Linux

This article is about Linux-based operating systems, GNU/Linux, and related topics. See Linux kernel for more extensive coverage of the Linux kernel itself, from which these systems derive their names.

Image:linux.jpg

Linux means, strictly, the Linux kernel part of a family of computer operating systems, but the name is also used to describe the entire free Unix-like operating system (also called GNU/Linux) that is formed by combining the Linux kernel with the GNU libraries and tools. The first version of the Linux kernel was written by Linus Torvalds, and released in 1991 combined with essential components from the GNU project (begun in 1983 by Richard Stallman). The kernel is not officially affiliated with GNU, but is independently developed and distributed under the terms of the GNU General Public License (GPL).

The term "Linux" is now even applied to whole Linux distributions, which typically bundle large quantities of software, from

web servers like Apache to
graphical environments like GNOME to
office suites like OpenOffice.org, all built on the
X Windows graphics subsystem,

with the core operating system.

Since its first release, the Linux operating system has experienced rapid growth in popularity, overtaking proprietary versions of Unix and even beginning to challenge the dominance of Microsoft Windows. It has been deployed in applications ranging from personal computers to supercomputers to embedded devicess such as mobile phones, supporting a remarkable variety of computer hardware.

The 'official' logo and mascot of Linux is Tux the penguin. There are many local Linux User Groups, worldwide; they serve as forums for users of Linux-based operating systems. The Linux trademark (SN: 1916230) is owned by Linus Torvalds, and applies -- officially -- to "Computer operating system software to facilitate computer use and operation."

Table of contents

1 Linux distributions
2 Applications of Linux-based operating systems
3 The scale of the Linux development effort
4 "GNU/Linux"
5 History
6 Litigation
7 Usability / Market Share

7.1 Configuration

8 References
9 See also
10 External links

10.2 General information
10.3 Newbie and user-friendliness in Linux
10.4 Linux distributions
10.5 Reference
10.6 Tutorials
10.7 Free online training
10.8 Online publications
10.9 Printed publications
10.10 Linux and GNU
10.11 Linux kernel
10.12 Linux hardware
10.13 Scholarship and grant support
10.14 Other

Linux distributions

There are many Linux distributions (distros), assembled by individuals, corporations, and other organizations, and each may include any number of additional system software and application programs, as well as a program to install the whole system on a new computer. The core of each distribution includes the Linux kernel, but also various software packages from the GNU project, and elsewhere, including a shell and utilities such as libraries, compilers and editors. Because the GNU facilities (without which the system would not resemble Unix from a user perspective) stem from a longstanding free operating-system project that pre-dates the Linux kernel itself, Richard Stallman of GNU/FSF asks that the combined system (regardless of distribution) be referred to as GNU/Linux. Some people do; most simply call the system "Linux."

Most systems also include non-GNU tools and utilities; examples include tools and utilities from the BSD and its descendents and the XFree86 open source implementation of the X Window System. The latter provides the most common foundation for a GUI interface on Linux systems.

Applications of Linux-based operating systems

Linux users, who traditionally had to install and configure their own system, have been thus more technologically oriented than Microsoft Windows and Mac OS users, often revelling in the tag of "hacker" or "geek." This stereotype, though often self-imposed, has been undermined in recent years by the increasing 'user-friendliness' of many Linux distributions. Some of them are at least as 'easy to install' as recent versions of Windows.

And with the adoption of Linux by several large PC manufacturers, computers with Linux distributions pre-installed on the first or second hard disk have become available. Linux has begun to make (slow) inroads into the high volume 'desktop' market. It has already made very considerable progress in the server and special purpose (eg, image rendering, Web services) markets.

Alternatively, some distributions (such as Knoppix, Gnoppix—the Gnobian version—and Gentoo) allow Linux to be booted directly from a CD (sometimes called a LiveCD), without modifying a hard drive. One can download CD ISO images for these and other distributions from the Internet, burn it to a CD, and execute Linux from the CD.

Still other possibilities include booting over a network, or (for a minimal system) from a few floppy disks, or network card NetBoot flash drivers (see Isolinux).

Linux is also the cornerstone of the LAMP server-software combination that has achieved widespread popularity among web developers.

Linux is also being used as an embedded operating system. The low cost of Linux makes it possible to use it in devices such as the Simputer, a low-cost computer aimed especially at low-income populations in developing nations.

