software. If you take that word "change" and substitute "improve," you'll see this a lot more clearly. You can get any version of the software very easily. You download it. That version comes with its source code (the code programmers work with), making it easy for you (if you know how to program) to change the source code and thereby change the program. (You might do this to fix a bug, to add a feature or to create a custom version. And "knowing how to program" doesn't mean you have to be a nerd who designs rockets on the side. Fixing program code or making changes can be relatively simple.)

You make your changes, and then YOUR new version is put on an Internet site somewhere along with ITS source code so that someone can fix a bug you missed (or one you inadvertently created), and so on.

What you get with this approach is very clear as soon as you look at the daily lists of Linux software on the Web: You get revised versions of Linux software every day of the year, many times a day, even on weekends and holidays. You get fixes for bugs usually within the same 24-hour period that the bugs were found. You get new functions and features sometimes within a few hours after someone posts a message explaining why such functions or features are needed.

This is all amazing if you're not used to the way the Open Source process works. Log onto one of the big Linux file sites in the evening and you'll see hundreds of new programs and other files. Log on again first thing in the morning and you're likely to see many revisions to those new programs, created overnight, often within minutes after one of the tens of thousands of Linux programmers around the world found a bug or pointed out that something wasn't working properly.

One of the sites that tracks new software and support files for Linux reported an average of 7,400 new files a day on a typical month in the late spring of 1999. In case you think that was a typographical error, I'll repeat it: 7,400 new files a day. Nothing like this phenomenon has ever happened; no one could have imagined it. It has got to make Microsoft chairman Bill Gates and his million-bucks-a-week buddies very nervous. Think about the disparity in operating systems for a moment. Microsoft Windows has 90 to 94 percent of the market in PC operating systems. There are perhaps 200 to 300 million PCs in daily use running Windows and only 8 to 12 million in daily use running Linux (numbers are almost impossible to get with any accuracy). Yet one site that tracks files, just one site, reports 7,400 new files a day for Linux PCs. There aren't that many new files in an entire week for Windows PCs, despite the lopsided ratio of Windows PCs to Linux PCs.

With this as a background to the skyrocketing interest in Linux, you should be able to see right away why so many Windows users are turning to Linux, and why so many companies that rely on PCs for their daily business are switching to an operating system that few people had ever heard of a couple of years ago.

Microsoft had it made. Microsoft could have done anything. That is, Microsoft could have done anything except what it actually did. And that's the lesson of history: When you're too big, too brazen, too high above what's really going on, sometimes you have the power to do anything you want, and more of than not you do precisely the wrong thing.

Microsoft did precisely the wrong thing. It had captivated the world and dazzled just about everyone with Windows. It sold so many copies of Windows that the money that came into Microsoft embarrassed the people who ran the company. They got so hyper-rich and their employees got so super-rich that no one wanted to talk about it any more. (Bill Gates was worth $90 to $100 billion -- that's "B" as in BIG in the middle of 1999. There may be as many as 2,000 Microsoft employees who are worth more than $1 million each.)

The wrong thing was to ignore what customers were saying. They were complaining that Windows crashed. That PCs locked up. That programs ran out of memory. That Windows itself ran out of memory. (Windows is called an operating system and it can't operate the computer. That's what "running out of memory" means.) They were complaining that Windows even ran out of memory when PC users went out and bought extra memory and spent all weekend putting it in. (I have a 384-megabyte Windows 98 PC that can run out of memory in just a few minutes.) They were complaining that Windows was full of bugs. That Windows couldn't run for more than a few days without crashing. That Windows corrupted its own files and refused to work. That Windows eventually will refuse to run no matter how carefully you run the PC.

And Microsoft didn't care. The money was rolling in and that was much more important.

When Windows 95 came out, Microsoft pulled off an amazing triumph of public relations buffoonery. The company convinced nearly everyone that Windows 95 was VERY different from Windows 3.1 (the previous version). It would be stable, would run forever, would handle memory properly, and on and on and on. (I get nauseated thinking about it now.) Of course, none of those things is true. Neither Windows 95 nor Windows 98 is stable, nor will either one run reliably for more than a few days. (In many business situations, they won't run reliably for more than a single day between reboots.) And Windows 95 and Windows 98 are textbook examples of how to mismanage memory.

