Feel free and welcome to contribute to this project. You can start with filing issues and ideas for improvement in GitHub tracker. Our favorite thoughts from The Zen of Python:

  • Beautiful is better than ugly.

  • Simple is better than complex.

  • Readability counts.

We respect the PEP8 Style Guide for Python Code. Here’s a couple of recommendations to keep on mind when writing code:

  • Maximum line length is 99 for code and 72 for documentation.

  • Comments should be complete sentences.

  • The first word should be capitalized (unless identifier).

  • When using hanging indent, the first line should be empty.

  • The closing brace/bracket/parenthesis on multiline constructs is under the first non-whitespace character of the last line.

When generating user messages use the whole sentence with the first word capitalized and enclose any names in single quotes:

self.warn(f"File '{path}' not found.")


It is challenging to be both concise and descriptive, but that is what a well-written summary should do. Consider the commit message as something that will be pasted into release notes:

  • The first line should have up to 50 characters.

  • Complete sentence with the first word capitalized.

  • Should concisely describe the purpose of the patch.

  • Do not prefix the message with file or module names.

  • Other details should be separated by a blank line.

Why should I care?

  • It helps others (and yourself) find relevant commits quickly.

  • The summary line will be re-used later (e.g. for rpm changelog).

  • Some tools do not handle wrapping, so it is then hard to read.

  • You will make the maintainers happy to read beautiful commits :)

You can get some more context in the stackoverflow article.


In order to experiment, play with the latest bits and develop improvements it is best to use a virtual environment. Make sure that you have all required packages installed on your box:

make develop

Create a development virtual environment with hatch:

git clone
cd tmt
hatch env create dev

Enter the environment by running:

hatch -e dev shell

Install the pre-commit script to run all available checks for your commits to the project:

pre-commit install


Every code change should be accompanied by tests covering the new feature or affected code area. It’s possible to write new tests or extend the existing ones.

If writing a test is not feasible for you, explain the reason in the pull request. If possible, the maintainers will help with creating needed test coverage. You might also want to add the help wanted and tests needed labels to bring a bit more attention to your pull request.

Run the default set of tests directly on your localhost:

tmt run

Run selected tests or plans in verbose mode:

tmt run --verbose plan --name basic
tmt run -v test -n smoke

Build the rpms and execute the whole test coverage, including tests which need the full virtualization support:

make build-deps
make rpm
tmt -c how=full run

This would install the freshly built rpms on your laptop. In order to run the full test suite more safely under a virtual machine run the full test suite wrapper against the desired branch:

cd tests/full
tmt run --environment BRANCH=target

Or schedule the full test suite under an external test system:

cd tests/full
tmt test export --fmf-id | wow fedora-35 x86_64 --fmf-id - --taskparam=BRANCH=target

Or run local modifications copied to the virtual machine. Because this requires changes outside of the fmf root you need to run make which tars sources to the expected location:

cd tests/full
make test

Similar as above but run only tests which don’t run for merge requests:

cd tests/full
make test-complement

To run unit tests in hatch environment using pytest and generate coverage report:

make coverage

To see all available scripts for running tests in hatch test virtual environments:

hatch env show test

To run ‘unit’ script for example, run:

hatch run test:unit

When running tests using hatch, there are multiple virtual environments available, each using a different Python interpreter (generally the lowest and highest version supported). To run the tests in all environments, install the required Python versions. For example:

dnf install python3.9 python3.11


When adding new unit tests, do not create class-based tests derived from unittest.TestCase class. Such classes do not play well with Pytest’s fixtures, see for details.


Tests which try various provision methods should use PROVISION_METHODS environment variable to select which provision methods they can utilize during their execution. This variable is likely to have default container or local and use adjust rule for how=full to add virtual method.


When submitting a change affecting user experience it’s always good to include respective documentation. You can add or update the Metadata Specification, extend the Examples or write a new chapter for the user Guide.

By default, examples provided in the specification stories are rendered as yaml. In order to select a different syntax highlighting schema add # syntax: <format>, for example:

# syntax: shell

Building documentation is then quite straightforward:

make docs

Find the resulting html pages under the docs/_build/html folder.

