Examples of continuous integration workflow using git

Question

Can anyone provide a rough outline of their git workflow that complies with continuous integration.
E.g. How do you branch? Do you fast forward commits to the master branch?

I am primarily working with Rails as well as client and server side Javascript. If anyone can recommend a solid CI technology that's compatible with those, that'd be great. I've looked into Jenkins but would like to check out other good alternatives.

To put some context into this, I am planning on transitioning from working as a single developer into working as part of the team. I'd like to start standardizing my own personal workflow so that I can onboard new devs quickly.

• Benjamin Bannier: Did you look how this is done commonly? Set up e.g. jenkins with gerrit and play with it.
• jcmeloni: Have you tried something that didn't work or didn't seem right, that you could describe & add to your question?

Answer 1

A simple workflow.

1. Central Git Repo used by the CI server
2. People push to a specific branch on that Repo (say Master)
3. CI Server detects changes in master branch or runs a timed build
4. CI Server runs unit tests
5. Depending on your rules, the CI server can then tag the repository, and a myriad of other things.

There are countless ways to do this, do what fits your workflow the best.

• Andrew Lauer Barinov: Do you use feature branches? How do you ensure a clean commit history on the master branch? Or are everyone's commits aggregated?
• sylvanaar: We do not use feature branches to trigger builds, but we do use them. We don't worry that much about clean commit history, we'd prefer it was accurate

Answer 2

• Work
• Commit to your local repo
• Push (to master repo)
• CI detects new revision
• CI checks out new revision
• CI builds and tests new revision
• CI updates Changelog (commit and push)
• CI creates a version file and then tags master repo
• CI packages build
• CI pushes package to artifact repo (in our case a YUM repo)
• rinse, repeat

We don't work of branches. We always commit to trunk, but then again we also do ATDD to mitigate the common issues like instability.

• Andrew Lauer Barinov: Is the CI server you're using a custom one or something that's available off the shelf?
• iteratingself: @Andrew Barinov: We are currently using Hudson, but most of the logic is in our Makefiles.