Docker Project Boilerplates
Slashdeploy is passionate about continuous deployment. Continuous deployment is the best way to build quality software. Achieving continuous deployment is no small task. Reaching the goal requires attention to detail at every step in the software process. Software starts in the development phase. The outcome here is simple: create software that works as expected and that there are no regressions.
Test Driven Development (or TDD for short) is the best way to achieve this outcome. The highest quality software is created by applying TDD to classes, functions, modules, and the other software artifacts. Most teams cover the first set of things but drop out on the second. How many times have you deployed code to production/staging/etc and seen errors because of some missing configuration or something that’s conditional to that stage? How many times have you seen your software fail because it was never actually started before going to production? Each of these failures (and others) are just like any other bug. They can be provably solved with regression tests.
This is where Docker and the associated boilerplate code come into the picture. Working with containers makes encapsulating the development and testing process easy for all types of testing. This is why we’re sharing these project boilerplate repos. They provide an excellent structure to build a production ready project. The boilerplates are build on
docker-compose. Each boilerplate includes:
- Docker based dependency management
- TDD setup
- Code linting
- An editorconfig file configured for best practices
- Dockerfile configured to run a web server
- Blackbox & whitebox tests for the web server
make test-ci: Run a full battery of tests
make test-smoke: Run a smoke test against production process
They generally follow the same structure.
docker is used to generate dependencies so they can be committed to source control. This removes upstream dependencies on upstream package hosts at build time.
docker-compose manages containers for various other targets. One container runs the process (the boilerplates use a simple web server as a common case). Another container is used for quick tests (for interpreted languages which don’t require a separate compile step). Another container is used to smoke test the application container.
All boilerplates are free to use and customize. Simply download or clone the repo and follow the
TODO’s inside. The repositories are continually updated to follow best practices learned from using Docker to ship the highest quality software possible.
Feedback and pull requests welcome as always. Good luck out there and happy shipping!