.. _`Development`: Development =========== This section only needs to be read by developers of the pywbem package. People that want to make a fix or develop some extension, and people that want to test the project are also considered developers for the purpose of this section. .. _`Repository`: Repository ---------- The repository for pywbem is on GitHub: https://github.com/pywbem/pywbem .. _`Setting up the development environment`: Setting up the development environment -------------------------------------- You may use any supported OS platform as the development environment for pywbem. On native Windows, you need to use a Windows command prompt in administrator mode. 1. It is recommended that you set up a `virtual Python environment`_. Have the virtual Python environment active for all remaining steps. 2. Make sure the following commands are available: - ``git`` - ``make`` (GNU make) - ``choco`` on native Windows (`Chocolatey package manager`_) 3. Clone the Git repo of this project and switch to its working directory: .. code-block:: bash $ git clone git@github.com:pywbem/pywbem.git $ cd pywbem 4. Install the prerequsites for pywbem development. This will install Python packages into the active Python environment, and OS-level packages: .. code-block:: bash $ make develop 5. This project uses Make to do things in the currently active Python environment. The command: .. code-block:: bash $ make displays a list of valid Make targets and a short description of what each target does. See the next sections for details. .. _virtual Python environment: https://docs.python-guide.org/en/latest/dev/virtualenvs/ .. _`Chocolatey package manager`: https://chocolatey.org/ .. _`Building the documentation`: Building the documentation -------------------------- The ReadTheDocs (RTD) site is used to publish the documentation for the pywbem package at https://pywbem.readthedocs.io/ This page is automatically updated whenever the Git repo for this package changes the branch from which this documentation is built. In order to build the documentation locally from the Git work directory, execute: .. code-block:: bash $ make builddoc The top-level document to open with a web browser will be ``build_doc/html/docs/index.html``. .. _`Testing`: .. # Keep the tests/README file in sync with this 'Testing' section. Testing ------- All of the following `make` commands run the tests in the currently active Python environment, and need to be invoked in the Git repo work directory. By default, the tests use the `pywbem` and `pywbem_mock` modules from the respective directories in the Git repo work directory. Pywbem 0.14.5 introduced a way to test installed versions of the pywbem package. For details, see :ref:`Testing installed versions of pywbem`. The `tests` directory has the following subdirectory structure: :: tests +-- unittest Unit tests | +-- utils Utility functions used by unit tests | +-- pywbem Unit tests for the pywbem package | +-- pywbem_mock Unit tests for the pywbem_mock package | +-- unittest_utils Unit tests for tests/unittest/utils | +-- functiontest Unit tests for tests/functiontest | +-- end2endtest_utils Unit tests for tests/end2endtest/utils | +-- servers Unit tests for tests/servers +-- functiontest Function tests +-- end2endtest End2end tests | +-- utils Utility functions used by end2end tests +-- manualtest Manual tests +-- server_definitions WBEM server definition file used by some tests and module | for accessing it +-- profiles Simple definitions of management profiles used by some tests +-- schema The CIM schema MOF files used by some MOF tests +-- dtd The CIM DTD file used by some CIM-XML validation tests There are multiple types of tests in pywbem: 1. Unit tests and function tests These tests do not require any WBEM server to be available, and the tests validate their results automatically. The distinction between unit tests and function tests as used in pywbem is that function tests exercise the entire pywbem client component or entire pywbem scripts, while unit tests exercise single modules. They are run by executing: .. code-block:: bash $ make test Test execution can be modified by a number of environment variables, as documented in the make help (execute `make help`). 2. End2end tests These tests are run against one or more WBEM servers, and the tests validate their results automatically. They are run by preparing a server definition file: :: tests/server_definitions/server_definition_file.yml from the provided example, and by executing: .. code-block:: bash $ make end2endtest Again, test execution can be modified by a number of environment variables, as documented in the make help (execute `make help`). 