But we fell for it. Then.

Fool me once, shame on you. Fool me twice, shame on me. Shame on Microsoft for diddling when it should have overhauled Windows. Shame on Microsoft for refusing to take its users seriously. Shame on Microsoft for playing all of us for fools. Shame on us for the intellectual laziness that allowed us to fall down and play dead.

Because one thing we all have in common is a need to get things done on our PCs. If you have only one car, the rust you see on the roof, the rattle you hear in the trunk and the hole you found in the fender might bother you but won't keep you from driving that clunker to the store. If you have two cars, suddenly the old one looks like it really is -- tired and old and uninviting. When we had no choice but Windows, the flaws in Windows seemed unimportant. We just worked around them as best we could. Now that we clearly have a choice, the flaws in Windows take on a new dimension: They are unacceptable, and they make Windows look like exactly what it is, tired and old and clunky and stuck together with baling wire and chewing gum.

It's no wonder that companies are looking elsewhere. Home Depot, the fastest growing store chain in the history of American commerce, is opening an average of two new stores a week, and has 90,000 Windows PCs at its stores helping manage things. The PCs are used for inventories, for company e-mail, for ordering and for dozens of other chores. Home Depot said in late spring of 1999 that it couldn't continue to use Windows 95 and Windows 98 because they are not reliable operating systems. It had a choice: Windows NT (the big brother of Windows 95 and 98) or Linux. Switching to Windows NT for all the PCs at its stores would cost Home Depot millions of dollars more than switching to Linux. (My calculations, which could be off in either direction, show that NT would cost $30 million to $50 million more than Linux for all those PCs.)

Home Depot's calculations include the cost of support, which is very high for NT and quite low for Linux, and the cost of what is called "down time" -- periods when a computer is not working because it crashed or failed in some other way. Linux PCs have almost no down time, whereas Windows NT is notorious for crashing under heavy loads.

Obviously, Linux is a much different operating system from Windows. Where it seems to work the same, it's actually quite different. The mechanisms it uses vary greatly from the way Windows does it, and, while it seems to look a lot like Windows, the real commonality is in graphical interfaces. The two newest ones for Linux -- Gnome and KDE -- do a much better job than Windows does of some of the things you need in an interface. (KDE has been around long enough for Microsoft's engineers to have seen it and even used it on Linux PCs in their PC laboratories, and the talk of the Internet among KDE fans recently was that Microsoft is copying some of the functions of KDE in the next version of Windows.)

I'll describe the way Linux works using KDE as its graphical interface. Technically, KDE is a "desktop environment" (hence the name -- the "K" doesn't mean anything) and not a graphical user interface (GUI) because Linux makes extensive use of a multipart graphical interface, and KDE is just one part of it. (The GUI in Linux is based on the Xwindows system, which is not related in any way to Microsoft Windows. Xwindows actually runs on all sorts of computers and a variety of operating systems, and you can even run Xwindows under Microsoft Windows if you're really desperate.)

The Linux version I'll describe is Caldera OpenLinux 2.2. Caldera is a company that packages the Linux "kernel" (the core of the operating system) with a lot of basic supporting software into a "distribution" of Linux. Many other companies also do this, including, as many PC fans know, Red Hat. "OpenLinux" is Caldera's trade name for its distribution of Linux. And "2.2" is the version number of the Linux kernel that is at the core of Caldera OpenLinux 2.2. (This should tell you that Caldera OpenLinux 1.3, which is still being sold in stores around the country, is an older version of Linux. Don't buy it.) You may be reading this a long time after OpenLinux 2.2 passes into history; if so, look for a later version.

What I describe about Caldera OpenLinux 2.2 will apply generally to all modern versions of Linux that have graphical interfaces, but keep in mind that there are some features and functions that Caldera handles differently. If you read about something here, don't run another version of Linux and then complain that you couldn't find the feature I referred to. I can only vouch for Caldera OpenLinux 2.2, since it's the only version I have worked with in the short time I've been using Linux. (That will change, but not too soon. I want to learn the basics of Linux thoroughly first, and that's keeping me busy enough without trying out different desktops and frills.)