Use the TMT_DOCS_THEME variable to easily pick custom theme. If specified, make docs would use this theme for documentation rendering by Sphinx. The theme must be installed manually, make docs will not do so. Variable expects two strings, separated by a colon (:): theme package name, and theme name.

# Sphinx book theme, sphinx-book-theme:
TMT_DOCS_THEME="sphinx_book_theme:sphinx_book_theme" make docs

# Renku theme, renku-sphinx-theme - note that package name
# and theme name are *not* the same string:
TMT_DOCS_THEME="renku_sphinx_theme:renku" make docs

Pull Requests

When submitting a new pull request which is not completely ready for merging but you would like to get an early feedback on the concept, use the GitHub feature to mark it as a Draft rather than using the WIP prefix in the summary.

During the pull request review it is recommended to add new commits with your changes on the top of the branch instead of amending the original commit and doing a force push. This will make it easier for the reviewers to see what has recently changed.

Once the pull request has been successfully reviewed and all tests passed, please rebase on the latest main branch content and squash the changes into a single commit. Use multiple commits to group relevant code changes if the pull request is too large for a single commit.

The following checklist template is automatically added to the new pull request description to easily track progress of the implementation and prevent forgetting about essential steps to be completed before it is merged. Feel free to remove those which are irrelevant for your change.

Pull Request Checklist

* [ ] implement the feature
* [ ] write the documentation
* [ ] extend the test coverage
* [ ] update the specification
* [ ] adjust plugin docstring
* [ ] modify the json schema
* [ ] mention the version
* [ ] include a release note

The version should be mentioned in the specification and a release note should be included when a new essential feature is added or an important change is introduced so that users can easily check whether given functionality is already available in their package:

.. versionadded:: 1.23

If the pull request addresses an existing issue, mention it using one of the automatically parsed formats so that it is linked to it, for example:

Fix #1234.


Pull request merging is done by one of maintainers who have a good overview of the whole code. Maintainer who will take care of the process will assign themselves to the pull request. Before merging it’s good to check the following:

  • New test coverage added if appropriate, all tests passed

  • Documentation has been added or updated where appropriate

  • Commit messages are sane, commits are reasonably squashed

  • At least one positive review provided by the maintainers

  • Merge commits are not used, rebase on the main instead

Pull requests which should not or cannot be merged are marked with the blocked label. For complex topics which need more eyes to review and discuss before merging use the discuss label.


There are several Makefile targets defined to make the common daily tasks easy & efficient:

make test

Execute the unit test suite.

make smoke

Perform quick basic functionality test.

make coverage

Run the test suite under coverage and report results.

make docs

Build documentation.

make packages

Build rpm and srpm packages.

make images

Build container images.

make tags

Create or update the Vim tags file for quick searching. You might want to use set tags=./tags; in your .vimrc to enable parent directory search for the tags file as well.

make clean

Cleanup all temporary files.


Follow the steps below to create a new major or minor release:

  • Run the full test coverage using tmt -c how=full run

  • Use git log --oneline --no-decorate x.y-1.. to generate the changelog

  • Update README with new contributors since the last release

  • Add a Release tmt-x.y.0 commit with the specfile update

  • Create a pull request with the commit, ensure tests pass, merge it

Release a new package to Fedora and EPEL repositories:

  • Move the fedora branch to point to the new release

  • Tag the commit with x.y.0, push tags git push --tags

  • Create a new github release based on the tag above

  • Check Fedora pull requests, make sure tests pass and merge

Finally, if everything went well:

  • Close the corresponding release milestone

  • Once the copr build is completed, move the quay branch to point to the release commit as well to build fresh container images.

Handle manually what did not went well:

  • If the automation triggered by publishing the new github release was not successful, publish the fresh code to the pypi repository manually using make wheel && make upload

  • If there was a problem with creating Fedora pull requests, you can trigger them manually using /packit propose-downstream in any open issue.

  • If running packit propose-downstream from your laptop make sure that the post-upstream-clone action is disabled in .packit.yaml to prevent bumping the devel version.