I’d like to invite you to read the last of three articles from the “Dump Windows” saga of writings. Two previous articles (part I, part II) stirred up quite a controversy and a variety of opinions. With this article I’m hoping for calmer and more meritorious responses. In the last article of this series we will touch on the aspects of work and usefulness of the console, remote access, logic behind the OS, pricing, and TCO.

Author: Borys Musielak

7. Power of the console

A console (terminal, terminal emulation) and a shell are equivalent to what is commonly known in Windows as the ‘command line’. For many it’s just a mysterious and unnecessary system tool, reserved only for the ‘1337′. For some, professional system administrators, although limited it’s indispensable. Linux console, in contrast with the Windows command shell, is an extremely powerful user tool, which allows one to perform virtually any operation on the system.
The combined power of the console and the shell (i.e. Bash) enable us to perform, among thousands, the following example tasks:

Pic.1 Software installation requires
no more than just one command.


Pic.2 Alsamixer - useful for
system sound settings changes.

• detailed and extensive system-wide configuration and administration,
• software installation, updates, and removal,
• file and folder administration and manipulation (using i.e. Midnight Commander),
• sharing, accessing, and transferring data between machines using SSH, NFS, or Samba,
• managing personal email (mutt, pine), web browsing (lynx, elinks), reading discussion groups (slrn),
• listening to the music (mpd/mpc, mpc, mp3blaster, xmms-shell),
• burning CD’s (cdrecord, cdrdao) and DVD’s (growiofs),
• chatting via IRC or via many popular instant messaging networks like Jabber, ICQ, etc!
• extensive system monitoring and troubleshooting (top, netstat, io, iptraf),

What differentiates a Linux console from a Windows command line?

Simply said: its architecture and abilities.
Let’s dig a little deeper… First of all, the shell in Linux is a completely system independent entity, it is not an integral part of the operating system. Thanks to this simple architecture we can use several shells for all kinds of different purposes. What’s worth noticing the most is that we can launch a large number of consoles for different tasks, using different shells if necessary. It’s the shell that controls the functionality of such events as command history, command line completion, but also any kind of conditional statements, loops, and all other programming concepts used in shell scripting. The most popular shells include: bash, zsh, and ksh.

The terminal emulator on the other hand is a program (usually with graphical user interface, including menus, etc) which helps us to effectively use the console. The most commonly used emulators are: gnome-terminal, konsole, xterm, rxvt, aterm, Eterm and multi-gnome-terminal. Most of them are simply called “console’ or ‘terminal’ in the system menu. The terminal emulator is responsible for extra stuff like transparency, having multiple terminals in tabs, key shortcuts, etc.

The real terminal and the emulator

In many Linux distributions (especially those dedicated for servers) the default user interface is pure terminal. The graphical interface (X Window System) is optional and can be run on top of it.

In such a case, we have the access to the console just after login so we don’t have to use a terminal emulator run from the GUI. In most cases however (all desktop-oriented Linux systems), the default setting is to use the graphical user interface. In these systems we can also switch to the raw console using a key shortcut CTRL+ALT+F1 to F6).

Such architecture, which does not require a graphical interface is a great advantage of GNU/Linux and other UNIX-like systems. Few remember that a similar division was in place in old versions of Microsoft systems as well. MS Windows, up to version 3.11 could not be run independently and required DOS (”the console interface”) to normally function.

In Windows 95 this solution has been partly abandoned. The graphical interface was running by default, eating up valuable resources for server systems (but you could still switch to the DOS mode). In later versions of Windows (2000, XP), it is not possible to run the pure text mode at all!

Being constrained to using graphical interface is also an important restriction in the case of system crashes.
For instance, in case of problems with a graphics card driver, when Windows does not want to run, we have practically no sensible way to correct the error (e.g. by removing the driver and replacing it with a working one). If the plug & play magic fails, there is no option but to call support or reinstall the system. At least for regular human beings (since there are some advanced tricks to get around it).

Why use the console if there is a GUI?

Indeed… Why the heck should you use the console (except for the system crash case)? Isn’t it way easier and faster to click through things? Linux was supposed to be as easy as Windows you said! And now you come up with this geeky stuff…

Take it easy. Nobody says you have to use the console. You can configure your system, perform all the standard operations in the GUI mode. But.. I’m not sure you really want to. Using the GUI is easy but remember these two things. Console is faster. Console is more efficient.

