Rumprun-packages is a work-in-progress repository for software running on the Rumprun unikernel.

Packaging is done in a BSD ports-like fashion, with individual packages as subdirectories of this repository.

NOTE to users: check the license of each individual package to make sure it suits your deployment needs. Unlike everything else provided by the rump kernel project, we do not guarantee a BSD, ISC or CC0 style license for every piece of 3rd party software linked from here.

We are working towards choosing a real packaging system, with support for versions and dependencies. Until we reach that point, this repo is meant to assemble the build scripts and patches for running various applications on Rumprun unikernels.

To get started:

1. Build a Rumprun toolchain (with build-rr.sh in the rumprun repo).
2. Add destdir/bin from Step 1 to your $PATH by . "$(pwd)/obj-amd64-<arch>/config-PATH.sh".
3. Copy the config.mk.dist file to config.mk and set RUMPRUN_TOOLCHAIN_TUPLE to specify the compiler toolchain to use for building, for example, x86_64-rumprun-netbsd or i486-rumprun-netbsdelf.
4. Refer to the package-specific README file for build instructions.

Tutorials you may want to complete for more in-depth knowledge:

• http://wiki.rumpkernel.org/Tutorial%3A-Building-Rumprun-Unikernels
• http://wiki.rumpkernel.org/Tutorial%3A-Serve-a-static-website-as-a-Unikernel

New packages are contributed by creating the package and, after sufficient testing, opening a pull request. If you are a contributor with push access to rumprun-packages, you may also push directly into the repository instead of going the pull request route. If you push directly, make sure you push only the commit(s) you meant to push. Opening a pull request and merging it from the GitHub web interface gives you an extra review step, so it is safer in that regard.

When creating new packages, include the following sections in the package-specific README:

• Overview: name of the package, version number, and a sentence or two on how the package can be used
  • Maintainer: maintainer of the package (you!). Required: github account. Optional: name, email address, irc nick.
• Instructions [if an application and not e.g. library]: free-form description of how to bake the packaged software
• Examples [optional but highly recommended]: a few examples on how to run/use the resulting package

We require that new packages have a maintainer. A maintainer should generally be interested in the welfare of a package by answering potential user questions, addressing problem reports, and updating the package especially when security vulnerabilities are discovered. All maintainers are given push access to the repository. In case you created a package but do not want to maintain it, submit a pull request anyway; we will mark the pull request as maintainer wanted, and if someone who needs the software comes along and is willing to be maintainer, they potentially do not have to start from scratch.

When creating new packages, renaming or removing existing packages, please update .travis.yml to reflect your changes, and .travis-install.sh if your package has any build dependencies not installed in the Travis CI environment we use (dist: trusty).

If something is unclear, see existing packages for examples or ask on the mailing list or on irc.

Due to the time taken for a complete build of all packages (3+ hours as of this writing) we do not use the traditional model of triggering a build on each commit. Instead, full builds of all packages are triggered twice a day, currently at 6am and 6pm CET.

Pull requests are not currently built as there is no straightforward way to get Travis to build only the subset of packages affected by a pull request.