debci development

Setting up a development environment

There are two ways to setup the development environment:

  • Vagrant. This method is recommended if your machine is not running Debian (or a derivative like Ubuntu), or if you don't want to install the development dependencies on your main system. vagrant will setup the development environment in a virtual machine for you. This will require you to have some minimal understanding of what Vagrant is and how it works.
  • Manual Setup. This method is recommended if your machine is running Debian (or a derivative like Ubuntu). It is faster and more practical, but will install a bunch of packages on your system.

You should only pick one of those.

Vagrant

Prerequisites

Install the virtual machine

Run this at the same path where VagrantFile is

$ vagrant up

Start debci

SSH into vagrant environment

$ vagrant ssh

Once inside vagrant, you can start run debci with the following commands

$ cd /vagrant
$ foreman start

Note: The other commands are the same as the ones mentioned in manual setup.

Manual Setup

Grab the dependencies and required software

Install the dependencies and build dependencies:

$ sudo apt-get build-dep .

If that fails with a complaint that any package it not recent enough, then you probably need to enable the backports repository and install those packages from there (replace <stable> with the current Debian stable release codename, and <PACKAGE> with the package you want to install):

sudo apt-get install -t <stable>-backports <PACKAGE>

One of the dependencies that you should have installed above is rabbitmq-server. You might not want to have it running at all times. To disable rabbitmq-server, you can run:

$ sudo systemctl disable rabbitmq-server

If you disabled rabbitmq-server, you will have to remember to start it before hacking on debci:

$ sudo systemctl start rabbitmq-server

Set up the test environment

After having the dependencies installed, you now have to do some setup. The exact steps depend on your goal with debci.

The most common case is that you want to work in aspects of the system that do not involve the actual test backends, e.g. the user interface, or the database. For that, you can use helper script to the setup for you:

$ ./tools/init-dev.sh

The above script will create:

  • a package seed_list in config/seed_list; this limits the set of packages that will be worked on, reducing the time it takes for processing everything on your local tests.
  • a configuration file at config/conf.d/dev.conf which sets architectures and suites. It also sets the debci backend to the fake backend, which is very fast (because it does not really run tests, it just produces "fake" test runs with random results)

Note: the fake backend gets package versions from your local system. So, for example if you are on Debian stable, when "running tests" for package foo, the fake backend will report as testing the version of foo that is available on Debian stable. If for some reason you want or need it to report, say, versions that look like the ones from Debian unstable, all you have to do is add a sources.list entry for Debian unstable.

If you want to work on an actual test backend, then you will want to modify config/conf.d/dev.conf to set the backend to the one you want to work on.

Get debci up and running

Now you need to compile a few files that will be part of the user interface:

$ make

Now initialize the database:

$ ./bin/debci migrate

debci is composed of a few daemons; you can run all of them in one shot by running:

$ foreman start

This will start:

  • one debci worker daemon, which runs tests.
  • one debci collector daemon, which receives test results, and generates data files and HTML for the web interface.
  • one web server daemon.
  • one indexer daemon, which generates the HTML UI from the data directory

Now leave those daemons running, and open a new terminal to continue working.

To visualize the web interface, browse to http://localhost:8080/

You will notice that the web interface looks a little empty. To generate some test data, run

$ ./tools/gen-fake-data.sh

The command above will submit one test job for each package on each suite and each architecture. If you changed the backend from fake to something else, you might not want to do this.

If you go back to the terminal that is running the debci daemons, you will see a few messages there related to test jobs you just submitted.

To schedule a single test run, run:

$ ./bin/debci enqueue $PACKAGE

debci web UI development

Starting out

If you are interested in working on the web UI, first make sure that you have a development environment setup and some test data.

The web UI is generated using Ruby and ERB templates. The Debci::HTML class in lib/debci/html.rb is responsible for generating all of the pages for the web UI by using the templates.

The templates contain HTML and debci Ruby API calls to display information in the interface.

The templates are contained in the lib/debci/html/ directory while the debci Ruby API is contained a directory lower in lib/debci/.

If you make changes to the documentation (HACKING.md, etc.), run the following to regenerate it:

$ make

With the web interface running, you should see your changes with a refresh of the web page.

NOTE: Try to keep lines under 80 characters in length unless it would cause the code to look weird or less readable.

Implementing new features for the debci web interface

If you are developing a new feature for the debci web UI, make sure that if you develop any new debci Ruby API calls that you add tests for them in the appropriate test file. (e.g. If you add a method to Debci::Job, make sure that the method has tests in spec/debci/job_spec.rb)

Running tests on your code

After adding tests for the new method in the appropriate test file, run the following:

$ make spec

This will run all tests using rspec. You should see output similar to the following:

rspec --color
................................................................

Finished in 0.05459 seconds
64 examples, 0 failures

If your code passed the appropriate tests, you will see that there are no failures reported by rspec.

Contribution guidelines