With desktop environments such as KDE and GNOME, Linux offers a graphical user interface much more like the Xerox Star and Alto, or Mac OS or Windows, than is the traditional Unix command line interface. Many no-cost (though not always open source/free) software packages available for Linux on various hardware offer most (or arguably all) the functionality of common commercial programs available on the other desktop operating systems.

The scale of the Linux development effort

One study of the Red Hat Linux 7.1 distribution found that this particular distribution contained 30 million physical source lines of code (SLOC). Using the COCOMO cost model, it could be estimated that this distribution required about 8,000 person-years of development time. Had it been developed by conventional proprietary means, it would have cost over $1.08 billion (1,000 million) to develop in the U.S. (in year 2000 dollars).

The majority of its code (71%) was in C, but many other languages were used including C++, shell scripts, Lisp, assembly language, Perl, Fortran, and Python.

Slightly over half of all its code (counting by line) was licensed under the GPL.

The Linux kernel contained 2.4 million lines of code, or 8% of the total, showing that the vast majority of a Linux operating system is not contained in the Linux kernel.

"GNU/Linux"

GNU/Linux is the term promoted by the GNU/FSF project and its supporters, most prominently by its founder and main activist Richard Stallman, to refer to the Linux operating system. Their basic argument is that GNU was an ongoing project to develop a free operating system that pre-dated the Linux kernel by almost a decade, and Torvalds' kernel was only the final missing piece completing that project. GNU's own kernel, the HURD, had been long delayed. Besides failing to credit the GNU project, some additionally argue that naming the whole system after the kernel alone encourages substantial technical confusion among the public. Nevertheless, the historical sequence of events and other factors have resulted in most people continuing to call the whole system Linux.

A popular misconception is that the FSF argues for GNU/Linux purely on the basis of the large number of GNU tools used in Linux. Actually, Stallman writes (in Linux and the GNU Project):

So if you were going to pick a name for the system based on who wrote the programs in the system, the most appropriate single choice would be GNU. But we don't think that is the right way to consider the question. The GNU Project was not, is not, a project to develop specific software packages. [...] Many people have made major contributions to the free software in the system, and they all deserve credit. But the reason it is an integrated system—and not just a collection of useful programs—is because the GNU Project set out to make it one. We made a list of the programs needed to make a complete free system, and we systematically wrote, or found people to write, everything on the list.

The name "GNU/Linux" was first used by Debian in 1994 as the name of their OS distribution based on the Linux kernel and GNU programs. (In 1992, the Yggdrasil distribution was called Linux/GNU/X). In GNU's 1994-June Bulletin, Linux is referred to as a "free UNIX clone" (with many GNU utilities and libraries). This is of course inaccurate, since the Linux kernel is not a copy of UNIX, but a work-alike. In the 1995-January edition, references to Linux were changed to "GNU/Linux". In May of 1996, Stallman released Emacs 19.31, changing the system target "Linux" to "Lignux". He argued that to give rightful credit to GNU, it is proper to use the terms "Linux-based GNU system", "GNU/Linux system", or "Lignux" to refer to the combination of the Linux kernel and the GNU system. Stallman later stopped using the term "Lignux" and used "GNU/Linux" exclusively.

Requests to call the system "GNU/Linux" have met with mixed success at best. Only a few distributions have followed the lead of Debian in calling their systems "GNU/Linux". The corporate world, including most media outlets, mostly doesn't. Amongst the users and developers in the free software and open source movements, some have use GNU/Linux; many don't. There is even some explicit opposition.

Some consider the term "operating system" to refer to only the kernel, with the rest being simply 'utilities' (regardless of the practical necessity that they be available and despite the large volume of code in such utilities) or 'applications'. In this sense, the operating system is called Linux, and a Linux distribution is based on Linux with the addition of the GNU tools and other software. On the other hand, both the name GNU and the name Linux are intentionally related to the name Unix, and 'Unix' has always notionally included the C library and userland tools as well as the kernel. Kernel-author Torvalds wrote, in the 1991 license statement for version 0.11 of Linux (which was not under the GPL until version 0.12):

Sadly, a kernel by itself gets you nowhere. To get a working system you need a shell, compilers, a library etc. These are separate parts and may be under a stricter (or even looser) copyright. Most of the tools used with linux are GNU software and are under the GNU copyleft. These tools aren't in the distribution - ask me (or GNU) for more info.

