Skip to content

Latest commit

 

History

History
330 lines (237 loc) · 11.9 KB

CONTRIBUTING.md

File metadata and controls

330 lines (237 loc) · 11.9 KB

Checklist (and a short version for the impatient)

  • Commits:

    • Make commits of logical units.

    • Check for unnecessary whitespace with "git diff --check" before committing.

    • Commit using Unix line endings (check the settings around "crlf" in git-config(1)).

    • Do not check in commented out code or unneeded files.

    • The first line of the commit message should be a short description (50 characters is the soft limit, excluding ticket number(s)), and should skip the full stop.

    • The body should provide a meaningful commit message, which:

      • uses the imperative, present tense: "change", not "changed" or "changes".

      • includes motivation for the change, and contrasts its implementation with the previous behavior.

    • Make sure that you have tests for the bug you are fixing, or feature you are adding.

    • Make sure the test suite passes after your commit. More information see testing below.

    • When introducing a new feature, make sure it is properly documented in the README.md

  • Submission:

    • Prerequisites:

    • Preferred method:

      • Fork the repository on GitHub.

      • Push your changes to a topic branch in your fork of the repository.

      • Submit a pull request to the repository in the puppetlabs organization.

The long version

  1. Make separate commits for logically separate changes.

    Please break your commits down into logically consistent units which include new or changed tests relevant to the rest of the change. The goal of doing this is to make the diff easier to read for whoever is reviewing your code. In general, the easier your diff is to read, the more likely someone will be happy to review it and get it into the code base.

    If you are going to refactor a piece of code, please do so as a separate commit from your feature or bug fix changes.

    We also really appreciate changes that include tests to make sure the bug is not re-introduced, and that the feature is not accidentally broken.

    Describe the technical detail of the change(s). If your description starts to get too long, that is a good sign that you probably need to split up your commit into more finely grained pieces.

    Commits which plainly describe the things which help reviewers check the patch and future developers understand the code are much more likely to be merged in with a minimum of bike-shedding or requested changes. Ideally, the commit message would include information, and be in a form suitable for inclusion in the release notes for the version of Puppet that includes them.

    Please also check that you are not introducing any trailing whitespace or other "whitespace errors". You can do this by running "git diff --check" on your changes before you commit.

  2. Sign the Contributor License Agreement

    Before we can accept your changes, we do need a signed Puppet Labs Contributor License Agreement (CLA).

    You can access the CLA via the Contributor License Agreement link

    If you have any questions about the CLA, please feel free to contact Puppet Labs via email at [email protected].

  3. Sending your patches

    To submit your changes via a GitHub pull request, we highly recommend that you have them on a topic branch, instead of directly on "master". It makes things much easier to keep track of, especially if you decide to work on another thing before your first change is merged in.

    GitHub has some pretty good general documentation on using their site. They also have documentation on creating pull requests.

    In general, after pushing your topic branch up to your repository on GitHub, you can switch to the branch in the GitHub UI and click "Pull Request" towards the top of the page in order to open a pull request.

  4. Update the related GitHub issue.

    If there is a GitHub issue associated with the change you submitted, then you should update the ticket to include the location of your branch, along with any other commentary you may wish to make.

  5. Responding to feedback.

    We may have feedback for you to fix or change some things. We generally like to see that pushed against the same topic branch (it will automatically update the Pull Request). You can also fix/squash/rebase commits and push the same topic branch with -force (it's generally acceptable to do this on topic branches not in the main repository, it is generally unacceptable and should be avoided at all costs against the main repository).

    The only reasons a pull request should be closed and resubmitted are as follows:

    When the pull request is targeting the wrong branch (this doesn't happen as often). When there are updates made to the original by someone other than the original contributor. Then the old branch is closed with a note on the newer branch This supersedes #github_number.

Testing

Getting Started

Our puppet modules provide a Gemfile which can tell a ruby package manager such as bundler what Ruby packages, or Gems, are required to build, develop, and test this software.

Please make sure you have bundler installed on your system, then use it to install all dependencies needed for this project, by running:

% bundle install
Fetching gem metadata from https://rubygems.org/........
Fetching gem metadata from https://rubygems.org/..
Using rake (10.1.0)
Using builder (3.2.2)
-- 8><-- many more --><8 --
Using bundler (1.3.5)
Your bundle is complete!
Use `bundle show [gemname]` to see where a bundled gem is installed.

