Branching and Releases
Essential knowledgeCore knowledge to understand Fluent Commerce
Changed on:
9 Nov 2026
Overview
[Warning: empty required content area]Key points
[Warning: empty required content area]Branch Model
| Branch | Purpose | Pipeline behavior |
`feature/*`, `fix/*` | One change each, short-lived | Build and test only |
`dev` | Integration branch | Merged through a merge queue; each merge builds, tests and deploys to the dev account |
`main` | Release candidates | Each merge builds, tests and tags the commit with the build number |
`release/<tag>` | One per release, cut from a tag | Builds and deploys to UAT on manual trigger, then to production on a separate manual trigger |
Guidelines
Keep feature branches short. Workflows are large JSON documents and merge badly. A branch that lives for weeks will conflict with every other workflow change, so merge small changes often.Use a merge queue for`dev` and `main`. The queue rebuilds each change on top of the latest target branch before merging. Without it, two PRs that each pass on their own can break together, and the first anyone hears of it is a broken dev account.Deploy only from shared branches. Feature branches build and test but never deploy, so the dev account always reflects `dev`, the one branch that everyone's work has been merged into.One person deploys to an environment at a time. Bitbucket deployment environments serialize deploys to the same environment. Leave this on.Tagging and release branches
- A merge to
`main`tags the commit with the build number. The tag is also the build identifier that`build-module.sh`appends to the module version, so the tag, the module version and the deployed version all match. - The tag pipeline cuts
`release/<tag>`from that commit. - The release branch pipeline builds the archive and holds it at a manual UAT deploy, then at a manual production deploy.
`dev` first and are then cherry-picked onto the release branch. Never fix only the release branch, or the next release will bring the defect back.