Incredulous? I’ve got examples:

• Let’s remove a file. For instance, the file test.txt located in our home folder
  • In terminal: rm ~/test.txt (the file name is auto-completed when you enter TAB)
  • In GUI: Double-click the “Home” icon on the Desktop. Then select find the proper file, select it and use the right mouse button to choose ‘Delete’.
• Let’s do some configuration now. For instance, let’s configure the Internet access via DHCP
  • In terminal: sudo dhclient eth0 (and enter your password)
  • In GUI: Click on the icon representing the Network on the desktop toolbar. A network configuration applet shows up. Choose a network card represented by eth0, and click Activate. If it does not work, then enter the specific configuration for this card, enter the root password, set the DHCP option (probably static IP has been set before), click OK, and again click Activate on the network card. Finally leave the applet.
• Let’s do some admin stuff now and start the Apache2 web server!
  • In terminal: sudo /etc/init.d/apache2 start (and enter your password)
  • In GUI: Choose System->Administration->Services from the GNOME menu. Then look for Apache2 server and select it. Click ‘Apply’ and close the applet.
• OK, the last example is the least geeky one. Let’s watch a movie in MPlayer!
  • In terminal: mplayer /home/movies/Aviator/AviatorCD1.avi (of course using the TAB completion option for file names!)
  • In GUI: In GNOME menu choose Applications->Sound & Video->MPlayer Movie Player. When the program shows up, choose ‘Open’ and navigate to folder /home/movies/Aviator. Then click the first part (AviatorCD1.avi). Alternatively, navigate in your file manager (like Nautilus) to the right folder and then double click the movie file (or, if MPlayer is not your default movie player, select ‘Open With’ and select MPlayer from the list).

Pic.3 Console tools in action: mp3blaster, Midnight
Commander, gcc, FvwmConsole and top

These are only four examples. In each one, performing exactly the same task was a couple times faster in terminal than in GUI. Of course it’s not always the case. Burning a DVD in a console is for hardcore hackers only, since the growisofs program requires many parameters and typing them all manually is harder than choosing ‘Burn as DVD’ in a program like K3B or GnomeBaker. Of course, this “hardcore hackers” expression should be treated as a metaphor. You don’t need to be a geek to put the program name with its parameters into a small script and run it ad infinitum, typing in a console or a terminal window the script’s name only. (Anyway you need to know how to do it!)

In general however, if we care for efficiency in our work, it is a very good idea to get familiarized with terminals. It will take time to get to know the basics, but for a Windows power user like you, the learning curve should be shallow and bring tangible benefits resulting in faster and more efficient work with your operating system.

What is worth mentioning, after many years of ignorance, the Microsoft guys finally realized that an efficient console would be good for Windows users as well. They have been working for a couple of years on the new Windows shell fiest called MONAD, then msh and finally Windows PowerShell designed to make the system administrator’s work easier.

If you don’t own a license (and if you think you need one for Vista, think twice!), you are left with Cygwin, a UNIX shell emulator for Windows, with all the drawbacks of this solution.

8. Remote access in GNU/Linux

If you have ever tried to access a Windows box remotely, it is very likely you suffered a lot of frustration. Remote desktop works great but only when you have a stable broadband connection to your remote machine. In GNU/Linux you (as usual) have a choice - you can connect remotely, both graphically and text-based (for maximum performance) using at least a few of the methods described in this article.

As you already know, Windows XP works only in a graphical environment. Thus, when you connect to it remotely, you are condemned to use a graphical application. This is called “remote desktop” and it works great… but only when you have a stable broadband connection to your remote machine. When you are on a modem (like GPRS, EDGE) or another type of weak connection, using the remote desktop in a comfortable manner is just not possible. In the case of previous versions of Windows, it’s even worse since the RDP connection is not available there and you are forced to use third party tools like TightVNC or shareware radmin, which does not offer comparable performance to the Remote Desktop.

In GNU/Linux you have a choice (as usual) - you can connect remotely, both graphically and in text mode using a few of the methods described below.

