The sustainability of packaging-farm depends on the number of its users. There currently are four: a company shipping a software suite to its customers ( Avencall with XiVO ), the software that initiated the development ( pokersource as found in Debian GNU/Linux with poker-eval, poker-engine etc), a university developed set of packages for bioinformatics ( biopp from université de montpellier ) and a single package web game in its early stages of development ( cardstori.es from farsides).
The problem packaging-farm is facing to get new users is described below in terms that may apply to other projects with a similar size and scope. No solution to the problem is found and there probably is none. The web site information was improved and a beginner guide written.
The target audience of packaging-farm is a software author or maintainer (who may be an individual, non-profit or company) who wants to distribute packages but don’t need to hire a release manager. It is best suited when packages need to be produced on a regular basis. If it only happens once or twice a year for a single package, it is better to use simpler tools such as pbuilder or sbuild. If the number of packages is large, to the point where it may require to hire a full time release manager, OBS or home made tools are more appropriate than packaging-farm. If the answer to most of the following questions is yes, it is worth giving packaging-farm a shot.
- Are you the author or maintainer of a software ?
- Do you want to maintain a package repository for more than one source package ?
- Do you have at least one source package ready for the software ?
- Are you planning to maintain packages for various distributions and architectures ?
- Eight hours per month is the maximum you can afford to spend to maintain the packaging-farm installation
- Twenty four hours is the maximum you can afford to spend understanding and configuring packaging-farm
- You need to maintain repositories enforcing a defined software life cycle for your customers
- The repository must be rebuilt as fast as possible after an upstream change
Some features related to packaging-farm are outside of its scope:
- It is not a continuous integration tool
- It does not provide automatic packaging such as alien
The development of packaging-farm was funded during four years by mekensleep and outflop for the need of the pokersource software suite. During this period, packaging-farm was a in-house tool it was difficult to run it without reading all the code base. It was then funded in 2011 by Avencall for the need of the XiVO software suite. It was simplified and documented to reduce the learning curve to a minimum.
There currently is no alternative solution to packaging-farm. OBS provides all the services packaging-farm does and more, but learning how to install it locally is best done if there is at least one person using it full time. Simpler tools such as pbuilder or sbuild lack the ability to handle cross distributions builds or package dependencies.
When packaging-farm could be used to save time, how can a company or non-profit know about it ? What would make them decide to give it a try ? How could they contribute to ensure the sustainability of the packaging-farm code base ? What would be their incentive to contribute ?
There are known recipes to solve these problems for larger projects, in particular when there is at least one person committed to it, either because she is employed by a company or because she decided to volunteer on a long term basis. However, they are not successfully applied in many smaller Free Software projects, such as packaging-farm. After eliminating the cases where this failure is caused by obvious reasons raging from a feature that nobody needs, an over complex solution to simple problem, a poorly tested code base that fails more often than it succeeds, etc. etc. a lot of projects remain.
For such a project, what can be done to find users without spending more in marketing than in development ?
In 2011 there has been a total of 45 days spent on the development of packaging-farm. Can packaging-farm get one contributing user if 8 days are dedicated to find such a user ? It is very unlikely and if such a work was to be subcontracted, there is no company or freelancer offering such services within such short delays. However, the fact that packaging-farm is ready to use for a number of potential users suggests that there may be a solution worth looking for.
Improved information on the web site
The packaging-farm http://packaging-farm.dachary.org/ was improved to better describe the audience and add contact information. A beginner guide was written to help potential users get a sense of what using packaging-farm entails.