Caldera OpenLinux 2.2 installs from Windows or from its own CD-ROM. In other words, if you are running Windows 95 or 98 (not Windows 3.1), you put the CD in the drive and run the installation program just as you'd do if you were installing a program from Microsoft or any other company. If you want, you can just install OpenLinux from the CD and forget about Windows. Your PC has to be able to boot from a CD-ROM in either case. Most can do this, but you may have to turn that feature on in the BIOS setup. (You get to the BIOS setup by pressing a combination of keys or maybe a single key just as the PC starts to boot up. You should see instructions on the screen that tell which key or keys to press.)

The Windows-based installation program puts a copy of a "lite" version of Partition Magic and a copy of Boot Magic on your Windows PC's hard drive. They're both made by PowerQuest, an old company in the software business. Partition Magic shows you how much space is available for Linux on your hard drive and then, if there is enough free space, prompts you to create a separate partition for Linux. You actually need two separate partitions for Linux -- one for the Linux files themselves and one for the Linux swap partition. (Unlike Windows, which uses a swap file, Linux prefers a swap partition, which is accessed faster than a swap file would be. Linux can also use a swap file in a pinch.)

A big flag should be waving over your head by now. Yes, you can install Caldera OpenLinux 2.2 side-by-side with Windows, but only if you have enough space to do that. (Please try to understand this. Files take up space, so if your hard drive is out of space, you don't have any space for new files, and you don't have any space for Linux, or anything else, for that matter. So clean out your Windows files before you start installing OpenLinux.) You'll need a lot of space to install all the Linux files you really need. You need about 1 gigabyte free to install everything except the source code and developer files. ("Everything" includes StarOffice, Netscape Communicator and WordPerfect 8.) Don't even put the Linux installation CD into your CD-ROM drive until you've cleaned out your Windows files. (If that's just not possible -- I know what I'm talking about here, having spent hours trying to pick up a few extra megabytes of space on one PC's smallish hard drive -- buy a larger hard drive. They're cheap.)

Once you have enough space on your C: drive, Partition Magic allows you to create Linux partitions. (It may create only one; the swap partition may be taken out of that single Linux partition later. This is not a problem.) Then the Caldera OpenLinux setup runs after an initial Linux bootup. You're asked to confirm the choices the setup program has made for your keyboard, mouse and so on. They're likely to be right, so don't change them unless you already know they're wrong based on experience.

You also will run through a video-card and monitor setup. I had so much trouble at this point that I should explain what happened and tell you how I got around it.

As far as I knew, because I had never installed or run Linux before (and had never installed or run Unix before, either), the Caldera installation was a bust. A dodo. A turkey. The first time I installed it, all I got was a black screen. The second time, all I got was a black screen. The third time. The fourth. (I'm patient!) The fifth.

An errata sheet enclosed in the CD-ROM box said that some video cards (including mine) confused the installation and caused it to fail. It said I should open a terminal and run XF86Setup. (It gave the location of XF86Setup, but the location it specified was wrong. No problem; I found it easily.)

This was both my salvation and my undoing.

I realized, after reading the note the 200th time, that Linux probably was running OK after all. I figured it must be the graphical part of the installation that had gone bad, not the installation of Linux itself. My hunch was correct, of course -- hunches based on reading something 200 times usually are -- so I read and reread everything I could find on how terminals work in Linux and discovered that pressing Alt-F2 allowed me to type command on the screen. (If I am remembering right, I could only do that if I also pressed something else first. Forgive me for lacking the specifics, but when you are frustrated, angry, upset and totally confused, you tend to hit a lot of keys at the same time, so I may have hit Esc, Q, PRNDL or some other combination to get Alt-F2 to work. Then again, I might have just pressed Alt-F2 to get Alt-F2 to work, too.)

Knowing that the black screen wasn't fatal meant I could do something about it. But that was my undoing. Running XF86Setup was a disaster -- always. It either started up with the display cut into four quadrants (all too fuzzy to view) or it started up with the wrong mouse settings. There was no way to get it to start up with a viewable display and the right mouse settings. (Apparently, XF86Setup was confusing my mouse with another model every time I told XF86Setup to ignore the current settings. This gave me a viewable display but made the mouse unusable. And there is no way change some of the settings without a mouse. I challenge anyone to try.)