Some of the reasons people refer to the system as "Linux" instead of "GNU/Linux" are because the former is shorter and easier to say, because Torvalds has called the combined system Linux since its 1991 release (at the suggestion of a friend), and because Stallman only began asking people to call the system "GNU/Linux" in the mid 1990s after the system had become popular and the name established. And, of course, since "Linux" is the most widespread name, many people simply copy the usage without learning the history or debate behind it.

One practical problem with the use of the word "Linux" to refer to both the kernel, as well as entire distributions, is that it has often led to confusion in the popular media (and hence among the general public). Thus, media sources frequently make erroneous statements such as claims that

the entire Linux operating system (in the popular sense) was written from scratch by Torvalds in 1991 (only the first version of the kernel was; all versions since have been by Linus and hundreds of others),
that Torvalds directs the development of other components such as graphical interfaces or the file systems or any of the GNU tools (he does not), or
that new releases of the kernel or the entire GNU/Linux system involve a similar degree of user-visible change as do new versions of proprietary operating systems such as Windows (changes in GNU/Linux are far more backward compatible than is typical for Windows).

A few use the Linux kernel but with few or no components of GNU (because they use BSD or other replacements, or simply run the kernel nearly 'bare'). These are mainly small, embedded, systems such as dedicated firewall products or 'appliances'. Everyone, including the FSF, agrees that GNU/Linux is not an appropriate name in that case.

History

The history of Linux is heavily tied to that of the GNU project, a prominent free-software project led by Stallman. The GNU project was begun in 1983 to develop a complete Unix-like operating system composed entirely of free software. By 1991, when the first version of the Linux kernel was written, the GNU project had produced nearly all of the components of this system, including a shell, a C library, and a C compiler. There was no kernel for the system, however, because the design of the GNU kernel (called the Hurd) was so ambitious that it proved unexpectedly difficult.

The Linux kernel was initially written as a hobby by a Finnish university student, Linus Torvalds, who was attending the University of Helsinki, as a free and modifiable Minix-like kernel (Minix was and is a Unix-like teaching project from Andrew Tannenbaum, deliberately made simple enough for teaching and without a 'production perspective'). Subsequently, thousands of volunteer computer designers/programmers throughout the world have participated in the project. (See the first Linux announcement, archived on Google.) Torvalds and other early Linux developers adopted the GNU components to work with the Linux kernel, creating a completely functional operating system.

Linux, thus, filled the last major gap in the GNU operating system. Although the Linux kernel is now licensed under the GNU General Public License, it is not part of the GNU project. The GNU project has a separate kernel development project, the HURD, whose completion is still eagerly awaited in some circles.

Litigation

The Linux kernel now presents a major competitive threat to the manufacturers of proprietary operating systems. There have been attempts to spread Fear/Uncertainty/Doubt about Linux -- widely thought to be in all cases, and shown to have been in some cases (including bogus 'research' reports), to have been deliberate actions by commercial competitors.

