Table of Contents
There are many distributions that decided to fork from a certain state of Debian. This is perfectly all right because Debian is completely free and everybody is allowed to do this. People who built those derived distributions had certain reasons to proceed this way.
If Debian should be used as the base for a commercial distribution like Linspire (formerly Lindows), Libranet or Xandros, there is no other choice than forking because these companies normally add some stuff that is non-free. While Debian Pure Blends might be interesting in technical terms for those commercial distributions by making it easier to build a separate distribution, these non-free additions are not allowed to be integrated into Debian, and thus integration into Debian is impossible.
As a completely free distribution Debian GNU/Linux is quite often a welcome starting point for derived distributions with a certain purpose that are as free as Debian but had certain reasons to fork. One main reason for a fork was that Debian was not flexible enough for certain purposes and some needed features had to be added. One reason for the Debian Pure Blends effort is to increase flexibility and to make the reason mentioned above void (if it is not yet void because of the general development of Debian). Some examples of forks from Debian that are probably now able to integrate back into Debian as a Debian Pure Blend are:
Mentioning SkoleLinux in the category of forks is more or less history. The merge back into Debian started with the SkoleLinux people really doing a great job to enhance Debian for their own purposes in the form of their work on debian-installer, and culminated with the formal merging of the Blend Debian Edu and SkoleLinux, so that they are now virtually equivalent. This is the recommended way for derived distributions, and the reasons for this recommendation are given below.
The Agnula project, which is founded by the European Community, (and in fact is the first Free Software project that was founded by the EU at all,) forked for the following reasons:
They had some special requirements for the kernel and configuration. This is more or less solved in the upcoming Debian release.
When DeMuDi started, not enough free programs in this field existed. This situation is better now.
Because of the founded status of the project, an extra distribution had to be developed. To accomplish this requirement, Debian Pure Blends plan to build common tools to facilitate building separate CDs with the contents of only a single distribution.
This shows that there is no longer a real need for a fork, and in fact, the organiser of the DeMuDi project was in contact to start bringing DeMuDi back into Debian. That is why DeMuDi is mentioned in the list of Debian Pure Blends above. Unfortunately the effort to merge back has stalled but it might be an interesting project to apply Blends techniques to support multimedia experts who want to use Debian.
LinEx is the very successful distribution for schools in the Region Extremadura in Spain. The work of the LinEx people perhaps made Debian more popular than any other distribution. The project was founded by the local government of Extremadura, and each school in this region is running this distribution. While this is a great success, the further development of LinEx has to face the problems that will be explained below. Because the creators of LinEx are aware of this fact they started joining the educational part of LinEx with Debian Edu which in turn led to an even stronger position of this Blend.
If developers of a non-commercial fork consider integrating back into Debian in the form of a Debian Pure Blend, it might happen that their field is covered already by an existing Blend. For instance, this would be the case for LinEx, which has the same group of target users as Debian Edu as explained above. On the other hand, some special adaptations might be necessary to fit the requirements of the local educational system. The specific changes that might be necessary would be called flavours of a Blend.
In general, a separate distribution costs extra effort. Because it is hardly possible to hire enough developers who can double the great work of many volunteer Debian developers, this would be a bad idea for economical reasons. These people would have to deal with continuous changes to keep the base system, installer, etc. up to date with the current Debian development. It would be more sane to send patches that address their special requirements to Debian instead of maintaining a complete Debian tree containing these patches.
Debian is well known for its strong focus on security. Security is mainly based on manpower and knowledge. So the best way to deal with security issues would be to base it on the Debian infrastructure, instead of inventing something new.
New projects with special intentions often have trouble to become popular to the user group they want to address. This is a matter of attaining the critical mass that was explained in Section 3.4, “General problem”.
Larger Free Software projects need certain infrastructure like web servers, ftp servers, (both with mirrors,) a bug tracking system, etc. It takes a fair amount of extra effort to build an entire infrastructure that is already available for free in Debian.
Forking would be a bad idea.
Debian has a huge user base all over the world. Any project that is integrated within Debian has a good chance to become popular on the back of Debian if the target users of the project just notice that it enables them to solve their problems. So there is no need for extra research on the side of the users, and no need for advertising for a special distribution. This fact has been observed in the Debian Med project, which is well known for many people in medical care. It would not have gained this popularity if it had been separated from Debian.
You get a secure and stable system without extra effort for free.
Debian offers a sophisticated Bug Tracking System for free, which is a really important resource for development.
There is a solid infrastructure of web servers, ftp servers with mirrors, mail servers, and an LDAP directory of developers with a strongly woven web of trust (through gpg key signing) for free.
By making changes to some packages to make them fit the needs of a target user group, the overall quality of Debian can be enhanced. In this way, enhancing Debian by making it more user friendly is a good way for the community to give back something to Debian. It would be a shame if somebody would refuse all the advantages to keeping a project inside Debian, and instead would decide to try to cope with the disadvantages because he just does not know how to do it the right way, and that it is normally easy to propagate changes into Debian. For instance, see Section C.1, “How to ask for packages which are not yet included”. This section explains how you can ask for a certain piece of software to be included in Debian. The next section describes the reason why Debian is flexible enough to be adapted to any purpose.
Debian is developed by about 1000 volunteers. Within this large group, the developers are encouraged to care for their own interests in packages they have chosen to look after. Thus, Debian is not bound to commercial interests.
Those who might fear this amount of freedom given to every single developer should realize that there are very strict rules, as laid out in Debian's policy, which glue everything together. To keep their packages in each new release, every developer must ensure that their packages abide by that policy.
One common interest each individual developer shares is to make the best operating system for himself. This way, people with similar interests and tasks profit from the work of single developers. If users, in turn, work together with the developers by sending patches or bug reports for further enhancement, Debian can be improved also for special tasks.
For instance, developers may have children, or may work in some special fields of work, and so they try to make the best system for their own needs. For children, they contribute to Debian Jr. or Debian Edu. For their field of work, they contribute to the appropriate Blend: Debian Med, Debian Science, and so forth.
In contrast to employees of companies, every single Debian developer has the freedom and ability to realize his vision. He is not bound to decisions of the management of his company. Commercial distributors have to aim their distributions at gaining a big market share. The commercial possibilities in targeting children's PCs at home are slight, so distributions comparable to Debian Junior are not attractive for commercial distributors to make.
Thus, single developers have influence on development - they just have to do it, which is a very different position compared with employees of a commercial distributor. This is the reason for the flexibility of Debian that makes it adaptable for any purpose. In the Debian world, this kind of community is called "Doocracy" - the one who does, rules.