If you already have those gems installed, make sure they are up-to-date:

% bundle update

With all dependencies in place and up-to-date we can now run the tests:

% bundle exec rake test

This will execute all the rspec tests tests under spec/unit as well as run puppet-lint. Rspec tests may have the same kind of dependencies as the module they are testing. While the module defines in its Modulefile, rspec tests define them in .fixtures.yml.

You can run the acceptance tests as well by issuing the following command

% bundle exec rake acceptance

Note however that this will cost you money as it launches resources in AWS.

Unit Testing with VCR

VCR is a utility for testing which is used to capture the requests and responses to the AWS APIs. The results of a transaction are stored in a YAML file that can then be loaded later and used to validate data structures etc without requiring access to AWS. This comes in handy when unit testing using Rspec.

Consider the following test for a new provider.

require 'spec_helper'

provider_class = Puppet::Type.type(:snazzy_new_type).provider(:v2)

describe provider_class do

  before do
    ENV['AWS_ACCESS_KEY_ID'] = 'something goes here'
    ENV['AWS_SECRET_ACCESS_KEY'] = 'something else goes here'
    ENV['AWS_REGION'] = 'us-west-2'
  end

  let(:provider) { resource.provider }

The _spec.rb file here begins with the usual bit of boiler plate. In the before block here, we define three environment variables that will get loaded into the environment for each test within scope. This is where you put the access credentials for AWS. You may wish to create a separate AWS account for testing purposes to avoid leaking data from your production systems.

  VCR.use_cassette('snazzy-setup') do
    let(:instance) { provider.class.instances.first }
  end

Continuing on, the VCR block here tells the code within in the block to refer to a "cassette" called snazzy-setup for all API calls. If this file is not present, then the API calls must be made so the requests may be stored in the given cassette. This will require the credentials be actually loaded above. Once the responses are cached, the access credentials above are no longer required.

The cassette here refers to a file in the fixtures/vcr_cassettes directory. So here, we specify snazzy-setup, and so the resulting file that gets created and loaded is located at fixtures/vcr_cassettes/snazzy-setup.yml.

This first use of VCR is here just to cache the data necessary for the rest of the tests.

  describe 'some_instance_method' do
    it 'should exist' do
      VCR.use_cassette('snazzy-setup') do
        instance = provider.class.instances.first
        expect(instance.someproperty).to eq(512)
      end
    end
  end
end

Wrapping up our example here, we test an instance method called some_instance_method. We do this by loading the snazzy-setup cassette, then calling the provider instances method and store the first result in a variable called instance. Now we can make assertions about what we expect from that instance to test our provider functionality.

The instance on which we are making assertions needs to actually exist when the initial VCR setup is done above. Once the results are cached in the YAML file, then you can tear it down and still test effectively.

You may notice that some tests make use of multiple cassettes. One might test the environment for creating a resource. One might test the environment for tearing down a resource. This is all an exercise left to the developer.

Tidying up! IMPORTANT

Its important to tidy up on spec tests before committing. Its easy to do, so just remember to tidy up. You don't want to commit your credentials to GitHub, as I have done this morning.

Firstly, replace the credentials in the before block within any spec files that you have written.

Secondly, replace the credential reference in the resulting cassette yaml files. I use something like the following.

sed -i '' -e 's/ABCEFGOOOOOOOOOOOOOOR/AAAAAAAAAAAAAAAAAAAAA/g' -e 's/123456789101/111111111111/g' fixtures/vcr_cassettes/*.yml

The above 'sed' replaces my account id of '123456789101' with a string of matching length which is not my account id. Also above, we replace the access credential of 'ABCEFGOOOOOOOOOOOOOOR' with a string of matching length. I don't believe keeping the length of the strings is required, but I do it anyway.

If you do both of the above, then you should not have account information, identity information, or credential information in either the _sepc.rb test files, or the VCR cassette files.

If you have commit access to the repository

Even if you have commit access to the repository, you will still need to go through the process above, and have someone else review and merge in your changes. The rule is that all changes must be reviewed by a developer on the project (that did not write the code) to ensure that all changes go through a code review process.

Having someone other than the author of the topic branch recorded as performing the merge is the record that they performed the code review.

Additional Resources