Rapid development workflow - tests with Docker Compose¶
Recently, I did a presentation about my development workflow for back-end applications.
English version of the video:
Polish version of the video:
Slides are here. The sample code is here.
Contents summary¶
Docker Compose setup with the sample HTTP/REST app (written in Python) and a SQL database
High-level tests using containers.
Reloading the app container on code file changes.
Running tests on code file changes.
Makefiles to describe common development tasks.
CI setup based on local tests.
Abstract¶
Containers have revolutionized many aspects of software development. In particular - testing.
Testing applications with the same data bases (e.g. Postgres, Redis), queues (e.g. Kafka), etc. that they’ll rely on in production grants a substantially higher degree of confidence in the software, than tests with mocks or in-memory fakes of real connectors.
In this presentation I’ll show how I use docker-compose for local running and testing of the app being developed. How to adjust the code and set the configuration to make your work faster, ease on-boarding of new team members, and, above all, to increase the quality. The presented techniques emerged from working on a couple projects across the last 5 years.
The example code will be a Python back-end app, but the techniques will apply in other programming languages, and not only for back-end applications.
The presentation assumes at least a passing familiarity with Docker and docker-compose.
The subjects covered will include:
separation of unit, integrated, and functional tests
organizing the project so that a local instance of the app can be run with two commands:
git clone && make runnot resetting the test environment between tests
local and CI test parity
debugging code running in containers
using “Docker mounts” to enable fast application reloading while editing code
changes to the production code that make testing easier