3. Manual tests There are several Python scripts and shell scripts that can be run manually. The results need to be validated manually. These scripts are in the directory: :: tests/manualtest/ and are executed by simply invoking them from within the main directory of the repository, e.g.: :: tests/manualtest/run_cim_operations.py Some of the scripts support a `--help` option that informs about their usage. The `run_cim_operations.py` script needs a particular MOF file loaded in the repository of the WBEM server that is used for the test. This can be done using the MOF compiler of pywbem: .. code-block:: bash $ mof_compiler -s tests/unittest/pywbem/test.mof To run the unit and function tests in all supported Python environments, the Tox tool can be used. It creates the necessary virtual Python environments and executes `make test` (i.e. the unit and function tests) in each of them. For running Tox, it does not matter which Python environment is currently active, as long as the Python `tox` package is installed in it: .. code-block:: bash $ tox # Run tests on all supported Python versions $ tox -e py313 # Run tests on Python 3.13 .. _`Logging during tests`: Logging during tests -------------------- Most unit tests perform entry and exit logging for the test functions at the DEBUG log level. Some unit tests perform additional logging during the test function, typically also at the DEBUG level. The Python logger used for this is named after the test module. Most functions of pywbem perform logging as well, to the pywbem-specific Python loggers. See :ref:`Pywbem logging overview` for details. The tests that are driven by pytest (e.g. ``make test``, ``make end2end``) can use pytest file logging to control the log level for all these loggers and the destination to a log file. The defaults for pytest file logging are defined in the file ``pytest.ini``. These defaults can be overridden in the command line using pytest options as documented in the ``pytest.ini`` file: .. literalinclude:: ../pytest.ini :language: ini For example: .. code-block:: bash $ TESTOPTS='--log-file-level=DEBUG --log-file=myfile.log' make test .. _`Testing from the source archives on Pypi or GitHub`: Testing from the source archives on Pypi or GitHub -------------------------------------------------- The wheel distribution archives on Pypi (e.g. ``pywbem-1.0.0-py2.py3-none-any.whl``) contain only the files needed to run pywbem, but not the files needed to test it. The source distribution archives on Pypi and GitHub (e.g. ``pywbem-1.0.0.tar.gz``) contain all files that are needed to run and to test pywbem. This allows testing pywbem without having to check out the entire repository, and is convenient for testing e.g. when packaging pywbem into OS-level packages. When installing these source distribution archives, the files needed for running pywbem are installed into the active Python environment, but not the test files. The following commands download the source distribution archive on Pypi for a particular version of pywbem into the current directory and unpack it: .. code-block:: bash $ pip download --no-deps --no-binary :all: pywbem==1.8.0 $ tar -xf pywbem-1.8.0.tar.gz Pywbem, its dependent packages, and packages needed for testing pywbem can be installed with the package extra named "test": .. code-block:: bash $ pip install .[test] When testing pywbem installations in Linux distributions that include pywbem as an OS-level package, the corresponding OS-level packages would instead be installed for these dependent Python packages. The ``test-requirements.txt`` file shows which dependent Python packages are needed for testing pywbem. .. _`Testing installed versions of pywbem`: Testing installed versions of pywbem ------------------------------------ By default, the tests use the pywbem and pywbem_mock modules from the respective directories in the Git repo work directory. Pywbem 0.14.5 introduced a way to test installed versions of the pywbem package. This is useful for example for testing a version of pywbem that has been packaged as an OS-level package. Typically, such a version would be installed into the system Python. Some words of caution: * Testing an installed version of pywbem with test cases from a pywbem repo of a different version can result in failing test cases for several reasons: - If a new version of pywbem has added functionality, its test cases are also extended accordingly. Running such newer test cases against an older installed version of pywbem may fail simply because the installed version does not yet have the added functionality. - Fixes in pywbem or in the test cases may change behavior in a subtle way that causes test cases to fail. - Unit test cases are particularly vulnerable to version mismatches because they test at the module level, including module interfaces that are internal to pywbem and thus can legally change incompatibly between versions. * If the version of the installed pywbem is before 0.14.5, some test cases that compile MOF will be skipped to avoid