Developing#

Development of Barista requires a couple of dependencies.

Crystal
Python (for docs)

After installing the dependencies

git clone https://gitlab.com/skinnyjames/barista && cd barista
shards install to install project dependencies
./spec.sh to run the specs

Warning

The specs include integration tests against a Barista::Behaviors::Brew::Project, which spins up and tears down persistent processes.

If working on Brew::Project, there is a chance of spinning up unterminated background processes. A docker-compose file is provided to assist in testing without worrying about doing this.

docker-compose build && docker-compose run spec
If you want to run locally, you can exclude the brew specs: crystal spec --tag="~brew"

Considerations#

The library structure encapsulates behaviors in their own directories

/src
  /barista # for base level features
    /behaviors
      /omnibus # for omnibus features
      /software # for general software features

Info

Since Barista is class/interface based, it is easy to expose new behaviors by including modules into a Project/Task

If a feature seems specific to the usage of Barista rather than the interface of it's Behaviors, consider keeping it in the consuming library.

Spec considerations#

The Barista specs provide helpers for integration testing the e2e flow. This includes

An HTTP server to publish and fetch cached artifacts and source archives.
Fixture paths for configuring Omnibus projects safely
Predefined cache callbacks to use when testing Omnibus projects
Ability to intercept output from an Orchestrated run

Documentation#

The documentation is generated with mkdocs. It requires python, and installs a virtual env in the project.

To watch/serve the docs, run make serve.
To buid the docs, run make build