Last updated: 2015-10-14 20:00:29 +0100

Upstream URL: git clone


View repository

View issue tracker

Contents of README follows

Ubuntu Service Pack Maker Prototype

Legal rubbish

The folders com, jeta.resources, META_INF and org are needed due to the RAD tool used to make the GUI, and thus are not part of the application itself and do not fall under my copyrights.

The actual application files (,, and and compiled files) have been put into the Public Domain. Do what you want with them. However, this tool is not finished and forking development would be ridiculous at this stage, so it is encouraged that the application as-is is not to be used outside of testing and development.

Introduction and Explanation

This is a non-functional prototype for a tool to enhance the usefulness of the current Debian package management system. Its purpose is to create isolated package repositories for user-specified pieces of software which contain all needed dependencies along with a meta-package which depends on them. This makes distributing software as packages much easier, since the individual qualities of the package management system are not changed (updates, dependencies, etc.), but the current hassles of distributing packaged software, along with needed dependencies such as libraries, (which currently involves either setting up a new repository, making users manually install large numbers of packages in a certain order, or collecting all of the needed dependencies into one huge, policy-breaking package) can be overcome with the use of Service Packs.

Service packs, as made by this tool, are small APT repositories which house all of the packages needed by their components (with user-specified assumptions allowed, for example that a package such as Ubuntu’s “ubuntu-desktop” is installed, which guarantees certain packages, or at a lower level that LSB functions will be available).

In each service pack, accompanying the repository, is a meta-package (a package which contains no files, but still has dependencies) with the same name as the service pack which depends on all of the packages specified to be in the service pack (but not their dependencies, which may change on updates). The meta-package can also depend on the meta-packages of other service packs, and thus assume that the contents of that service pack are also installed (this would allow incremental service packs, like monthly updates, to be kept small as they only contain changes from date of the last service pack, not from scratch).

Upon installation of the meta-package (which sits in a directory outside the repository, and thus is easily accessible) the repository is temporarily added as an APT source, the package is installed along with its dependencies, which if not already available will be taken from the service pack’s repository, then the service pack is removed as an APT source and the service pack has finished installing.

The main point to note about the functioning of this tool is that it does not propose a new way of installing software. It does not propose any new formats for saving software. It exists solely to use the existing APT framework in a new way. For installing software, using a service pack is essentially the same as using an online APT repository, except that it is all saved locally so machines with no network can use it (yes, APT repositories can be local too, but require more setting up. This tool essential eases that setting up) and the repository handling is setup and cleaned up automatically during installation of a service pack (ie. there is no need to enter and clean up repository information manually, as is the case when using a CD as an APT source for example). A disadvantage of service packs compared to regular repositories is the lack of updates (installed packages will still get updated via any regular repositories used, but for packages only distributed via service packs there will be no updates until a new service pack is released).

Technical Info for Developers

This prototype is written in Java, only because that is the language I am most familiar with due to its use in my computing courses. The final implementation will be in Python and GTK/QT, since a small one-off-use tool like this would not cause any massive performance drain using an interpreted language.

Here is a description of the components and their current status: - This is the runnable class (ie. running “java MainInterface” in a console will execute it) The “Advanced” button should really be a tab. A “browse” button should sit alongside the “location” entry. Standalone (not in rpository) packages should be addable, perhaps by specifying a full path instead of just a package name The createPack method requires work (using PackageList to give Packages dependencies, etc.). - This represents a service pack, and also contains a static method for creating packages based on text containing their names. - This represents a Debian package. Dependency adding needs to be sorted out. - This handles the system’s APT information, namely a large list of the available packages and their associated information (dependencies, etc.) Needs work opening and using the list files.

More Information

Any help is appreciated. A specification for this tool can be found on the Ubuntu Wiki at

Thanks, Warbo