permission denied errors when ply attempts to re-generate its parsing table files in the pywbem installation directory in case of ply version mismatches. Starting with pywbem 0.14.5, it has tolerance against ply version mismatches. In order to not clutter up the system Python with Python packages needed for running the pywbem tests, the following steps use a virtual Python environment that uses the packages of the system Python. That way, the installed version of pywbem becomes available to the virtual Python environment from the system Python, while any additional packages that are needed but not yet available that way, will be installed into the virtual Python environment. Follow these steps to run pywbem tests against a version of pywbem that is installed into the system Python: 1. Verify that the following commands are available when the system Python is active: .. code-block:: bash $ virtualenv --version # Python virtualenv package $ pip --version 2. Create and activate a virtual Python environment of the intended Python version, that is based on the system Python: .. code-block:: bash $ virtualenv --system-site-packages .virtualenv/test $ source .virtualenv/test/bin/activate The pywbem project is set up so that Git ignores the ``.virtualenv`` directory, so use that directory name for ease of Git handling. 3. Verify that in that virtual Python environment, pywbem comes from the intended installation: .. code-block:: bash $ pip show pywbem 4. Ensure a fresh start of the make process. This should be done whenever switching between the installed version of pywbem and the local directories: .. code-block:: bash $ make clobber 5. Run the pywbem tests with environment variable ``TEST_INSTALLED`` being set: .. code-block:: bash $ TEST_INSTALLED=1 make test This will assume that the pywbem package and any prerequisite Python packages and OS-level packages are already installed. This will also move the current directory (i.e. the repo working directory) to the end of the module search path, so that the installed version of pywbem is used when importing it into the test scripts. Setting ``TEST_INSTALLED=DEBUG`` causes some debug messages to be printed that allow verifying from where the pywbem and pywbem_mock modules are loaded. This also works for the pywbem end2end tests: .. code-block:: bash $ TEST_INSTALLED=1 make end2end Note that tox does not support creating its virtual Python environments based on the system Python, so at this point, tox cannot be used for this approach. .. _`Updating the DMTF MOF Test Schema`: Updating the DMTF MOF Test Schema --------------------------------- Pywbem uses DMTF CIM Schemas in its CI testing. The schema used is stored in the form received from the DMTF in the directory ``tests/schema`` and is expanded and compiled as part of the unit tests. Since the DMTF regularly updates the schema, the pywbem project tries to stay up-to-date with the current schema. At the same time, earlier schemas can be used for testing also by changing the definitions for the dmtf schema to be tested. The schema used for testing can be modified by modifying the test file: :: tests/unittest/utils/dmtf_mof_schema_def.py .. _`Developing Ipython Notebooks`: Developing PyWBEM Ipython Documentation Notebooks ------------------------------------------------- The pywbem developers are using ipython notebooks to demonstrate the use of pywbem. Today we generally have one notebook per operation or group of operations including definition of the operation, references back to the pywbem documentation, and one or more examples (hopefully examples that will actually execute against a wbem server) These can easily be extended or supplemented using a local ipython or jupyter server by: 1. Install ipython or Jupyter software using pip or pip3. The notebook server may be installed as root or within a python virtual environment. For example: .. code-block:: bash $ sudo pip install ipython or $ sudo pip3 install ipython or $ sudo pip install jupyter The notebook server may be installed as root or within a python virtual environment. 2. Start the local ipython/jupyter notebook server in the notebook directory (`pywbem/docs/notebooks`) referencing that directory in the command line call: .. code-block:: bash $ ipython notebook docs/notebooks or $ jupyter notebook docs/notebooks This will start the local ipython/juypter notebook server and on the first page displayed in your web browser all existing pywbem ipython notebooks will be available for editing. New ones can be created using the commands on that ipython server web page. New and changed notebooks must go through the same contribution process as other components of pywbem to be integrated into the github repository. .. _`Git workflow`: Git workflow ------------ * Long-lived branches: - ``master`` - for next functional version - ``stable_M.N`` - for