scikit_image-0.24.0.tar.gz

How to make a new release of ``skimage`` ======================================== While following this guide, note down all the times that you need to consult a previous release manager, or that you find an instruction unclear. You will, of course, make a PR to update these notes after you are done with the release! ;-) Before you start, make sure you have all the required write permissions (if not, you will need to ask an owner to grant you access), specifically to: - https://pypi.org/project/scikit-image/ - https://github.com/scikit-image/skimage-web - Check ``TODO.txt`` for any outstanding tasks. We use a variant of "semantic versioning", where version numbers are classified as v<major>.<minor>.<patch>. By default, releases are made from the main branch as part of a linear release history and, as described below, are triggered by pushing a git tag to the scikit-image repository on github. If a patch release is required, a branch can be created from the appropriate point in main and the following instructions are still apt. Example `version number` - 0.23.0rc0.dev0 # development version for 0.23.0 first release candidate - 0.23.0rc0 # 0.23.0 first release candidate - 0.23.0rc1.dev0 # development version for 0.23.0 second release candidate - 0.23.0 # 0.23.0 release - 0.24.0rc0.dev0 # development version for 0.24.0 first release candidate ## Process - To debug building and testing wheels via CI, push to a branch named `maintenance/anything`. Once everything works as intended, remove those branches and proceed. - Set release variables: export VERSION=<version number> export PREVIOUS=<previous version number> export ORG="scikit-image" export REPO="scikit-image" If this is a prerelease: export NOTES="doc/source/release_notes/release_dev.rst" If this is release: export NOTES="doc/source/release_notes/release_${VERSION}.rst" git rm doc/source/release_notes/release_dev.rst - Autogenerate release notes: changelist ${ORG}/${REPO} v${PREVIOUS} main --version ${VERSION} --out ${NOTES} --format rst changelist ${ORG}/${REPO} v${PREVIOUS} main --version ${VERSION} --out ${VERSION}.md - If you want, you can review the release notes: - Scan the PR titles for highlights and mention these in the relevant sections of the notes. Ideally, all changed API objects are mentioned by name, e.g. a new parameter or a deprecated function. - Check for duplicate names in the automatically generated list of contributors and reviewers. - Append the release date to the title similar to previous iterations. - If this is a release: - (Only needed if you create a new branch) update the file ``skimage/data/_fetchers.py`` to point the pooch repository to the new branch instead of main during testing. Look for the parameter ``version_dev="main",`` for ``pooch.create`` and change it to the newly created branch name. - Edit ``doc/source/_static/docversions.js`` and ``doc/source/_static/version_switcher.json`` in order to add the release, move the key value pair `"preferred": true` to the most recent stable version, and commit. git add skimage/data/_fetchers.py doc/source/_static/docversions.js doc/source/_static/version_switcher.json - Update ``__version__`` in ``skimage/__init__.py``. - Commit changes: git add skimage/__init__.py ${NOTES} git commit -m "Designate ${VERSION} release" - Tag the release in git: git tag -s v${VERSION} -m "signed ${VERSION} tag" (If you do not have a GPG key, follow the tutorial to set it up: https://help.github.com/articles/signing-commits-with-gpg/) - Push the new meta-data to github: git push --tags origin main where `origin` is the name of the ``github.com:scikit-image/scikit-image`` repository - Create release from tag - go to https://github.com/scikit-image/scikit-image/releases/new?tag=v${VERSION} - add v${VERSION} for the `Release title` - paste contents (or upload) of ${VERSION}.md in the `Describe this release section` - if pre-release check the box labelled `Set as a pre-release` - Update https://github.com/scikit-image/scikit-image/milestones: - close old milestone - ensure new milestone exists (perhaps setting due date) - Update ``__version__`` in ``skimage/__init__.py``. - Commit changes: git add skimage/__init__.py git commit -m 'Bump version' git push origin main - Update documentation on the web: The documentation is kept in a separate repo: scikit-image/docs - Wait for the CI service to deploy to GitHub Pages - Sync your branch with the remote repo: ``git pull``. - Update the docs git switch gh-pages git branch -C gh-pages gh-pages-backup cp -a dev ../. git reset --hard <commit from last release> git rm -rf dev cp -a ../dev . git add dev cp -a ../dev <release name, e.g., 0.23.x> git add <release name, e.g., 0.23.x> rm stable ln -s <release name, e.g., 0.23.x> stable git add stable git commit -m "Add <release name, e.g., 0.23.x>" git push -f origin gh-pages # force push---be careful! - Update the web frontpage: The webpage is kept in a separate repo: scikit-image/skimage-web - Highlight new release in the news section git pull # Edit the ``News`` section of ``index.rst`` git add index.rst git commit -m "Highlight ${VERSION} release" git push ---- - Post release notes on user & dev forums, blog, Twitter, etc. - https://forum.image.sc/tag/scikit-image - https://discuss.scientific-python.org/c/contributor/skimage - scipy-user@python.org - scikit-learn@python.org - Update the version and the release date on Wikipedia https://en.wikipedia.org/wiki/Scikit-image - Make a PR to scikit-image with any updates you might have to these notes - If making a patch release (v<major>.<minor>.<patch>), forward-port the release notes to the main branch and make a PR. Conda-forge ----------- **Note**: conda-forge now has an automated bot who makes the below PR for you. Now all you have to do is to look at pull requests at https://github.com/conda-forge/scikit-image-feedstock/pulls and check for a new one for this version. Wait for all the continuous integration checks to go green, then merge. The manual instructions remain below in case the bot fails for some reason. A scikit-image build recipe resides at https://github.com/conda-forge/scikit-image-feedstock. You should update it to point to the most recent release. You can do this by following these steps: - Fork the repository at https://github.com/conda-forge/scikit-image-feedstock, and clone it to your machine. - Sprout a new branch, e.g. ``v<major>.<minor>``. - Find out the SHA256 hash of the source distribution. You can find this at https://pypi.org/project/scikit-image/, or use the following commands: - ``sha256sum path/to/scikit-image-*.tar.gz`` (Linux) - ``shasum -a 256 dist/scikit-image-*.tar.gz`` (macOS) - ``CertUtil -hashfile dist\scikit-image-*.tar.gz SHA256`` (Windows) - Edit the file ``recipe/meta.yaml``: - Update the version number on the first line. - Update the SHA256 value on line 9. - If necessary, reset the build number to 0. (line 12) - Update any requirements in the appropriate sections (build or run). Note: don't remove ``numpy x.x``. This tells conda-smithy, conda-forge's build system, that the library must be linked against NumPy at build time. - Commit these changes. - Update the infrastructure around the recipe with ``conda-smithy``: * Install conda-smithy either with conda or pip * Run ``conda-smithy rerender`` in the root of the repository, and commit the changes. - Push to your fork, and submit a pull request to the upstream repo. Debian ------ The below instructions remain here for completeness. However, the De