• Connect via SSH
  SSH is a secure shell, the secure descendant of the telnet protocol. SSH enables you to access a shell on a remote machine in a secure manner (the transmission is encrypted). What can you do with the shell? In short: everything that you can do with the GUI plus a lot more. Check out the article The power of the Linux Console to see the details. Via SSH you can also run remote graphical apps on your local machine, provided that you have an X-server installed locally (this is the default in GNU/Linux and BSD and easily available in MacOSX; for Windows you need to run additional program for that, e.g. the X-server from the free cygwin package or commercial WinaXe). SSH is the most popular and the safest way to work remotely on a Linux box. However, if you need to access your whole desktop remotely, you can use one of the technologies described in the following points.

Pic.4 Skype run over an SSH tunnel — a viable solution for
restricted networks with blocked ports

• Pure XDMCP connection
  This X-server function is disabled by default in most of the Linux distributions due to the low level of security it offers (no transmission encryption). However, XDMCP connection via an SSH tunnel can be a good way to access your home desktop from a remote machine. It has to be mentioned that it needs a pretty fast Internet connection to work flawlessly.
• Desktop Sharing in KDE and GNOME
  K Desktop Environment has an option called “desktop sharing” which is a client-server solution provided by the kdenetworks package (krdc + krfb). Desktop sharing is realized through the RFB protocol (better known as VNC) and is compatible with any other VNC client. KDE Desktop Sharing harnesses its GUI interface for setting itself up both on the server and client level. We can use a neat wizard to enable remote access to particular machines and users, set proper passwords, etc.

On the other hand, GNOME provides its own framework for remote connections called Vino. The solution is also VNC-based. Vino is just another VNC server which is strictly integrated with GNOME. It allows you to access your GNOME desktop remotely in order to work or perform system diagnosis.

As far as the performance is concerned I haven’t noticed any significant difference between GNOME and KDE desktop sharing.

Pic.2 KDE Desktop sharing
— configuring a connection

• Connect using external tools like FreeNX or VNC
  Using external tools like different VNC programs (RealVNC, TightVNC, Vnc4 and more) or a less popular but much more efficient FreeNX, gives you another possibility to access your machine remotely. Both technologies enable you to log in remotely or access an existing X session (read-only or read-write). These two technologies are platform independent and can be used in any combination between GNU/Linux and MS Windows systems.

So, what do we get as extraordinary benefits when using GNU/Linux? The key thing seems to be the option to access a remote shell in a text-based way and the possibility to run applications remotely (without having to download the whole environment) thanks to SSH and tunneling. Desktop sharing in Windows and Linux is on a similar level. I could even take the risk of saying that the default Windows Remote Desktop is a superior solution to XDMCP or desktop sharing in GNOME and KDE. When platform-independent technologies are used (VNC, FreeNX or non-free NX client/server) the performance is similar in both environments. It has to be mentioned that the commercial NX package is the best solution when you are short on bandwidth.

Further reading:

• Secure Shell — the article in Wikipedia
• Linux XDMCP HOWTO — all you need to know about XDMCP
• KDE Desktop Sharing Handbook
• Remote Desktop Administration Using Vino

9. The Case of The Philosophy

The term philosophy is short for Operating Systems Philosophy which in turn comes from Systems Philosophy theory. In such a case the “philosophy” with reference to the operating systems means a form of systems thinking.

The cited definition is sine qua non of understanding the term, more so, it is frequently used without supplementing words or specifying its context. In this light the term “philosophy” might be seen as wacko hype and out of touch with reality or, if accepted, a bit too lofty.

I tend to view it as a set of those aspects of an operating system which a user gains an appreciation of after obtaining the system, and which refer to the underlying design of the system and its applications. Thus the philosophies of the Windows and the GNU/Linux systems are diametrically opposed. When I write “GNU/Linux systems” I have in mind the most popular distributions which are distributed free of charge. All these considerations would be meaningless if it was not for the fact that the system philosophy significantly exerts its influence upon the ways the system can be used by us and upon the system’s capabilities.