fix stream of released version M.N. * We use topic branches for everything! - Based upon the intended long-lived branch, if no dependencies - Based upon an earlier topic branch, in case of dependencies - It is valid to rebase topic branches and force-push them. * We use pull requests to review the branches. - Use the correct long-lived branch (e.g. ``master`` or ``stable_0.8``) as a merge target! - Review happens as comments on the pull requests. - At least two +1 are required for merging. * GitHub meanwhile offers different ways to merge pull requests. We merge pull requests by creating merge commits, so the single commits of a topic branch remain unchanged, and we see the title line of the pull request in the merge commit message, which is often the only place that tells the issue that was fixed. .. _`Releasing a version`: Releasing a version ------------------- This section describes how to release a version of pywbem to PyPI. It covers all variants of versions that can be released: * Releasing a new major version (Mnew.0.0) based on the master branch * Releasing a new minor version (M.Nnew.0) based on the master branch * Releasing a new update version (M.N.Unew) based on the stable branch of its minor version This description assumes that you are authorized to push to the remote repo at https://github.com/pywbem/pywbem and that the remote repo has the remote name ``origin`` in your local clone. Any commands in the following steps are executed in the main directory of your local clone of the ``pywbem/pywbem`` Git repo. 1. On GitHub, verify open items in milestone ``M.N.U``. Verify that milestone ``M.N.U`` has no open issues or PRs anymore. If there are open PRs or open issues, make a decision for each of those whether or not it should go into version ``M.N.U`` you are about to release. If there are open issues or PRs that should go into this version, abandon the release process. If none of the open issues or PRs should go into this version, change their milestones to a future version, and proceed with the release process. You may need to create the milestone for the future version. 2. Run the Safety tool: .. code-block:: sh make safety If any of the two safety runs fails, fix the safety issues that are reported, in a separate branch/PR. Roll back the PR into any maintained stable branches. 3. Check for any `dependabot alerts `_. If there are any dependabot alerts, fix them in a separate branch/PR. Roll back the PR into any maintained stable branches. 4. Create and push the release branch (replace M,N,U accordingly): .. code-block:: sh VERSION=M.N.U make release_branch This uses the default branch determined from ``VERSION``: For ``M.N.0``, the ``master`` branch is used, otherwise the ``stable_M.N`` branch is used. That covers for all cases except if you want to release a new minor version based on an earlier stable branch. In that case, you need to specify that branch: .. code-block:: sh VERSION=M.N.0 BRANCH=stable_M.N make release_branch This includes the following steps: * create the release branch (``release_M.N.U``), if it does not yet exist * make sure the AUTHORS.md file is up to date * update the change log from the change fragment files, and delete those * commit the changes to the release branch * push the release branch If this command fails, the fix can be committed to the release branch and the command above can be retried. 5. On GitHub, create a Pull Request for branch ``release_M.N.U``. Important: When creating Pull Requests, GitHub by default targets the ``master`` branch. When releasing based on a stable branch, you need to change the target branch of the Pull Request to ``stable_M.N``. Set the milestone of that PR to version ``M.N.U``. This PR should normally be set to be reviewed by at least one of the maintainers. The PR creation will cause the "test" workflow to run. That workflow runs tests for all defined environments, since it discovers by the branch name that this is a PR for a release. 6. On GitHub, once the checks for that Pull Request have succeeded, merge the Pull Request (no review is needed). This automatically deletes the branch on GitHub. If the PR did not succeed, fix the issues. 7. On GitHub, close milestone ``M.N.U``. Verify that the milestone has no open items anymore. If it does have open items, investigate why and fix (probably step 1 was not performed). 8. Publish the package (replace M,N,U accordingly): .. code-block:: sh VERSION=M.N.U make release_publish or (see step 4): .. code-block:: sh VERSION=M.N.0 BRANCH=stable_M.N make release_publish This includes the following steps: * create and push the release tag * clean up the release branch Pushing the release tag will cause the "publish" workflow to run. That workflow builds the package, publishes it on PyPI, creates a release for it on GitHub, and finally creates a new stable branch on GitHub if the master branch was released. 