Development

This section provides information on development process for the project, including instructions on how to set-up a development environment or run the tests locally.

Preparing development environment

There is a number of different ways a development environment can be set-up. The process outlined here is centered around virtualenvwrapper. Instructions have been tailored for a GNU/Linux system.

Before proceeding, ensure you have the following system-wide packages installed:

With those in place, do the following:

  1. Clone the git repository:

    git clone https://code.majic.rs/gimmecert/
    
  2. Change directory:

    cd gimmecert/
    
  3. Create Python virtual environment:

    Warning

    Make sure to specify Python 3 as interpreter.

    mkvirtualenv -a . -p /usr/bin/python3 gimmecert
    
  4. Install development requirements:

    pip install -e .[devel]
    
  5. At this point, any time you want to run tests etc, you can easily switch to correct environment (this will also put you in project root directory) with:

    workon gimmecert
    

Testing

Project includes both unit tests (within tests/ directory) , and functional tests (within functional_tests/ directory).

Tests can be run in a number of different ways, depending on what you want to test and how.

To run the unit tests via setup script, run the following command from repository root:

python setup.py test

To run the unit tests directly, run the following command from repository root:

pytest

Tests can be also run with coverage checks. By default coverage is configured to report to standard output. Report will only list files which lack coverage. For each file a percentage will be shown, as well as line numbers that were not covered by tests. To include coverage checks, run tests with:

pytest --cov

Warning

Gimmecert project has 100% coverage requirement. Anything below will trigger failures when coverage checks are run.

Should you desire to generate coverage report in HTML format, run (coverage report will be put into directory coverage/):

pytest --cov --cov-report=html:coverage/

Functional tests must be run explicitly (since they tend to take more time), with:

pytest functional_tests/

In addition to proper linting, implemented code should be pruned of unused imports and variables. Linting should be conformant to PEP8, with the exception of line length, which is allowed to be up to 160 characters. Linting and code sanity checks are not executed automatically, but can easily be run with:

flake8

Documentation should be buildable at all times. Documentation build can be triggered with a simple:

cd docs/
make html

Tests can also be run using tox:

Note

When running tests via tox, functional tests and coverage checks are included as well.

# Run full suite of tests on all supported Python versions.
tox

# List available tox environments.
tox -l

# Run tests against specific Python version.
tox -e py35

# Run documentation and linting tests only.
tox -e doc,lint

Running tests on all supproted Python versions

With a range of different Python versions supported, it might be somewhat difficult to run the tests against all the posible versions of Python depending on distribution used for development.

The projects comes with a Vagrantfile to make this easier. To run all tests within a Vagrant machine, perform the following steps:

  1. Go to project root directory:

    workon gimmecert
    
  2. Bring up the Vagrant machine (this may take a while since it will build the necessary Python versions):

    vagrant up
    
  3. Log-in into the Vagrant machine:

    vagrant ssh
    
  4. Change directory:

    cd /vagrant
    
  5. Run tests against all available environments:

Warning

The --workdir option should be used in order to avoid mixing of Python cache files from the host and the Vagrant virtual machine, and to avoid getting the error ERROR: Could not install packages due to an EnvironmentError: [Errno 39] Directory not empty: '__pycache__' during installation of Gimmecert package inside of Tox virtual environment. It is unclear at time of this writing why such an error would be triggered, but it could have something to do with the vboxsf filesystem used for sharing the /vagrant directory.

tox --workdir /tmp/

Building documentation

Documentation is written in reStructuredText and built via Sphinx.

To build documentation, run:

cd docs/
make html

Resulting documentation will be stored in HTML format in directory docs/_build/html/.

Versioning schema

Project employs semantic versioning schema. In short:

  • Each version is composed of major, minor, and patch number. For example, in version 1.2.3, 1 is the major, 2 is the minor, and 3 is the patch number.
  • Major number is bumped when making a backwards incompatible change.
  • Minor number is bumped when new features or changes are made without breaking backwards compatibility.
  • Patch number is bumped when backporting bug or security fixes into an older release.

