How to use semversioner in your release flow

Recently, we’ve marked a few of our branches in a project as protected which results in local changes to the branch not being push-able to the remote branch.

This also means that when you are using git flow, you need to come up with a way to push your changes into a pull request and merge in the repo’s frontend.

This lead me to believe that my readers would appreciate an article about how we use semantic versioning in our release flow of python projects.

Semversioner

semversioner is a tool that allows you to do updates to your code, document them and have changelogs and version numbers generated automatically on release.

This is quite easy when you add a short method to your .bashrc or .zshrc:

def sem() {
  TYPE=$1  # patch, minor, major
  shift
  DESCRIPTION="$@"

  # Test git changes
  pre-commit run
  if [ $? -ne 0 ]; then
    return $?
  fi

  semversioner add-change --type ${TYPE} --description "${DESCRIPTION}"
  git add .changes

  git commit -m "${DESCRIPTION}"
}

After making a change to your source code, all you need to do is:

git add list-modified-files
sem patch|minor|major "Description of the change"

to have semversioner automatically deal with the change internally. All the changes pending for the next release will be located in .changes/next-release and semversioner will also deal with your release numbering.

Makefile

Together with the snippet above, we usually add the following to our Makefile to make effective use of git flow:

semver: semver-release semver-updates

semver-release:
	-semversioner release

semver-updates:
	semversioner changelog > CHANGELOG.md
	$(eval CURRENT_VERSION = $(shell semversioner current-version))
	sed -i "s/^__version__.*/__version__ = \"$(CURRENT_VERSION)\"/" setup.py
	-git add .changes setup.py CHANGELOG.md
	-git commit -m "semverioner release updates" --no-verify
	-git flow release start -F $(CURRENT_VERSION)
	-git flow release publish $(CURRENT_VERSION)
	git flow release finish -F -s -m "Release $(CURRENT_VERSION)" --noff-master --pushtag --pushdevelop --nopushproduction --keepremote --nokeeplocal $(CURRENT_VERSION)

release: semver clean build check

The make release does the following for your:

1. Create a new changelog and store it in CHANGELOG.md.
2. Obtain a new version number depending on the pending changes.
3. Update your setup.py to include the new __version__.
4. Commit the latest changes of setup.py and CHANGELOG.md.
5. Start a release flow with the new version number
6. Publish the new release branch
7. Finish the release

No, as you can see, the finishing of the release uses a bunch of options --noff-master --pushtag --pushdevelop --nopushproduction --keepremote --nokeeplocal. The sole purpose of those is to do be able to finish the release locally, but not push it the remote as we have a protected master branch.

After that, all that’s left to do is to go to your repository management and manually create a pull/merge request and have that merged into your production branch.