11. Verify the publishing Wait for the "publish" workflow for the new release to have completed: https://github.com/pywbem/pywbem/actions/workflows/publish.yml Then, perform the following verifications: * Verify that the new version is available on PyPI at https://pypi.python.org/pypi/pywbem/ * Verify that the new version has a release on Github at https://github.com/pywbem/pywbem/releases * Verify that the new version has documentation on ReadTheDocs at https://pywbem.readthedocs.io/en/stable/changes.html The new version M.N.U should be automatically active on ReadTheDocs, causing the documentation for the new version to be automatically built and published. If you cannot see the new version after some minutes, log in to https://readthedocs.org/projects/pywbem/versions/ and activate the new version. .. _`Starting a new version`: Starting a new version ---------------------- This section shows the steps for starting development of a new version. This section covers all variants of new versions: * Starting a new major version (Mnew.0.0) based on the master branch * Starting a new minor version (M.Nnew.0) based on the master branch * Starting a new update version (M.N.Unew) based on the stable branch of its minor version This description assumes that you are authorized to push to the remote repo at https://github.com/pywbem/pywbem and that the remote repo has the remote name ``origin`` in your local clone. Any commands in the following steps are executed in the main directory of your local clone of the ``pywbem/pywbem`` Git repo. 1. Create and push the start branch (replace M,N,U accordingly): .. code-block:: sh VERSION=M.N.U make start_branch This uses the default branch determined from ``VERSION``: For ``M.N.0``, the ``master`` branch is used, otherwise the ``stable_M.N`` branch is used. That covers for all cases except if you want to start a new minor version based on an earlier stable branch. In that case, you need to specify that branch: .. code-block:: sh VERSION=M.N.0 BRANCH=stable_M.N make start_branch This includes the following steps: * create the start branch (``start_M.N.U``), if it does not yet exist * create a dummy change * commit and push the start branch (``start_M.N.U``) 2. On GitHub, create a milestone for the new version ``M.N.U``. You can create a milestone in GitHub via Issues -> Milestones -> New Milestone. 3. On GitHub, create a Pull Request for branch ``start_M.N.U``. Important: When creating Pull Requests, GitHub by default targets the ``master`` branch. When starting a version based on a stable branch, you need to change the target branch of the Pull Request to ``stable_M.N``. No review is needed for this PR. Set the milestone of that PR to the new version ``M.N.U``. 4. On GitHub, go through all open issues and pull requests that still have milestones for previous releases set, and either set them to the new milestone, or to have no milestone. Note that when the release process has been performed as described, there should not be any such issues or pull requests anymore. So this step here is just an additional safeguard. 5. On GitHub, once the checks for the Pull Request for branch ``start_M.N.U`` have succeeded, merge the Pull Request (no review is needed). This automatically deletes the branch on GitHub. 6. Update and clean up the local repo (replace M,N,U accordingly): .. code-block:: sh VERSION=M.N.U make start_tag or (see step 1): .. code-block:: sh VERSION=M.N.0 BRANCH=stable_M.N make start_tag This includes the following steps: * checkout and pull the branch that was started (``master`` or ``stable_M.N``) * delete the start branch (``start_M.N.U``) locally and remotely * create and push the start tag (``M.N.Ua0``) .. _`Contributing`: Contributing ------------ Third party contributions to this project are welcome! In order to contribute, create a `Git pull request`_, considering this: .. _Git pull request: https://help.github.com/articles/using-pull-requests/ * Test is required. * Each commit should only contain one "logical" change. * A "logical" change should be put into one commit, and not split over multiple commits. * Large new features should be split into stages. * The commit message should not only summarize what you have done, but explain why the change is useful. * The commit message must follow the format explained below. What comprises a "logical" change is subject to sound judgement. Sometimes, it makes sense to produce a set of commits for a feature (even if not large). For example, a first commit may introduce a (presumably) compatible API change without exploitation of that feature. With only this commit applied, it should be demonstrable that everything is still working as before. The next commit may be the exploitation of the feature in other components. For further discussion of good and bad