So I ran the non-graphical setup program that the Linux loader (or some other part of Linux) mentions when there is an obvious problem with the Xwindows setup. After going through all of its many options three times and having the same problems that I had to begin with, I gave up on the idea that I would ever get Caldera OpenLinux 2.2 to load properly with its graphical interface. I wept for a long time, gnashed my gnteeth, wrote out my will and gave up.

As you will know if you are male or have known men at some time in your life (being born is also a qualification), there is no such thing as "I give up" if you are a testosterone junkie. What "I give up" actually means to most men is "I think I'll come back to this tomorrow or maybe in five minutes, whichever comes first." So that's what I did. I went through the entire installation procedure again. Then again. And again. I think I did it more than 30 times. Each time, I specified a different video mode for my monitor, hoping it would work.

Finally, it did work. I was ecstatic.

My ecstasy lasted two hours. I realized when I was about to write down the video-mode setting that I had finally chosen as the ONLY successful one that I had no idea what that mode was. I was paying no attention to it whatsoever.

So I did the installation again and paid attention this time. It only took five more tries to repeat the success. (I never did write down the proper mode. I didn't need to. I have it engraved on my frontal lobe.)


It's much easier to install OpenLinux if you're not worried about what to do with Windows. You go into the BIOS setup of your PC and make sure your PC can boot from the CD-ROM. If it can't, I don't know what you'd do. (Sorry. All my PCs can boot from the CD.) Then save the settings, put the Caldera CD in the drive and reboot. Linux will start loading and the setup screens will appear.


OpenLinux 2.2 uses the KDE desktop. KDE is a consortium of volunteers who work under the Open Source principle, creating software that everyone has access to and anyone can fix or improve. The main competitor is the Gnome desktop, part of the GNU project. GNU is another organization. GNU simply means "GNU is Not Unix." (GNU is the main force behind the Open Source movement.) Gnome probably would never have come about were it not for an oddity that Open Source fans will understand right away while the rest of us scratch our heads. KDI uses programming code from a commercial company, under a friendly arrangement. (The folks at KDE had permission.) This comapnyother source was not making its code available under the Open Source guidelines, so the GNU people had a conniption. They decided to design a totally new desktop that didn't use any non-Open Source code, and Gnome came about. (It's still a little buggy.) Meanwhile, back at the ranch, the KDE people got the other company to back down and release its code to KDE with an Open Source license, so Gnome changed from being a Big Necessity to Open Source users to being Just Another Interesting Desktop. For a while, everybody seemed to be jumping ship to Gnome, but that lasted only a few months. KDE is back in good graces and is a solid "product." (Whether something that is free is a "product" in the classic sense isn't clear, but it's surely a product in every other shading of the word, so let's not quibble.)

In some helpful ways, the KDE desktop looks and works like the Windows desktop. Icons arrange themselves along the left (the wrong place, I believe, but you can move them to the right just as you can in Windows), and the KDE launch menu, corresponding to the Windows Start Menu, is right where the Start Menu would be on a Windows PC. The launch menu pops up from a button that has a KDE logo, and the tooltip that pops up when your mouse hovers over that button is a play on a tired Microsoft slogan. It says "Where do you want to go tomorrow." The launch menu opens from the left edge of a launch bar, not from a task bar. (KDE has a separate task bar.) The KDE launch bar is about twice as thick as the task bar, which sits right on top of the launch bar. Windows 98 and all versions of Windows 95 that have Internet Explorer 4 or 5 also have a launch bar and a task bar, so the KDE arrangement will have a familiar look to many Windows users. (And, as you can do in Windows, you can turn off the launch bar or make it smaller.)

The launch bar has a few other icons on it besides the KDE menu icon, and you can add other icons easily -- or sort of easily, I suppose. You can try dragging program icons down to the launch bar. Sometimes that will work and sometimes it won't. I'm not sure why, except that it obviously has something to do with the type of icon you are dragging. If it's a KDE link icon (a shortcut icon already, in other words), it probably will copy itself to the launch bar OK. It's not hard to make KDE link shortcuts, so you'll get used to the way KDE does it fairly easily. (Windows does this better, for sure.)

The launch bar has tiny arrowheads on its left and right ends. Click one of them and the launch bar slides out of sight in the direction of the arrowhead. The task bar then slides down to fill the space left by the launch bar, and the launch bar leaves only a tiny button to show that it still exists. Clicking on that button slides the launch bar back out. You can also make the launch bar or task bar hide just like the Windows versions can hide.

