tetra – builds RPMs for Java software

tetra is a tool to help you build RPM packages for Java projects.

Packaging of Java projects is sometimes hard - mainly because build tools like Maven partially overlap with distro ones like RPM.

See MOTIVATION.md for further information.

Prerequisites

Ruby 3.1 or later;
git with your credentials set in ~/.gitconfig (name and email);
some basic Unix commands: bash, unzip, tar;
a JDK > 8 that can compile whatever software you need to package;
Fedora only: packages to compile native gems, use yum install ruby-devel gcc-c++ zlib-devel;

Installation

You can install tetra via RubyGems or by building and installing it yourself.

# via RubyGems (currently 2.0.6 from 02/2020)
gem install tetra

# from source
rake install

Tests

Test can be found in the spec/ folder and are divided into fine (more similar to unit tests) and coarse (more similar to integration tests). To run tests manually, simply execute rake. With every Pull Request against master and consecutive pushes to that PR, several GitHub Actions tests will run automatically.

Workflow

Building a package with tetra is quite unusual — this is a deliberate choice, so don't worry. Basic steps are:

tetra init <package name> package_sources.tar.gz to initialize the project and unpack original sources;
cd into the newly created <package name> directory
if anything other than ant and mvn is needed in order to build the project, add it to the kit/ directory in binary form;
execute tetra dry-run, which will open a bash subshell. Build your project, and when you are done conclude by exiting the subshell with Ctrl+D;
cd into the directory that contains pom.xml and execute tetra generate-all: tetra will scaffold spec files and tarballs.

Done!

How can that possibly work?

During the dry-run tetra:

saves your bash history, so that it can use it later to scaffold a build script;
keeps track of changed files, in particular produced jars, which are included in the spec's %files section;
saves files downloaded from the Internet (eg. by Maven) and packs them to later allow networkless builds.

Note that with tetra you are not building all dependencies from source - build dependencies are aggregated in a binary-only "kit" package. This is typically sufficient to fulfill open source licenses and to have a repeatable, networkless build.

A commons-collections walkthrough

First, grab the sources:

wget https://archive.apache.org/dist/commons/collections/source/commons-collections4-4.5.0-M2-src.tar.gz

Second, create a new tetra project named commons-collections based on those sources:

tetra init commons-collections commons-collections4-4.5.0-M2-src.tar.gz
cd commons-collections/src/commons-collections4-4.5.0-M2-src

Third, you need to show tetra how to build your package. Run tetra dry-run and a new subshell will open, in there do anything you would normally do to build the package (in this case, run Maven):

tetra dry-run
mvn package
^D

Note that you don't even need to install Maven - tetra bundles a copy in kit/ and uses it by default! Also note that this being a dry-run build, sources will be brought back to their original state after it finishes to ensure repeatability.

Finally, generate build scripts, spec files and tarballs in the packages/ directory:

tetra generate-all

Note that tetra will generate files for the commons-collections package and all binary-only build-time dependencies in the packages/ folder.

In-depth information

In more complex cases building a package will require some special tweaks. We are trying to cover the most common in the SPECIAL_CASES.md file.

An in-depth discussion of this project's motivation is available in the MOTIVATION.md file.

Status

tetra is considered stable. At the moment tetra is tested on openSUSE and on Ubuntu with GitHub Actions.

Security

Please see SECURITY.md for more information.

Sources

tetra's Git repo is available on GitHub.

License

MIT license.

uyuni-project/tetra