practices regarding commits, see: * `OpenStack Git Commit Good Practice`_ * `How to Get Your Change Into the Linux Kernel`_ .. _OpenStack Git Commit Good Practice: https://wiki.openstack.org/wiki/GitCommitMessages .. _How to Get Your Change Into the Linux Kernel: https://www.kernel.org/doc/Documentation/SubmittingPatches .. _`Creating and submitting a change to pywbem`: Creating and submitting a change to pywbem ------------------------------------------ All changes to pywbem are made through Github with PRs created on topic branches and merged with the current master after successful group review. To make a change, create a topic branch. You can assume that you are the only one using that branch, so force-pushes to that branch and rebasing that branch is fine. When you are ready to push your change, describe the change for users of the package in a change fragment file. That is a small file in RST format with just a single change. For more background, read the [towncrier concept](https://towncrier.readthedocs.io/en/stable/markdown.html) (which uses Markdown format in that description and calls these files 'news fragment files'). To create a change fragment file, execute: For changes that have a corresponding issue: .. code-block:: sh towncrier create ..rst --edit For changes that have no corresponding issue: .. code-block:: sh towncrier create noissue...rst --edit For changes where you do not want to create a change log entry: .. code-block:: sh towncrier create noissue..notshown.rst --edit # The file content will be ignored - it can also be empty where: * ```` - The issue number of the issue that is addressed by the change. If the change addresses more than one issue, copy the new change fragment file after its content has been edited, using the other issue number in the file name. It is important that the file content is exactly the same, so that towncrier can create a single change log entry from the two (or more) files. If the change has no related issue, use the ``noissue...rst`` file name format, where ```` is any number that results in a file name that does not yet exist in the ``changes`` directory. * ```` - The type of the change, using one of the following values: - ``incompatible`` - An incompatible change. This will show up in the "Incompatible Changes" section of the change log. The text should include a description of the incompatibility from a user perspective and if possible, how to mitigate the change or what replacement functionality can be used instead. - ``deprecation`` - An externally visible functionality is being deprecated in this release. This will show up in the "Deprecations" section of the change log. The deprecated functionality still works in this release, but may go away in a future release. If there is a replacement functionality, the text should mention it. - ``fix`` - A bug fix in the code, documentation or development environment. This will show up in the "Bug fixes" section of the change log. - ``feature`` - A feature or enhancement in the code, documentation or development environment. This will show up in the "Enhancements" section of the change log. - ``cleanup`` - A cleanup in the code, documentation or development environment, that does not fix a bug and is not an enhanced functionality. This will show up in the "Cleanup" section of the change log. - ``notshown`` - The change will not be shown in the change log. This command will create a new change fragment file in the ``changes`` directory and will bring up your editor (usually vim). If your change does multiple things of different types listed above, create a separate change fragment file for each type. If you need to modify an existing change log entry as part of your change, edit the existing corresponding change fragment file. Add the new or changed change fragment file(s) to your commit. The test workflow running on your Pull Request will check whether your change adds or modifies change fragment files. You can review how your changes will show up in the final change log for the upcoming release by running: .. code-block:: sh towncrier build --draft Always make sure that your pushed branch has either just one commit, or if you do multiple things, one commit for each logical change. What is not OK is to submit for review a PR with the multiple commits it took you to get to the final result for the change. .. _`Core Development Team`: Core Development Team --------------------- Anyone can contribute to pywbem via pull requests as described in the previous section. The pywbem project has a core development team that holds regular web conferences and that is using Slack for offline communication, on the Slack workspace: https://pywbem.slack.com. The web conference and the Slack workspace are by invitation, and if you want to participate in the core team, please `open an issue `_ to let us know.