Unlike the Windows launch bar and task bar, the ones in the KDE desktop know how to behave when you place them at the top of the screen. I moved mine up there immediately -- that's also where my Windows launch bar and task bar are located -- because that forces the launch menu (the KDE start menu, in other words) to drop down instead of pop up. That seems much more natural to me. I've also found that sliding the mouse up to the top of the screen to do something is an easier movement than sliding it down to the bottom of the screen. My wrist doesn't do the downward arc as well as it moves in an upward arc.

The designers of the Macintosh and of Digital Research's GEM operating system and even the designers of Windows seem to agree with me; they all have their menus at the top, not the bottom, of their windows. Rumor has it that Microsoft put its launch bar and task bar at the bottom just to make Windows look different from the Mac, but in fact the reason is much more mundane: Windows is not able to control the way program windows open and move on the desktop. As a result, windows that take up the area along the top sometimes cover up the launch bar and task bar. This failing of Windows is well known, and Microsoft had plenty of time to fix it yet refused to do so. In order to keep program windows under control, I've had to use a utility that pushes windows of any kind back onto the screen when they slide off (as many do) and forces windows out of the launch bar and task bar area when they move up too far. This is called Shove-It. You can find it at any big Windows file site.

KDE won't let windows misbehave, so it's a no-brainer to move the launch bar and the task bar to the top of the screen. I don't think they can be dragged up there, as you can do under Windows, but they can be positioned from the Control Panel.

The task bar behaves just like the task bar in Windows 98 (not Windows 95). In KDE and in Windows 98, clicking once on the task bar entry for an open window minimizes that window, and clicking on the task bar entry when the window is minimized restores it to what it was before. (Chances are, most Windows 98 users who read this are now playing with their task bars to confirm that this is, indeed, the way the task bar behaves in Windows 98. As is typical from Microsoft, you never get any clue that Windows behaves this way until you stumble on it. In Windows 95, clicking on the entry for a minimized window has no effect whatsoever.)

Icons behave the way icons behave in most other graphical interfaces. You can select them by clicking on them or you can lasso more than one at a time by clicking on a blank area outside the group of icons you want to select and then holding the button down while you draw a "marching ants" box around the icons. All mouse clicks are single clicks in KDE. On other words, a single click on the icon for Netscape runs the Netscape browser, no matter where that icon is located. (In Windows, the standard mode requires double clicks in some instances and single clicks in others. KDE makes everything a single click.) Single clicks also select icons, so you'll find yourself opening windows when all you wanted to do was select the icon. Get around this by holding down the Ctrl key when clicking.

The right mouse button does the same sort of thing the right button does in OS/2 and in Windows. OS/2 had it more or less first -- after collaboration of some kind between IBM's OS/2 programmers and the team of two or three genius-level programmers who pioneered the modern desktop in the outstanding GeoWorks system -- but Windows deserves credit for turning the right click into a vastly more useful tool than it was in OS/2. A major difference between the way KDE does normal file-and-folder dragging operations turned out to be very welcome. When you drag a file or folder in Windows, you have no idea what's really going to happen unless you are a Windows expert (and even the experts don't know how this works some of the time). What's really going to happen depends on a lot of things -- are you dragging a folder to another drive? Then it will be copied. Are you dragging a folder to another location on the same drive? Then it will be moved. Are you dragging an executable file? Then it won't be copied or moved no matter what; you get a shortcut instead. There are even more exceptions and quirks, all signs of a programming operation in which a thousand people make the decisions that three should make. (Too many cooks have spoiled the broth, in other words.) The fact that this has remained a "function" of Windows 98 long after the introduction of Windows 95 proves that Microsoft has no idea how to fix something once it's obvious that it's not right. (Even if the company looked only at its bottom line and realized that getting rid of this nonsensical behavior would save millions of dollars in support costs, it would not change the way Windows does this. Microsoft does not like to admit that it did something wrong, even when it would save money by doing so.)

