Open Source, Free Software and Other Beasts

The free software movement (before it was called this way) started organically from individuals who distributed code they wrote under the Public Domain or what would now be considered open source or semi-open source licences.

AT&T UNIX that started at 1969 was the first showcase for this movement. Several Bell Labs Engineers led by Ken Thompson developed UNIX for their own use, and out of legal restrictions AT&T faced, decided to distribute it to academic organizations and other organizations free-of-charge with the source included. (that licence did not qualify as open-source but it was pretty close). UNIX eventually sported the C programming language, which enabled writing code that would run on many platforms easier, and the UNIX sources included a C compiler that was itself written in C. Around the early 70’s the only computers capable of running UNIX were main-frames and the so-called “mini-computers” so there initially weren’t as many installations as only large organizations could support buying computers to deploy UNIX on.

That changed as integrated circuits, and computers became cheaper and more powerful. Very soon, cheap UNIX-based servers and workstations became commonplace and the number of UNIX installations exploded. ^[1]

Nadav Har’El has prepared a coverage of the BSDs and early AT&T UNIX history.

The University of California at Berkeley (a.k.a UCB) forked its own version of AT&T UNIX and started re-writing parts of the code, and incorporating many changes of its own. The parts that the Berkeley developers wrote on their own had originally been licensed to UCB and kept as non-FOSS (= “free and open source software”) “All Rights Reserved” licence. The BSD system became very popular (perhaps even more than the AT&T one).

When Arpanet, the predecessor to the Internet was disbanded due to inadequacy, the Internet converted to running on top of 32-bit UNIX boxes such as the VAX architecture by Digital Equipment Corporation (now part of Hewlett-Packard). This caused a merging of the UNIX culture with the Arpanet enthusiasts who exchanged code on the Arpanet, and UNIX programmers started sharing code for various components and add-ons of UNIX on the Internet.

After a while, the legal restrictions posed on AT&T subsided, and it started to “smell money” and believe it can do better selling UNIX commercially. It created the AT&T System V system, touted it was better than AT&T UNIX and the BSDs, and sold it to vendors. System V was sold under a very restrictive licence, that forced them to hold the source code for themselves. Even cooperation between two different vendors was not allowed.

Gradually, vendors licensed the System V source code and ported it to their own architectures. This caused an explosion of proprietary UNIX systems. Sun Microsystems and other vendors took the BSD source code, diverged from it and distributed it without full access to the code to all customers. A similar thing happened with other software distributed under similar licences.

To answer this threat, a new phenomenon sprang into existence: the “free software” movement, the GNU project and the copyleft licences, all led by one dynamic personality: Richard M. Stallman.

Richard Stallman (aka RMS) published the GNU Manifesto in 1984, which coined the term “free software”, and explained the rationale behind it. The Manifesto was also a creed for the the GNU project which aimed to be a complete UNIX-compatible replacement for UNIX systems, while being completely original work. The software of the GNU project was released as free software, under the terms of the GNU General Public License (or GPL for short).

Gradually, the GNU project created more and more C code to replace the UNIX and BSD utilities. It was already installable and usable on various flavours of UNIX, and became a fully independent system once the Linux kernel was written.

The GPL licence is a free software licence that has many fine points. The most important concepts in it are:

The incentive to restrict a software this way rather than following the more traditional public domain or public-domain-like licences (as used by such software as the TeX typesetting system), was to make sure that the core GNU system would always remain free as well.

Encouraged by Stallman’s growing momentum behind the Free Software Foundation and the GNU project, Berkeley University changed the licence of the parts that they have originated, to a a free software licence which is now called “The Original BSD License”, which qualified as free software, but as opposed to the GPL was public-domain-like. ^[3] To add to this effort, some UCB students decided to rewrite the remaining parts that were licensed to AT&T under the BSD licence. This task was eventually completed that resulted in a BSD system that was entirely under the BSD licence.

However, AT&T did not stand by, and pressed charges against UCB and some other organisations, for claiming they actually own parts of the BSD operating system. This brought uncertainty into the BSD world, which would not be resolved until the 1990s, when the law-suit was decided mostly in favour of UCB. As a result of this uncertainty, the status of some spin-offs of BSD (such as 386BSD, and its derived operating systems such as FreeBSD or NetBSD) was in a legal limbo.

