SCUBA

Simple Container-Utilizing Build Apparatus

SCUBA is a simple tool that makes it easier to use Docker containers in everyday development. It is intended to be used by developers in 'make' or 'scons' based build environments, where the entire build environment is encapsulated in a Docker container.

Its purpose is to lower the barrier to using Docker for everyday builds. SCUBA keeps you from having to remember a complex docker run command line, and turns this

$ docker run -it --rm -v $(pwd):/build:z -w /build -u $(id -u):$(id -g) gcc:5.1 make myprogram

into this:

$ scuba make myprogram

Installation

Install via pip

Scuba is hosted at PyPI.

To install:

$ sudo pip install scuba

To install with argcomplete (for bash completion support):

$ sudo pip install scuba[argcomplete]

To uninstall:

$ sudo pip uninstall scuba

Install from source

Scuba can be built from source on Linux only (due to the fact that scubainit must be compiled):

1. Run make to build scubainit
2. Run ./run_nosetests.py to run the unit tests
3. Run sudo python setup.py install to install scuba
4. Run ./run_full_tests.py to test the installed version of scuba

If musl-libc is installed, it can be used to reduce the size of scubainit, by overriding the CC environment variable in step 1:

CC=/usr/local/musl/bin/musl-gcc make

Note that installing from source in this manner can lead to an installation with increased startup times for Scbua. See #71 for more details. This can be remedied by forcing a wheel to be installed, as such:

$ export CC=/usr/local/musl/bin/musl-gcc    # (optional)
$ sudo pip install wheel
$ python setup.py bdist_wheel
$ sudo pip install dist/scuba-<version>-py3-none-any.whl

Configuration

Configuration is done using a YAML file named .scuba.yml in the root directory of your project. It is expected that .scuba.yml be checked in to version control. Full documentation on .scuba.yml can be found in doc/yaml-reference.md, and specific examples can be found in the example directory.

An example .scuba.yml file might look like this:

image: gcc:5.1

aliases:
  build: make -j4

In this example, scuba build foo would execute make -j4 foo in a gcc:5.1 container.

Environment

Scuba defines the following environment variables in the container:

• SCUBA_ROOT the root of the scuba working directory mount; the directory where .scuba.yml was found.

Bash Completion

Scuba supports command-line completion using the argcomplete package. Per the argcomplete README, command-line completion can be activated by:

• Running eval "$(register-python-argcomplete scuba)" manually to enable completion in the current shell instance
• Adding eval "$(register-python-argcomplete scuba)" to ~/.bash_completion
• Running activate-global-python-argcomplete --user to install the script ~/.bash_completion.d/python-argcomplete.
  • NOTE: The generated file must be sourced, which is not the default behavior. Adding the following code block to ~/.bash_completion is one possible solution.

    for bcfile in ~/.bash_completion.d/*; do
        [ -f "$bcfile" ] && . "$bcfile"
    done

• Running activate-global-python-argcomplete as root (or sudo) to use argcomplete for all users

License

This software is released under the MIT License.