- trunk-based development
As much as possible, everything that can be done on CI server should be doable on any machine. Testing, publishing, deploying etc... In practice the action can be controlled with security credentials, eg. deploying to prod. CI tool specific information should be kept to a minimum. fundamentally discrete steps should be replicable in a "local" machine.
Examples
- github action config uses
make
commands already in the repository - naming of artefacts according to CI source
changelog.md will be treated the same as any other information. there will be no manual steps to combine changelog. Conflicts are meant to be resolved.