In 1992, Linus Torvalds, then a student at Helsinki University, began writing the “Linux” kernel - a 32-bit kernel for UNIX-like operating systems. The kernel development advanced rapidly and was released under the GPL licence starting from an early stage. To complete the system and make it into a usable UNIX system, the Linux developers used various existing user-land utilities and libraries from the GNU project and other sources (such as the X-Windows system), and wrote a few user-land utilities from scratch.

From an early stage, this entire system was dubbed “Linux” as well. Richard Stallman instead has advocated the name “GNU/Linux” (pronounced “ggnoo-Linux”) which acknowledges the fact that the GNU project contributed the lion’s share of the system (including some pre-requisites of the Linux kernel itself). Most people haven’t consistently followed this piece of advice.

The importance of the Linux kernel was that it was the last brick in materialising a fully GNU system. Since GNU tools tend to be more complete, feature-rich and generally superior to tools of other systems, this has made Linux one of the most powerful UNIX systems available. Nowadays, most UNIX servers out there, many UNIX workstations, (and many embedded devices) run the GNU/Linux system. Linux was, thus, the spearhead that guided the acceptance of free software into the mainstream.

Debian GNU/Linux was a Linux distribution that was eventually endorsed by the GNU project. One of the aspects that made it unique was the fact it distinguished between “free” and “non-free” packages as far as the user is concerned. The guidelines for determining which software is “free” in the Debian sense were phrased by Bruce Perens.

Note that they deviate from the Free Software Definition (which was only published later on) and include some licences that are not free. I.e: “Debian Free” is a superset of free software according to the Stallman definition.

This fact is important because later on, the Debian Free Software Guidelines formed the basis for the open-source definition.

Eric Steven Raymond (now also known as ESR) wrote an essay titled “The Cathedral and the Bazaar”, and presented it to the Linux Kongress at 21 May 1997. This contrasted the Bazaar way of managing a software project to the old “Cathedral” way, that was used by almost all non-free projects and (until that point at least) by most free ones.

“Bazaar” projects are characterised by frequent and incremental release schedules, treating the users as co-developers, and generally getting a lot of peer review, ideas, input and cooperation. Despite a common misconception, the core group of the project contributors still usually remains relatively small except for some of the larger projects.

The article is considered one of the seminal works on free software, and was followed by other works in what is collectively known as the “Cathedral and the Bazaar” (or CatB for short) series. It has made Eric Raymond a famous person, at least among the community of free software hackers.

In February 3, 1998, in Palo Alto California, a brainstorming session which Raymond attended, coined the term “open source” as an alternative for “free software”. Their incentive was that when talking to a businessman, either free software will be understood as gratis software, or it will be associated with the relatively anti-Capitalistic views held by Richard Stallman (who claims non-free software is immoral). They decided that the term “open source” would be a better candidate for acceptance in the corporate world.

Consult the opensource.org history document for further coverage of the history of the term.

During the following week, Eric Raymond, and Bruce Perens launched the opensource.org web-site, and formed the Open source definition. This was based on the Debian Free Software Guidelines.

The term “open source” caught on. Very soon, Richard Stallman decided to reject it on the premise that the freedom of software is more important than the “openness” of its code. While he does not oppose the openness of the code, and acknowledges the fact that free software is open source as well, its freedom remained more important. For more about this stance, read the document “Free Software for Freedom” on the GNU web-site.

While some people have continuously stuck to the term “free software” and a few others converted to using “open source” entirely, most knowledgeable people don’t completely reject either term, and use each one whenever they see fit. Nevertheless, the term “open source” is more commonly used by both open source developers and even more so by non open source developers. See Eric Raymond’s “Terminology Wars” for more details.

Since 1997, Linux and other open-source systems have become more and more popular. Linux saw a lot of success in the server market, where cheap PCs that can be bought in stores can serve as an almost full replacement for more costly UNIX servers by installing Linux. Even if the latter are used, they very often run open source servers and other open source programs, utilities and frameworks.

Linux has become the number one choice for constructing clusters, a large set of computers that are networked together to form a fast computation system, with powers that rival or exceed super-computers. There are various kinds of clusters around. Some of them are performed at a relatively high level. Others, try to make the system believe it has as many processors as there are nodes.

Linux also had a lot of success in the embedded market, serving as the framework for creating software that is embedded in hardware.

