This repository contains a sapling (skeleton) for building a Java Module using Maven that can be uploaded to Maven.
Modules uploaded to Maven can be used by other modules or Java applications using the Java Application sapling.
-
Click "Use this template", and name your new creation.
-
Clone the resulting project to your computer
cd ~/code git clone git@github.com:your-org/my-module.git
-
Remove this
README.md
file, and replace it with your own -
Edit the
pom.xml
- Update the
groupId
andartifactId
sections - Add your dependencies
- Update the
-
Add your code
- Create a folder with the same name as your module
- Run
mvn compile
to sync your local dependencies
Dependencies installed, but not listed in your lock file will not be installed in CircleCi!
-
Commit your code
- See the section on Workflow for Git workflow processes and best practices
-
Add your project to CircleCi
-
Party!
See maven here and see java here
If you've created a project called yodelling_coach
and had just freshly minted release 0.9.9
,
your artifact would be called yodelling_coach
version 0.9.9
.
Using this package in another pom.xml
would look like:
...
<dependency>
<groupId>com.nxtlytics.team</groupId>
<artifactId>com.nxtlytics.team.yodelling_coach</artifactId>
<version>0.0.9</version>
</dependency>
...
If you've created a project called awesome_sauce
and were working off snapshot 0.7.4-SNAPSHOT
in the develop
branch,
your artifact would be called awesome_sauce
version 0.7.4-SNAPSHOT
.
This would be considered a prerelease module by Maven, so to reference it in a pom.xml
of another project, you would need to
reference it with it's exact name.
...
<dependency>
<groupId>com.nxtlytics.team</groupId>
<artifactId>com.nxtlytics.team.awesome_sauce</artifactId>
<version>0.7.4-SNAPSHOT</version>
</dependency>
...
The initial Git bootstrap workflow is:
- Clone the sapling
- Add your code in any state (working or not)
- Commit your code
- Update
pom.xml
to reflect the version of the tag you are about to create. - Tag the
master
branch with a semver (0.0.0
for initial commit is suggested)
CircleCi will build a 0.0.0
version (based on the version declared in pom.xml
) of your library which depending on the state of your initial code commit may be all you need.
Once you've decided you need to make updates or fix bugs, you've got two options:
-
feature-*
branches ->develop
branch ->master
branch
Have multiple contributors? Want separation of duties and potentially a benevolent merge-dictator?
Use this method if so. -
develop
branch ->master
branch
Just one person working on the project? No time for process, just need to get builds now?
Alright, alright. I hear ya. Use this method instead.
Short and sweet version: Don't commit directly to master
branch.
Setting up feature branches is easy.
After you've got the initial bootstrap done, create a develop
branch and commit it.
Then create a feature branch (does not need to be named anything in specific, just keep it short and descriptive) and work from there.
As you work, make commits to the feature branch and push them to Github as you see fit. CircleCi will run any unit tests and help you spot any issues along the way.
Once you've completed your feature, send a Pull Request from Github to merge into the develop
branch.
The project admin (maybe you?) will review this PR and decide to merge it if they see fit.
Once the PR is merged into the develop
branch, the project admin will usually allow CircleCi to build a
prerelease module that can be used in an integration test for any projects that require this module.
After all integration tests are completed, the project admin will merge develop
into master
and cut a new git tag.
Don't have time for the whole feature branch method? That's fine. You can shortcut it and migrate to feature branches later if required.
After the initial bootstrap, simply create and work on your develop
branch. Commit directly to the develop
branch and
merge (PR or manually) into master
when you're ready to create a new production-ready version by tagging off master
.