In KDE, whenever you drag an object and let go of the mouse button, you get a pop-up menu that has three choices (and one implied choice) -- Copy, Move or Link (with Cancel implied -- just hit the Esc key). This is the way Windows should work. There is no way in KDE to move something when you thought you were copying something, no way to copy something when you thought you were moving something, and no way to get a shortcut when you thought you were moving or copying. (Whew! You'd think Microsoft would have caught on by now.)

By default, KDE uses "smart" placement of windows on the screen. The first window is always placed at the upper left (another reason to move your icons to the right!) and each subsequent window is placed, as much as possible, in a location that keeps any windows from overlapping. You can turn off "smart" placement and use any of three other, less automatic, methods. I think you'll like "smart" placement best.

Windows can be moved and resized the same way they are in Microsoft Windows. However, KDE adds a popular Unix-desktop feature that you may find indispensable: Clicking in a window while holding down the Alt key lets you drag the window anywhere. You can turn this feature off if you want. (Don't. You'll regret it.)

By default, each user of OpenLinux 2.2 has four desktops, but the maximum is eight per user. You can switch from one desktop to another by clicking a button in the launch bar. Windows that are open on one desktop are not open on any of the others unless you make them "sticky" by clicking a button on the window. (This is an outstanding feature.) You can move a window from one desktop to another by clicking on a menu item. Each desktop has its own background and color scheme, also. Desktop icons remain on all desktops. A "pager" program shows your desktops in miniature. They're not iconic representations but the real things, as you'll see if you make the pager window larger: You'll start to see the actual miniatures of windows on each desktop. You can drag windows around on the real desktop by dragging them within the pager's mini desktop, and you can reposition windows on the real desktop by moving them in the pager window. It's all very nice.

Desktop backgrounds work just like they do in Windows but have many more options. You can tile small images or center large ones, and you can enlarge any image to fill the screen in a different ways. You can also place an image that's smaller than the screen against a brick background or another patterned background. Icon captions can have solid backgrounds or transparent backgrounds. (Windows does not give you this choice by itself, although you can get free utilities that do it.)


KDE comes with a marvelous file manager called kfm -- all KDE programs are named in lower-case letters, as you quickly find out when you run Linux -- that works like an improved version of Windows Explorer. The improvements first:

Now for the only real complaint. The kfm file manager is slow browsing Web pages. It does surprisingly well at all the basic things a browser is supposed to do, and no doubt there are many users who would be happy using kfm as their main Web browser. (After all, you can hardly ask for an easier way to handle everything. The same program you use to view files and folders on your hard drive is also the program you use to view files and folders on somebody else's hard drive -- on the Web).

But kfm needs some attention if it's going to be satisfactory as a replacement for Netscape Communicator. I'd love to see that happen. There's an undeniable pleasure in using a well designed Open Source program, but no one can deny the other pleasure we all enjoy -- the one that comes from using a superbly designed program, from any source. And Netscape Communicator is just that kind of program.


The KDE desktop comes with an extraordinary file browser. It works so well that as soon as you run it for the first time, you wonder why Microsoft didn't do the same thing. It opens a cascading list of folders and files of just certain areas of your system (such as your personal folders, for example) or the entire storage system (so that you can browse every file and folder).

You'll know right away how to use this file browser because it works just like the Windows Start Menu. A major difference, of course, is that the KDE browser can look through all folders, not just the one called "Start Menu" -- that's how the Start menu actually works, by popping open a cascading list of files and folders under in a folder named "Start Menu" in the Windows folder. If this lights a bulb in your head telling you that Microsoft surely must know how to create a file browser like the one in KDE, you can switch the lamp back off and congratulate yourself for being so smart and so dumb at the same time. Microsoft does know how to do this, but the reason it has not done it is not related to knowing how it's done; Microsoft just doesn't care how it's done. (Tell me a company that cared how its operating system works would make one year after year that crashes every few days and I'll listen to your dissenting opinion.)


Microsoft never really called the area at the bottom right where the clock is located the "System Tray." Officially, it's the "notification area," a better name, perhaps, because programs that need to work in the background can use flashing icons or little lights in the tray to tell you something. (I have a couple of utilities for Windows that do that, and it's a great idea.) KDE has the same thing, except that it's larger. (It seems twice as big, but I didn't measure it.) Just as with Windows, some programs seem to want to use the tray and others don't, while a few programs can be put there if you try hard. No doubt the tray will be improved at some point.

