The idea was that in order to continuously deliver whilst keeping Semantic Versioning, we needed an automation step. This aims to do that, it's not perfect, but it's better than tagging manually or worse, forgetting to tag. It was mainly designed to work with GitLab, it can be quite easily extended to also work with other tools.
In order to make this work you need a user that has access to your repository and have it's SSH private key in a environment variable, as well as the public key added to its account.
Bumping the patch version is quite simple, however how could we decide to bump the minor or major? Quite frankly
it would be quite difficult to guess it based on the code changes, it would be a cool project for another day though!
For now the only way to automatically bump the version is to include a keyword in the last commit message. The options
are to use #major to bump the major and #minor to bump minor. This keyword can be anywhere in the last commit
message.
Below is an example of using it in GitLab CI. First we inject the SSH key of the user with sufficient rights to push to a repository.
Then we call the auto-semver command to do its thing. This step will only be triggered on the branch main assuming
that this is the branch you merge your development branch to.
Given that auto-semver actually pushes to your repository it will trigger a new pipeline, steps such as deploy are
then activated but not publish.
Ideally you should change latest to a pinned version.
auto-semver:
stage: publish
image: freddegier/gitlab-auto-semver:latest
script:
- mkdir -p ~/.ssh && echo "$SEMVER_SSH_KEY" > ~/.ssh/id_rsa && chmod -R 700 ~/.ssh
- auto-semver
rules:
- if: $CI_COMMIT_REF_NAME == 'main'
deploy:
stage: deploy
script:
- echo "Deploying"
rules:
- if: $CI_COMMIT_TAGI was inspired by this article to create this tool. https://threedots.tech/post/automatic-semantic-versioning-in-gitlab-ci/