The Internet boom not only made free software more essential for its operation , but also enabled more and more users and developers to share their code, get help and work together for advancing it.

At the moment, Linux had a much more limited success as a choice for a desktop system. While it used to be the only operating system that was gaining market share (at least until the renaissance of Apple with its Mac OS X), it still has a very low one, in comparison to Microsoft solutions. Many projects started to supply users with desktop and GUI environments and applications. Some of them are very mature, usable and successful. Only time can tell if and when Linux becomes the default solution for the desktop.

Apple’s Mac OS X was released and is based on Darwin, which is an open-source BSD-derived system. Mac OS X can run UNIX applications natively, and supports the X-Windows system, which is the de-facto GUI framework for UNIXes (including Linux). It is therefore a popular UNIX choice for PowerMac computers (and more recently for Macintosh computer based on Intel-based chips), albeit not the only one since Linux, and various open-source BSD clones and other UNIXes can run there as well.

The recent recession in the information technology market, did not seem to slow down the development of open source software. Freecode (formerly Freshmeat) is still busy as ever with releases of new software, and since the recession started, many important new releases were done for a lot of major applications and even many more less important ones.

While open source software has existed for DOS and Microsoft Windows practically since the beginning, and some of it was relatively popular among people, most of the software available for these platforms has been non-open-source binary-only software, a lot of it from Microsoft.

This has started to change recently. The Firefox browser from the home of the Mozilla Foundation (and now also the Mozilla Corporation), is an open-source, modern and sophisticated browser, that has been virally publicised by various means such as the various “Spread Firefox” campaigns. It has become popular and as of July 2006 has passed the 10% usage in web site hits according to some firms, and in some countries much more so. It is still gaining some market share, even if its growth has declined somewhat.

Other cross-platform open source software includes OpenOffice.org, a powerful and usable office productivity software for Windows, Linux and other platforms, the GIMP (GNU Image Manipulation Program), a sophisticated raster image editing program, and Inkscape, a vector editing program, many open-source music and media players such as VLC, and also most Peer-to-Peer networking clients. These have probably seen less popularity than Firefox, but are still providing cheap, open, modifiable alternatives to traditional binary-only software.

In 2003, a study was published that estimated that by 2004, more software developers will write software for Linux than for Windows. While it definitely does not mean that more people will use Linux at home, it is still a good indication for its general mainstream acceptance and usefulness.

Another important recent trend was the rise of open content. The first edition of this article, included a small section about “open content”, where I concluded by saying that “Only time can tell whether other elements of open source besides its freely distributable nature will have an impact in other areas of creative arts besides software.”. Now, about 3 years later, I can say that by all means open content has already proven to be a great success.

Among the landmarks of open or semi-open content are:

Proprietary software is legal, but illegitimate and immoral. Manufacturing and using proprietary software causes a lot of unhappy social and psychological side-effects. The knowledge that a software cannot be shared causes people to become reluctant to sharing, which is a natural and good part of living in a human society. The inability of people to modify software for their own needs makes them feel helpless, and at the mercy of external software.

Free software, on the other hand, is the natural conclusion derived from the basic facts of information, computing and software, and is highly moral. People, companies and other organizations can modify it, customise it and distribute it for their own use should the need arise, and so it actually benefits them.

Proprietary software is not illegitimate, just problematic from the economic sense. Open Source software gives many advantages to the end-users and is a generally a good thing. Copyleft licences may be important in making sure certain software is not abused. (Note, however that even Raymond recently voiced his opinion that the GPL is no longer a wise choice as licence for new code). It is not immoral to use proprietary software, it’s just risky. Using or producing software that is not 100% open-source but pretty close, can be a good idea, depending on its licence and the general attitude of its developers.

This approach basically says this:

“I like free software because I can get a lot of useful software without charge. I may like contributing to free software because it helps other people, makes me happy, and may indirectly benefit me technically or financially. But proprietary software is perfectly valid as well, if it’s done right, and I may choose to use it or contribute to it.

In short: write code, use whatever tool you wish, and be happy.”

The most prominent figure who holds this view is Linus Torvalds, but there are many others, some of them quite prominent. Such figures, however, tend to be less loud than the “religious” advocates of the other two views, and thus it may seem that they are at a minority. Part of the reason is that many of them inherently tend to value productive coding and decision-making over advocacy.