Some of the Linux programs I run show themselves in the tray. They include the audio mixer, which shows a speaker icon in the tray; the desktop-properties applet, which shows an appropriate desktop-looking icon; a system monitor, which shows, instead of an icon, a chart of either CPU or Linux kernel activity; a program that displays the phase of the moon in the tray; a clipboard utility; a sticky notes program and a personal information manager.


The version of KDE that comes with Caldera OpenLinux 2.2 has color schemes and sound schemes but not desktop themes. In other words, you can change the sounds that play when you do something like run a program or close a window, either individually or by choosing a theme, and you can change the color schemes of the windows and desktop background the same way (by changing individual items or choosing a theme). Desktop themes are different; they change sounds and icons and pointers and window colors all at once. They can be added but were not included in OpenLinux.


Linux is not Windows, nor is it a Macintosh or an Amiga or anything else. The KDE desktop defaults to a look and feel that is different from Windows and the Mac. But you can make your KDE Linux desktop look very Mac-like or very much like Windows with a couple of mouse clicks. This is done through the Control Panel, which has many more options and is even easier to use than the excellent one in Windows.

Through a freeware utility program, my Windows PC has had Mac icons on the desktop and it has even run for a few days with program windows that looked like they came from a Mac. This kind of change is easy in Windows if you know how to change icons and know how to get different shapes and colors for windows themselves. But that's just cosmetic. KDE lets you make cosmetic changes, too -- far more extensive ones than you can do with utilities for Windows -- but that's only the beginning. KDE lets you change the behavior of the graphical interface so that it mimics the way Windows works or the way the Mac works. Or, of course, you can choose the KDE standard interface. You can, for example, switch the interface almost entirely over to a Mac operation, with the Macintosh menu bar at the top of the screen. (As a big fan of the GEM interface on the Atari ST and TT computers of the 1980s, I was also a fan of the Mac-style menu bar at the top of the screen, since both GEM and the Mac used that design. If you've never used a Mac or GEM on an ST or TT, you might be wondering what I mean by a menu bar at the top, so I'll explain briefly. Under Microsoft Windows and most other graphical user interfaces, each window is self-contained; menus are located within each window, and windows ordinarily do not take up the entire desktop. The Mac and GEM treat this differently: Windows that are in the foreground place their menus at the top of the screen. This has a huge advantage in normal use, because you never have to wonder what program is in the foreground; the menu bar tells you right away. It has two disadvantages: A window that is not in the foreground can be hard to find (it could be hidden away and accessible only through a separate list), and a window that's a lot smaller than the screen's overall display size will present an challenge in ergonomics, because you're working within one area of the screen while manipulating menus that are far away. To me, this is the main reason this kind of interface is a bad thing. I suspect the newest Macintosh operating system, due out at the end of 1999, will abandon the menu bar at the top or make it optional.)

You can switch KDE over to the Mac look without also getting the menu bar at the top. I find this a delightful escape from the generally ugly industrial-design look of the standard Windows interface and the uninspired but serviceable look of the normal KDE look and feel. You can also mix and match, adding Windows-style check boxes (a good idea, since they are easy to read from a distance) to Mac-style window treatments (with the close box at the upper left, for example) and Motif-style "pin" buttons (which make a window visible on all desktops).

Basic window and mouse behavior is also customizable. A left double click on a window title bar can roll up a window (or roll it back down if it's already rolled up), or it can toggle a window to its maximum size or back again. Rolling up means making a window disappear into its title bar the way a window shade would roll up into a valence. Modern Macs can do this and so can Windows if you add a utility. Linux and Unix computers have had this feature for a long time. You can change the layout of all window gadgets (the buttons at the corners of windows) and you can even get rid of parts of selected windows based on rules -- if a window with "Control Panel" in its title opens up, for example, you can have KDE automatically strip off the title bar. (Tricks such as this can help produce a completely customized desktop.)

Alas, the slick KDE interface with all its customizing tricks can't make Cinderellas out of some of the programs you're likely to run under Linux. These programs will still look and work like the wicked sisters. Let me explain.

As anyone who has used Windows a long time knows, there are Windows programs and there are DOS programs and there are oddball programs that don't seem to be Windows or DOS. (Often, the oddballs are games.) Nothing that Windows can do will make a DOS program look and work like a Windows program, and nothing Windows can do will make one of those oddball weirdo programs look normal either. You're just stuck with things the way they are.