The Closed Source versus The Open Source

Microsoft Windows operating systems are distributed in binary form robbing us of an insight into their source code. This type of selling is based on customers’ full confidence in the manufacturer. Only developers from Redmond are granted the right to read the code and have the chance to correct the code and apply patches. This approach has its advantages and disadvantages. All know the notorious tenet security through obscurity which says in short that if a source code is a well-kept secret, the binaries can be full of security holes, because to make use of them one has to know them.

Detections of the security holes in closed source software is highly hampered. On the other hand users do not know how many security holes or bugs exist in their systems. Due to the “code obscurity” all the holes can go unnoticed (not for the “hard-working” crackers) and unpatched for years. Keeping such information secret not always easy.

It’s another matter with open source software. Nothing can be hidden here. The code is available to all and anyone with sufficient programming knowledge can make use of it. In theory, this looks like a big security hole in a system’s defenses. In practice, making the code public affects its quality. Mr Eric S. Raymond’s well known aphorism states that “Given enough eyeballs, all bugs are shallow.”

There’s no need to argue with the fact that if professionals and not necessarily a program’s authors are able to look thru the program’s code, then the probability of mending the security vulnerabilities increases enormously. It happens rather frequently that errors discovered in F/OSS software are accompanied with appropriate security patches, sent by a user who found the bug(s).

Nothing can illustrate the differences between open and closed software better than backup keys to our house. We can hide the keys in a place known to us only (e.g. under the doormat). This is an example of a blatant security breach. If a burglar checks the hiding place first we will be finished. But if we hide the keys in difficult and improbable place, the probability that no one will find them and will not make use of the breach is very high.

Another method to deal with the problem is to trust the keys with our neighbors. Here we face another problem. Our neighbor could be a crook and would give the keys to burglars. However, we can introduce our acquaintances to the secret or we can make the information public (nailing an appropriate note to door), and for all that there is a fair chance that our activities will not lead to multiple break-ins, even so the knowledge of where the keys are being kept is overt.

Windows Licensing versus GNU GPL

Basically there are two licenses the Windows systems are being sold with: OEM and FPP.

OEM license is cheaper one but has one essential flaw - “it integrates with hardware” with which it was bought. If we buy a PC with Microsoft Windows XP with OEM license we’ll be able to use the system only with the computer. Even if we uninstall the system we will not be allowed to install it on another PC. What’s more, the matter becomes nasty if we decide to exchange one part of the PC, e.g. old graphics card with a new one. Any hardware improvement ends up with a phone call to Microsoft Technical Call Center to get permission to use the license again but not without lengthy explanations. If the part which happened to fail was a PC’s motherboard we are in a deep trouble as the OEM license is tied up to the motherboard. “Deep trouble” means we will have to buy a new Windows system.

FPP license is a bit more liberal one but simultaneously nearly two times more expensive. It doesn’t mean a user was bestowed with more rights. Far from that. FPP is tied up to a user, not to a hardware. Thanks to that we can use the system on a new PC after throwing away our old junk. But again only after we make a call to Microsoft Technical Call Center and we tell them why we’d like to register once more.

In fact the whole idea behind the Copyright “magic word” can be boiled down to a list of bans. The forbidden actions (bans) simply tell users what can’t be done with the bought software.

Thank God, the GPL license with its famous Copyleft declaration made by RMS a “Copyright flipped over” has nothing to do with Windows licenses. GPL does not take user’s rights away but grants them. In particular it bestows a user with the four basic freedoms:

• Free Software Definition
• the right to run the program for any purpose (freedom 0),
• the right to study how the program works and adapt it to your needs (freedom 1),
• the right to redistribute copies so you can help your neighbor (freedom 2),
• the right to improve the program, and release your improvements to the public so that the whole community benefits (freedom 3)

The only condition the GPL license requires users to follow is that the corrected or improved programs have to be distributed under the same license. This condition prevents “closing” the free programs and prevents amending the code without making the changes known to the public.

The Microsoft licenses and those of the other firms using Copyright law present an entirely different philosophy to that of GNU GPL. They take a totally different approach to the software market and software availability. There is no doubt that GNU and Free Software Movement approach is decidedly more friendly to home users. But not only to them. Many companies like Red Hat and Novell make a lot of money creating Free Software. They make a profit both selling their products and selling their technical support.