Note: I have prepared a longer (and unofficial) manifesto for this view which you may wish to consult for further information.

While some figures out there prominently stick to either ideology, most people hold a mixture of the three (or more?) approaches, or are just happy using free software or contributing to it, without thinking too much about its philosophy.

The terms themselves are used interchangeably by many people. “Open source” has become more common, partly because free software can mean software that is given free of charge. (the standard “free as in free speech” or “free as in free beer” distinction). Moreover, both the Free Software Foundation, and the people associated with the Open Source Institute are on friendly terms with each other and answer questions, give feedback, and accept contributions, from each other or from people that do not belong to either camp.

Like I said earlier, the fact that some licences would qualify as open-source and not as free software is usually a negligible fact. While some esoteric software has been released under custom licences that are open-source while not being free software, most of the important software applications out there (and most applications generally started by individuals) is free as well. ^[4]

Making a program free is not necessarily enough to make it compatible with the GNU GPL licence. The GPL makes some restrictions regarding which licences it can link against, and some otherwise free software is not compatible with them. Examples for incompatible licences are the Mozilla Public License, the Qt Public License, and even the original BSD licence. It is advisable that, whenever possible, a developer or vendor should choose a licence that is compatible with the GPL, because otherwise there may be problems integrating his code with GPLed one or using both a GPL and a non-GPL compatible library. (I am not a lawyer, so I cannot conclusively say when it is legal or not).

Mozilla is an example for a large project that started out with its custom (albeit now relatively common), non-GPL compatible licence, and recently adopted a triple licence of the Mozilla Public License, the GNU General Public License, and the GNU Lesser General Public License in order to make it compatible with the GPL and to standardise its integrability. The Qt library whose commercial vendor and originator is Troll Tech Inc., also had adopted the GPL as well as its own Qt Public License, to relieve the various legal problems that KDE (a desktop system for UNIXes which is based on it) faced when using GPL code.

One relatively recent issue with the GNU General Public Licence had been the formation of Version 3 of both the GPL and the LGPL, which made programs that were only version 2 of the GPL (and not a later version) incompatible with those of the GPL version 3 or even the LGPL version 3. While some programs has been relicensed or sublicensed under version 3 of the GNU licences, a lot of software packages out there are stuck at being version 2 without an option for a later version, which make them mutually incompatible with the formerly licensed packages. Such FOSS packages include a large amount of the code of the Linux kernel, as well as Ghostscript and xpdf, which are commonly used for rendering the PostScript and PDF standards for ready-to-print-documents.

The copyleft definition

Copyleft means that a derived work of a copyleft software, that are not used for internal or personal use, must include the source code and released under the same terms of the original work. Copyleft is common in many licences including the GPL, the Lesser General Public License, the QPL, etc.

Many licences are not Copyleft - most notably the various BSD licences and MIT/X11 licences. Software released under such licences can be derived into a proprietary software product by a third party, and often have been.

Not any software application that is accompanied by its source is open source, albeit many people who are new to the term would be tempted to think that. It is possible to write non-open-source software while accompanying it with the source.

Examples for such cases are:

None of these packages qualify as free software, but they are all accompanied with the source. There are many others around. A quick search on the Freecode directory of UNIX software will find many such packages.

In order for a program to be open-source it needs to be free of various restrictions as specified in the open-source definition. To be free software as well, it must be also free of some other restrictions. ^[5]

I believe the term open-source is a bit dangerous in this regard. Then again, free software may not automatically be associated with freedom and liberty, so it isn’t perfect either. But I guess finding a description that accurately describes it in a short space is not very possible, so these terms will have to do.

This analogy is not new but very deceptive. First of all, there’s nothing anti-Capitalistic about sharing something voluntarily. While in a Capitalistic country, goods are generally sold and have to be paid for, people can voluntarily dedicate their time and money for any cause they wish, possibly altruistic. Communism in fact forces the sharing of all good, including physical ones that take time and money to manufacture each unit of.

Secondly, because manufacturing and distributing a unit of software costs practically nothing, it is not necessary that it will be sold. While the development cost can be very large, a developer of the software will not be encumbered by it being used by a million people instead of a thousand.