And there is a possibly major legal problem with Linux. Early in 2003, SCO Group filed a lawsuit against IBM claiming that IBM had included portions of SCO's intellectual property into the Linux kernel in violation of IBM's license to use UNIX, now claimed to be held by SCO. Additionally, SCO reportedly sent letters to many companies warning them that their use of Linux without a license from SCO may be actionable. Red Hat has now filed a lawsuit against SCO, seeking to stop SCO's intellectual property claims, and seeking legal redress for the harm done to Red Hat by such claims. IBM has counterclaimed as well in the case filed by SCO. On November 13, 2003 SCO Group served subpoenas on Stallman and Linus Torvalds. Discovery is ongoing and, to the extent that SCO's specific claims are known (not at all for many months; not very much and that only by accident it seems in more recent months), there is near universal industry opinion -- outside of SCO -- that they are groundless. In particular, Linus commented on one of SCO's claims by noting that the source code cited was in fact written by him long before SCO acquired any color of claim to any UNIX source code and furthermore was rather embarrassingly defective; he had corrected the design mistake many years ago. See SCO v. IBM Linux lawsuit for details. SCO has announced that IBM's license to use the UNIX code has been cancelled, and more recently it has done the same with respect to SGI for related reasons.
SCO's claims derive from its reading of the history of UNIX. UNIX was developed at Bell Laboratories and was, at that time without question, the property of AT&T. AT&T licensed that UNIX, non exclusively, to many other companies (eg, Mt. Xinu, Fortune Computers, Dual Computers, SGI, HP, DEC, IBM, Microsoft, Sun Microsystems, University of Califonia at Berkeley, ...). Until the System III release (?), those licenses included access to the source code. Many of these firms sublicensed their version of UNIX, with or without access to the source code, under various names to various others. AT&T made several official attempts to merge the various strains of UNIX into a single release; the last version it was responsible for, 'UNIX System V Release 4' (SVR4), included material from among others, the CSRG at Berkeley, Microsoft, Sun Microsystems, ... AT&T even sued the UC Regents/CSRG and a spinoff company, BSDI, over their use of the source code; the issue appears to have been copyright on the code used in assorted BSD releases. The settlement of that suit (in 1994?) confirmed that BSDI and the Regents had the right (via license, of course) to use and release the source code of their UNIX version essentially as they pleased. Later still, AT&T transferred the rights to Unix from its Western Electric division to the Unix System Labs division (after System III), and then sold the rights to UNIX to Novell. Novell sold some (all?) of its rights in Unix to SCO (the Santa Cruz Operation); the name UNIX eventually ended up with one and then another industry standards group -- currently the Open Group. Microsoft also sold its version, Xenix, to SCO, completely returning that branch. But SCO went bankrupt and its assets were acquired by Caldera, a spun off division of Novell.
Caldera had previously developed and sold (under the GPL) a Linux distribution of their own (OpenLinux) and joined with other Linux distributors (eg, TurboLinux and SuSE) to form UnitedLinux, a company devoted to producing an industry standard Linux distribution. As well, Caldera had acquired the rights to DR-DOS (a widely respected MS-DOS work-alike developed originally by Digital Research and thereafter also acquired by Novell), and sued Microsoft for alleged illegal interference with DR-DOS' sales. That suit has been settled out of court for a reputedly large amount.
Caldera has since changed their name to SCO Group (NB, this is a different, though a successor, corporate entity than the original SCO -- the Santa Cruz Organization), and decided to enforce their rights to the copyrighted UNIX source code not long thereafter. Hence the letters to assorted users demanding licensing and license fees, and the suits against some of those not complying. Perhaps significantly, Microsoft is said to have complied with the first licensure demand. In 2003, Novell registered (for copyright) major parts of the UNIX source code with the USPTO, and announced that their agreement with SCO did not include the copyright on UNIX' source code.
Since the UNIX source code base has had inclusions from many (hundreds if not thousands) of sources at various times over approximately 30 years, not least of them the University of California at Berkeley, and since few of those additions are (or ever were) copyrighted in the same manner and by the same owners as the original UNIX source code from AT&T, the copyright status of any specific part of the UNIX source code is, at best, hard to discern, and at worst open to substantial dispute. The disagreement between SCO and Novell on this point would appear to be more fundamental than a disagreement about which line of source code came from whom and under what conditions. SCO has since sued Novell over this.

Linux users, or prospective users, are left in a state of uncertainty. Most industry technical observers, especially those with some intimate knowledge of the source code involved, are nearly unanimous in concluding that SCO's claims are shaky at best. Linux' development has been done is public view for more than a decade; who contributed which lines and when are easily determinable. Where, however, these or those particular lines came from may be in question. Algorithms, designs, program fragments, techniques, and ideas were frequently shared among UNIX (and UNIX derivative) developers / designers and programmers for a very long time, going back well before the Linux project existed; some of this was official, some of it unofficial, some of it substantial, some trivial. The extent, if any, to which any of those transfers were documented at the time and which if any of them can be credibly traced now is unknown, and perhaps unknowable, to anyone. And some of the code involved was in the public domain from the moment of creation having been developed under US Goverment research grants which for all(?) of this period included a public domain release requirement (eg, much of the work done at UC Berkeley). This is, however, an engineers' perspective and may not be legally compelling.
Commercially, Caldera had released (via the Internet) 'Ancient Unix' (meaning apparently editions 1-7, System III, and 32V) to all those interested under, it appears, more than one license, one of which was an educational/research license. SCO (the original entity) also permitted the publication of John Lions' commented edition of the UNIX 6th edition source code in the mid-90s(?); AT&T had insisted that the book be withdrawn when it was originally published some decades earlier. These, and other, copies of assorted releases of assorted versions of UNIX source code make claims of trade secret violations for code contained in any of these releases difficult to sustain at present. Both SCO (pre-bankruptcy) and Caldera, prior to acquiring SCO assets, have both released Linux distributions, both under the GPL. And Caldera was a founding member of United Linux, the venture charged with creating a common, GPL'd, Linux source code base. But this is a kind of commercial history, neither an engineering issue, nor a legal perspective. It may not be legally compelling either.