Upgrading Systems

New Windows versions are being released once each 3-4 years. Microsoft publishes security patches and so called service-packs between new releases. Service-Packs are in fact aggregated packages containing both patches and new code which enhance the systems’ functionality. Generally all security patches are available free of charge and can be downloaded with theWindows Update application.

But the upgrade to a new Windows version is payable and it costs little less than a new product. What’s worse, Microsoft supports only the last three versions of its flagship product. Earlier systems like Windows 98 or 95 are left without support. Even Windows 2000 is serviced with security patches only. So users have to make a decision to buy or not to buy a new system. Otherwise they are forced to use their old systems fully laden with unpatched security holes.

In the meantime Linux distributions have a distinctly shorter release cycle (perhaps excluding Debian GNU/Linux). New versions appear in a six monthly or yearly cycle. New security patches are being published for quite some time (several years on average) for every release. For free systems like Ubuntu, Fedora Core, and OpenSUSE every upgrade is free of charge. Thus the users are not compelled to run older versions of their systems (often left without support). They are free to choose the time of the upgrade to suit their needs. And because new Linux versions run without problems even on very old machines (in contrast with Windows systems), the upgrade is not connected with lost efficiency.

This time I won’t draw any conclusions. It seems to me that the right inferences come to mind by themselves.

10. TCO And The Price

Last but not least — the price of the “operating systems”. Here GNU/Linux and other free of charge systems like BSD family are without competition. Most Linux distributions are both free as in freedom, as well as free as in beer. Even the cheapest OEM versions of MS Windows systems are priced for about 100$. And buying a new PC with Windows on board we rarely see the price tag as it is included in the selling price. But it should be kept in mind that there are alternative systems to Microsoft products. Many Internet shops and malls started selling personal computers with GNU/Linux preinstalled. Even Dell - one of the world’s biggest PC manufacturers - is equipping its PCs with GNU/Linux lately. PCs without operating systems can be bought as well. We can install Linux by ourselves and use the saved money according to our fancy, e.g. for additional RAM or a ring for our fiancée.

As the price disparity between the two systems is a fact, then the question of TCO (Total Cost of Ownership) is under never ending debate. The most notorious example of the debate is the duplicitous Microsoft advertising campaign Get The Facts focusing mainly on delivering false data about Linux systems (FUD). The answer to the “Get the Facts” are alternative campaigns led by IBM and Novell (among others) and intended to separate the truth from some Linux myths and debunk the tripe about the system.

Certainly the TCO matter concerns medium to big corporations more than home users as they implement Linux or Windows system on a large number of PCs and servers. But surely, also in the case of a home PC the eventual decision to migrate to a new system shouldn’t be taken without consideration of the system’s price and the cost related to learning and administering it. The costs vary a lot depending on the tasks we want to perform on our computer.

One way or another, we learn Windows system only once whereas we have to pay many times if we want to upgrade to newer versions. One expenditure of 100-200 $ in several years for the world’s most popular system seems to be small. But if your home budget is tight I would suggest reconsidering the decision to buy a new Windows version and give a try to one of the free of charge GNU/Linux operating systems. The risk is minimal and being patient we can save quite a bit of money while simultaneously improving security, comfort, and effectiveness of our home system.

Summary

If we draw a comparison between the two operating systems considering licensing, selling, and upgrading models, and if we add the supreme configurability (resulting from the system’s philosophy), and a low pricing then it will appear the GNU/Linux is decidedly more user friendly in terms of software then any other system existing on the market at present (perhaps apart from niche systems like GNU/Hurd and Haiku). Microsoft Windows systems are still dug in on the other pole with their restricted licensing allowing users to — at the most — conditionally run the systems (not even in every case) and their applications.

By friendliness we mean something more general that the ease-of-use and availability of applications! It is about not constraining the users but granting them full freedom to exert their rights to use software in their own way. It is about a freedom of choice — the possibility to use any window manager, any file system, and any package system the user wants to choose, about the possibility to work in a console or GUI environment, locally or remotely. It’s all about the security schema, well designed and overt, instead of “security by obscurity” model which has been compromised over and over. Do the users need such freedom? Certainly 90% PC users will do well without it. For the remaining 10%, including me, the GNU/Linux systems have appeared to be the only solution.

Proof-read by chaddy