And you're just stuck with the same kind of thing when you run Linux. You have command-line programs that run in a sort of DOS-looking window (or that run full-screen, just as DOS can run in a window or full screen). You have programs that come with their own graphical interface. You have Xwindows programs, which use the Xwindows basic interface. And you have KDE programs.

Command-line programs are extremely important to Linux. You'll just have to learn how to use some of them. Don't get carried away: Books on Linux tend to be written by long-time Linux users who revere all that command-line stuff, yet KDE has taken over a lot of the chores by making menued programs out of many of the command-line ones. (A great example is the file manager, which takes over brilliantly from a group of command-line programs.) The best thing you can do is learn how to use bash -- yes, that's the name, in lower case -- the shell program in Linux. There are other shells you can choose, and OpenLinux comes with two others. But bash is ideally suited to Linux. There are times when you have to use bash to do something important.

Programs that come with their own interface should be avoided whenever possible. Sometimes you just can't avoid them. One such essential but troubling program is StarOffice, which uses the standard Xwindows conventions in most ways but has a customizable window interface. You can choose a Windows 95 look, a Mac look, a Unix-Motif look or a StarOffice look. The problem is that ALL of the choices are unsatisfactory in small ways. You'd also find it odd to have KDE set up with a Windows 95 look and feel, yet have the Mac look turned on in StarOffice. The best setting for StarOffice is the StarOffice setting, which makes the fewest changes between the KDE look and feel and the StarOffice window behavior.

Xwindows programs are quaint, showing evidence of a very old interface, yet they're perfectly functional. KDE programs work seamlessly with the KDE desktop and behave the way modern programs should.


Linux can be a jolt to anyone who is used to Windows. Perhaps the biggest difference in daily use is the way file and folder permissions work. You can't do anything in Linux unless you're given permission to do it. In a single-user Linux system, you're two users, not one: You are a standard user, "joe_user" or whatever you call yourself, and you are "root." That's the supervisor, or administrator, of the PC. No PC can use Linux without a supervisor, sometimes called a super user. (There is a Linux command called "su." Most users probably think it means "super user." It doesn't -- it actually means "substitute user" but all anyone generally uses it for is to log in as the "super user" when something drastic needs to be done.) So a one-user Linux PC has at least two users, "joe_user" and "root." In truth, a typical Linux PC has many users, but most of them are not expected to be humans; they are more likely to be operations that are given permission to do certain things, and, to make things easy on the Linux accounting system, they're listed as users.

This requirement that each PC have an administrator can get in the way big time. I found it to be the single most annoying thing about Linux. Old Linux hands will laugh, but the problem I had will be faced by hundreds of thousands of users in the next few months and millions in the next year. It's a crazy, needless and mindless problem. My PC should be smart enough to know how to handle security without causing me pain and suffering. Linux needs major changes in this area -- not to reduce security but to increase user friendliness.

And there's the lack of support within the operating system for a lot of the things many Windows users take for granted, such as on-board video and accelerated graphics cards. My second Big Annoyance with Linux has to do with the graphics card -- ATI All in Wonder Pro -- in my Linux PC. I'll leave out the technical part (which would simply explain that it's not really Linux that's the problem but the "Xserver" system -- you see why I'm skipping this part?) and tell you that the screen images are stupendous and wonderful and I love what I see and all that. But this card is an accelerated graphics card, meaning it has a chip that speeds things up by a tremendous amount if the chip is told to do so. Alas, under Linux this chip is left unawakened. Double alas, the TV capabilities and video recording functions of this card are totally ignored under Linux. This may change before long -- the ATI company has at last started to help put a working system together -- but that's not much cheer right now.

This hasn't stopped me from using Linux. They're annoyances, and I've lived with Windows annoyances long enough -- look at my Windows tips for a sympathetic view -- to know what will ring my tolerance buzzer. And nothing I've seen so far in Linux will make that buzzer go off. The operating system is just too good to let annoyances get in the way. And the Open Source system will find fixes for all problems; it always has.

 Image courtesy of Adobe Systems Inc.technofile: [Articles] [Home page] [Comments:]