As the problems which have arisen are in some aspects legal ones, and as some of them have already been taken to court, the views of lawyers and other legal types are also relevant. On the one hand, some lawyers with knowledge of the case suggest that SCO is likely to prevail at least in part; one such is an SCO outside counsel, David Boies. But some lawyers with good knowledge of the issues think the first group must be talking about something else altogether; Professor Eben Moglen of Columbia Law School is an example. Non-lawyers using Linux may take some comfort in the fact that the situation is quite confused and SCO (nee Caldera)'s straightforward reading of it may not prevail. At the least the complex copyright history, SCO/Caldera use of the GPL for their own Linux releases, Caldera (now SCO)'s public release of much historical UNIX source code, and finally Caldera (now SCO)'s participation in United Linux have made the SCO case difficult. One group of lawyers (or another) may finally convince a judge otherwise (or not); perhaps after a few appeals to higher courts.
There is some prospect that requests for summary judgement by IBM and Red Hat may stop (all or part of) this litigation early in 2004. On the other hand, perhaps not. SCO may find court support for their claims or, in the alternative, may be found liable to assorted counter-claimers for various sums from large to token.

Usability / Market Share

Linux, once viewed as an operating system only computer geeks could appreciate, is today a much more user-friendly system that companies, public and private organizations, and consumers can use as easily as competitive proprietary operating systems (like Microsoft's Windows XP). That's the core finding of a study published by Relevantive, a Berlin-based company specializing in consulting companies on the usability of software and Web services.

On the other hand, Linux has been criticized for having inconsistent and unpredictable development schedules, thus making consumers less confortable with Linux than they might be with another operating system (Marcinkowski, 2003). The fragmentation of Linux distributions creates too many choices for consumers, further confusing them. In practice, this is not necessarily a disadvantage, as the dominant supplier (Microsoft) has hardly been a paragon of reliable release dates or backwards compatibility.

In 2004, however, the question of when Linux would (or had already) experience(d) significant market share on desktops, was debated. During the Linux.Conf.au conference at the University of Adelaide in January 2004 (Linux Australia, n.d.), Torvalds made a statement of his own opinion on the issue (as cited in Gedda, 2004): "This year there will be a lot of desktop users..." On the other hand, later that month he said, "I mean it's going to take literally five to ten years before 'normal users' start seeing [the] Linux desktop" (as cited in Mackenzie, 2004). As with most other things Linux, opinions differ and everyone can contribute.

For people who have always used Microsoft Windows (or the MacOS especially before MacOS X), using Linux can present problems because many things do not work 'exactly' the same. Some things will have to be relearned. It is also usually easier to find local tech-support people who understand some of the technical aspects of Windows (or the MacOS) than it is to find similar folk for Linux. Most non-technical users don't particularly care which operating system they use, so long as it has all of the programs they need and they can sucessfully use it. Unless they find major problems in other operating systems, it's widely believed they're unlikely to switch to Linux. On the other hand, Linux, once installed successfully, is far less likely to crash, lock up, be attacked by a virus or be taken over by an attack 'from the network' than is any version to date of Windows.

At this writing, on personal desktops, Linux is still used mainly by people who 'know more' about computers. As long as computer manufacturers and distributors supply Windows as the default operating system, this is not likely to change rapidly. Microsoft has shown little inclination to modify the terms of its Windows' licensing to computer manufacturers.

Configuration

Linuxconf is a sophisticated administration system for the Linux operating system, written in the C++ and may be operated with your favorite Web browser. Linuxconf works for a.out and ELF systems; it installs currently directly on the majority of Linux distros, including Debian.

gnome-linuxconf is a GTK-based graphical frontend to Linuxconf.