Furthermore, by making a software package open-source and keeping it so, it is possible to gain other economical and psychological advantages: you’ll make sure it is maintained, gain feedback and admiration of others, and may be able to eventually receive input and contributions from the outside. Distributing software as commercial proprietary packages does not automatically yield good advantages and it takes a lot of time and money to make it usable as well. Such a vendor is actually risking that his software will work at all, and not be out-competed by something better.

This analogy was rejected and treated in a semi-jokily manner by most people who did not oppose Capitalism as a whole. A true understanding of why open-source does not contradict the liberal ideals of Capitalism and Individualism originated from Eric Raymond’s “Homesteading the Noosphere” and “The Magic Cauldron”, which are a very good read anyway you look at it.

The facts themselves do not give way to it either. The free software movement is almost entirely limited to liberal Capitalistic countries, which are free enough to allow unrestricted programming and distribution of software to prosper. Furthermore, I cannot name a single prominent figure who is Socialist or Communist, or otherwise entirely anti-Capitalistic (albeit some may hold certain views of this kind). Lastly, open source or some open source software was recently endorsed by many IT and non-IT businesses who greatly benefit from it, including some vendors of proprietary software.

Finally, even assuming that free and open source software or partly-free-and-open cultural works (e.g: the various Creative Commons licences) have some elements of Communism or Socialism, does it really matter? After all, we enjoy the fruits of the many people who contributed to open source, the Wikipedias and other Wikimedia projects and various other free or mostly free works of culture and code, which would probably not have happened without them being free. As a result, even if open-source is indeed communistic or socialistic, it is nevertheless “good for the people” so to speak and should not be discouraged.

All of that put aside, it is clear that copyright laws do not and must not prevent people from applying not-fully-restrictive (nicknamed “All Rights Reserved”) licences to their original works, so eventually there will be enough people who out of either desiring to “help their neighbour” and/or out of motives that have more an aspect of a rational self-interest and without being forced to do so against their will, will be willing to release their works under open or semi-open licences.

It is unlikely that assuming Open Source becomes the dominant paradigm, it will imply that programmers will “starve to death”. As Eric Raymond notes in the Magic Cauldron, the vast majority of software applications out there are one without a sale value: be it software that large organization like banks, insurance companies or militaries use to power their critical systems, customizations, scripts or code used within smaller organizations (Microsoft Access customizations, spreadsheets’ formulae and macros, Perl or shell scripts), embedded software whose source code is not released to the public, software that powers web-sites and was not released to the public, etc.

The majority of programmers out there are employed for developing such code, whose codebase dwarfs that of the marketplace software that includes all commercial and open-source software put together.

In due times, packages are developed and become available that makes some tasks that were once hard to do internally almost straightforward to set up and run. Nevertheless, these packages still require a clueful person to operate, diagnose problems, communicate with the vendor or developer and manage the configuration. For example, a complete computer beginner will probably not know what to do with a spreadsheet program (such as Microsoft Excel) without thoroughly studying it. Afterwards, it becomes a very useful tool.

Even if programmers do become out of job as a result of free software, then it will not necessarily a bad thing. It means that it solved problems that otherwise required extra hands, and so those programmers can be allocated for something else entirely that is more productive.

One can very often hear many myths or generally accepted “truths” about open source software and Linux, some of which negative and other positive. Examples include:

1. Open source software is less secure than software whose source code is not revealed, because people can find bugs at it by looking at the code.

2. Open source software is more secure than closed-source software because more people can review the code and discover bugs in it.

3. Linux is harder to use than Windows.

4. Linux is not compatible with Windows.

5. Open source gives way to forking more easily.

And many others. The Linux Myth Dispeller attempted to answer some of these, with a focus on negative myths. Myth #1 is completely false as bugs can still be found by analysing the disassembly of the machine code. Also often such bugs are found by accident due to a certain valid use of the software) There were many closed-source packages out there in which many bugs have already been discovered. (like Microsoft Outlook, Microsoft IIS or Microsoft Internet Explorer). Some of these already became widely exploited a long time before a vendor patch was made available.

Myth #2 has a grain of truth in it. However, some open source packages nevertheless had very poor security records out of poor programming practices. Some closed-source offerings, on the other hand, have a very good security record. In most packages, security bugs occurred due to sloppy programming practice, or lack of auditing of the code. They can be mostly avoided whether or not the package’s source code is available to the public.

