Debian Pure Blends Manifesto
For years, Debian has described itself as the universal operating system. Through its open development processes, it has exploded in size to contain at present more than 20,000 packages. Debian contains the best free software available for businesses, programmers, scientists, doctors, lawyers, dentists, desktop users and professionals of all kinds -- and the depth and breadth of the software Debian contains is only increasing.
Following the wild success of Debian, businesses, governments and individuals seized onto Debian as both a distribution model and a source of code. The creators of live CDs like Knoppix and Morphix have harnessed the power and consistency of Debian to create bootable linux distributions.
Other groups have chosen to work within the organization and technical structures of Debian. These groups have started subprojects to create so called Debian Pure Blends (formerly known as Custom Debian Distributions).
(work in progress)
/[should be short definitions, max 3/4 lines]/
Debian Pure Blend
(Targeted to a specific use, 100% Debian, customized package selection, pre-configuration of some packages.)
A Debian Pure Blend (in short Blend) is a subset of Debian that is tailored to be used out-of-the-box in a particular situation or by a particular target group.
Any changes necessary to support this are merged back into Debian whenever possible, improving Debian as a whole.
(we should solve the andreas/cobaco dispute about if a blend is not 100% debian compliant should be considered an ... (find name here) Blend or a derived distribution)
Same as Debian Pure Blend or kind of further specialization of a Blend (see http://lists.debian.org/debian-devel/2003/debian-devel-200312/msg00337.html) ?
Derived Debian Distribution
Non 100% compliant Debian derivative, à-la Knoppix, Morphix, Lindows?
- relation to users/purpose
- each Blend is aimed at a particular target group/purpose
- working out-of-the-box for that target group
- possibility of having a full Debian distribution pre-configured for a certain purpose
- possibility to bui target-specific support and know-how
- relation to Debian as a whole
- aim to be a subset, not always possible yet
- focus on certain target group/purpose
- integrate improvements back into Debian
- mediate between debian and non-technical realities
- help specific communities with a debian distribution better customized for them
- bring ideas and creativity from non-technical communities into Debian
- relation to companies
userlinux paper has ideas we could take:
- commercial distributions not catering to the sophisticated needs of certain (small) target groups with particular needs result in hampering the adoption of Linux
- economics of open source working worsts for commerial distros (hm, links with Enrico's idea of externalities?)
- the idea of a shared, commonly developed core, with companies delivering support/services using that core (something we're trying to create for Skolelinux, in Norway this is happening already). That this structure establishes economic limits on the service companies. While they can make a living, they can't make a killing
- Trust in a brand will come easier when the brand is one developed for a target group, by that target group
- The idea of Blends being of interest to many target groups that would like an inexpensive and reliable GNU/Linux system, with payment for service and engineering rather than "seats"
- (enrico's take about externalities)
- enrico's waiting for people.debian.org to become sane again to put up slides for a talk he's made about that.
It basically talks about gathering into Debian the externalities derived by the activity of developing free software. The externalities coming from different Blends into Debian get then part of a network and are subjected to Metcalfe's Law, contributing to a powerful creative process and improving by great lengths the quality of Debian as a whole. This improvement of Debian, then, automatically feeds back into the Blends, who get their externalities back, and augmented. [I'm not an economist, though, and I'd like to submit this idea to some review].
- relations between Blends
- common problems + any solutions present/proposed (some stuff about this on the wiki)
- communication/cooperation between Blends
Texts extracted from debian-devel threads