In addition to versioning schema, project employs a specific nomenclature for naming the branches:

  • All new development (both for features and bug/security fixes) uses master branch as the base.
  • Features and bug/security fixes are implemented in a local branch based on the master branch. Local branches are named after the lower-cased issue number. For example, if the issuer number is GC-43, the implementation branch will be named gc-43. Normally these branches are only local, but if necessary they can be pushed to central repository for collaboration or preview purposes.
  • Patch releases are based off the maintenance branches. Mainteance branches are named after the MAJOR and MINOR number of the version - maintenance/MAJOR.MINOR. For example, if a new release is made with version 1.2.0, the corresponding branch that is created for maintenance will be named maintenance/1.2 (notice the absence of .0 at the end).

Backporting fixes

From time to time it might become useful to apply a bug/security fix to both the master branch, and to maintenace branch.

When a bug should be applied to maintenance branch as well, procedure is as follows:

  1. Create a new bug report in issue tracker. Target version should be either the next minor or next major release (i.e. whatver will get released from the master branch).
  2. Create a copy of the bug report, modifying the issue title to include phrase (backport to MAJOR.MINOR) at the end, with MAJOR and MINOR replaced with correct versioning information for the maintenance branch. Make sure to set correct target version (patch release).
  3. Resolve the bug for next major/minor release.
  4. Resolve the bug in maintenace branch by backporting (cherry-picking if possible) the fix into maintenace branch. Make sure to resign (cherry-picking invalidates OpenPGP signature) and reword (to reference the backport issue) the commit.

Release notes

Release notes are written in parallel to resolving project issues, in the docs/releasenotes.rst file. In other words, any time a new feature, bug fix etc is implemented, an entry should be created in the relase notes file. This applies for tasks and user stories as well.

By ensuring the release notes are always up-to-date, the release process is simpler, faster, and less error prone.

Release notes are always added under section title NEXT RELEASE. This placeholder section title is replaced during the release process.

Release notes for each version consist out of two parts - the general release description, and listing of resolved issues.

General description provides a high-level overview of new functionality and fixes included in the release, and points to any important/breaking changes.

The listing of resolved issues is split-up based on issue type, and lists all issues that have been resolved in the given release. Each issue in the list is provied as URL link pointing to issue URL in the issue tracker, with the link text in format ISSUE_NUMBER: ISSUE_TITLE. Both issue number and issue title are taken from the issue tracker.

To provide a more visual example, template for single release note is as follows:

NEXT RELEASE
------------

[General description of release.]

Resolved issues:

- **User stories**:

  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
- **Feature requests**:

  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
- **Enhancements**:

  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
- **Tasks**:

  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_
  - `ISSUER_NUMBER: ISSUE_TITLE <ISSUE_URL>`_

Release process

The release process for Gimmecert is centered around the use of included release.sh script. The script takes care of a number of tedious tasks, including:

  • Updating the version information in release notes and setup.py.
  • Tagging a release.
  • Starting a new section in release notes.
  • Switching version back to development in setup.py.
  • Publishing changes to origin Git repository and publishing release to PyPI.

When releasing a new major/minor version (from the master branch), the release script will take care of setting-up a maintenance branch as well.

Patch releases should be done from maintenance branches. New major/minor releases should be done from the master branch.

Warning

Keep in mind that the release script is interactive, it cannot be run unattended.

Perform the following steps in order to release new version of Gimmecert:

  1. Make sure that Git is correctly set-up for signing using GnuPG, and that the necessary key material is available.

  2. Verify that there are no outstanding issues for this release in the issue tracker.

  3. Switch to project virtual environment.

  4. Ensure that the repository is synchronised with origin, and that a correct branch is checked out (master or maintenance).

  5. Go through release notes for NEXT VERSION, and ensure the general description and listed issues look fine. Make any necessary changes, commit them, and push them to the origin.

  6. Prepare the release:

    Warning

    Make sure to provide correct version.

    ./release.sh prepare VERSION
    
  7. Verify that the preparation process was successful.

  8. Publish the release:

    Warning

    Make sure to provide correct version.

    ./release.sh publish VERSION
    
  9. Build release documentation on Read the Docs project page, and update the default version if this was a new major/minor release.

  10. Verify documentation looks good on Read the Docs documentation page.

  11. Mark the release issue as resolved in the issue tracker.

  12. Release the version via release center in the issue tracker. Upload source archive from the dist/ directory.

  13. Archive outdated releases in the issue tracker.