Myth #5 is not entirely true. While it is possible to fork a piece of open-source software, most packages have not been actually forked. Eric Raymond covers the customs that relate to forking a package in “Homesteading the Noosphere”, and Rick Moen explains why when major packages forked, it was not necessarily a bad thing in his “Fear of Forking essay”.

Moreover, many times proprietary software was forked as well. There are many flavours of System V UNIX out there, and there used to be many more. Microsoft released three different lines of Windows flavours with two or more simultaneously, and has many localised versions. (Which are many times incompatible with one another.)

Software patents are patents that cover algorithms used inside a computer program, and prevent a competitor from implementing it as well. Software patents can often be very generic, can be trivial to think about, can have some prior art upon acceptance, and their cost of issuing and cost of patent litigation are beyond the reach of most small-time open-source developers.

There have been many cases of patent litigation in the past by large companies, and there are some companies which consist entirely of lawyers who get hold of some patents and sue large companies for infringement, hoping to make some money. The English Wikipedia has coverage of the software patent debate

Some recent anti-copyright infringement laws such as the notorious Digital Millenium Copyright Act (or “DMCA” for short.) or the more recent ACTA, can be used to prohibit the distribution of some open source software, such as those that can be used to break the copyright-protection measures of some content providers.

The Electronic Frontier Foundation has published a document titled “Unintended Consequences: Twelve Years under the DMCA”, which contains a list of “cases where the anti-circumvention provisions of the DMCA have been invoked not against pirates, but against consumers, scientists, and legitimate competitors”.

Many source code licences were approved as free software or open source by either the Free Software Foundation or the Open Source Initiative. The problem is that if one encounters a software package that is distributed under one of these licences, they will have to become aware of this licence’s peculiar restrictions. Furthermore, it will be difficult to tell whether it will be compatible with code under a different licence.

As a result, there’s the problem of licence proliferation.

Some projects done by companies, or other individuals, require copyright assignment on the part of their contributors, so the copyrights will remain under the same copyright owners, who may then be able to relicense their code under different terms. Many people have been reluctant to contribute to projects that require such copyright assignment, because they would like to keep their own copyrights. As a result, this undermines the Bazaar model of development and the atmosphere of community in general.

Copyright assignment is a complete non-issue with projects licensed under permissive licences (a.k.a BSD-styled ones), whose licences allow the copyright owners, or every other party, to sublicense such a project.

There is a risk that open-source packages will become unmaintained, or spin-off a non-open-source version that will be more actively maintained. This is often seen as the risk of BSD-style licensed software, but it is not completely absent with GPLed one (see the story of the Nessus Vulnerability scanner). Even if the licence is a copyleft one, then there’s a risk of the originators of the programs stopping to update them, and no one stepping up to maintain them instead.

This problem is not specific to the FOSS world, because proprietary software also has become under-maintained or discontinued, but naturally the problem still exists.

Joel on Software describes a situation of eliminating sesame seeds:

Such a situation may occur with open-source software or free content resources, where people gradually eliminate features or text, or introduce more bugs until the software is too unusable. A prime example for such case is the so-called ”Deletionism” in Wikipedia and other collaborative online wikis, where people gradually remove pages or parts thereof that they do not like.

Another bad aspect of that is driving away the contributors who donated these features, who feel demotivated by such an attitude.

Many people in the open-source community are known for their hostility towards people whom they don’t like. I voiced some criticism of the “usability” of the Perl Online World for Newcomers, and there are similar sentiments about problems in treating people in the “HOWTO Encourage Women in Linux” document. I’ve also witnessed how one very important open-source project, which I was involved in, stagnated, because its developers were incredibly rude on their on-line forums and scared away most potential contributors.

Such hostility may be detrimental to a project’s success or the success of the open-source community in general.

With some grain of salt, I refer you to the other essays I have written about free and open source software (FOSS). Especially of interest is my essay “How to start contributing to or using Open Source Software”

Shlomi Fish, http://www.shlomifish.org/

Thanks to Chen Shapira for taking the time to read several early drafts of this document and provide useful comments. Thanks to Nadav Har’El for summarising the early history of UNIX and the BSDs for me. Thanks to Richard Stallman for some factual corrections on earlier versions of this document.

This document is copyrighted by Shlomi Fish under the Creative Commons Attribution Unported Licence version 3.0, or, at your option, any greater version of this licence.