Automated build for Bitbucket Pipelines Docker image
Setup for local development
- Make sure your system has Anaconda installed
conda env createin the root folder of this repo, this will create the Python environment for this project as specified in the environment.yml file
source activate deeplearningin the root folder of this repo, this makes your Python environment point to the environment you created in the previous step
Description of build pipeline setup
- We use Bitbucket Pipelines to automatically lint our code and execute all tests upon each push to the repository.
- The pipeline depends on our own custom Docker image (mathiaseitz/deeplearning) defined by the Dockerfile at the root of this repository. We use this custom Docker image only to enforce that all required conda packages are already available in conda's cache in order to speed up the build.
- Whenever the Dockerfile on the master branch changes, an automated build for the Docker image on DockerHub is triggered automatically and the resulting image is pushed to DockerHub as mathiaseitz/deeplearning:latest. Note that changes to the Dockerfile on other branches than master do not trigger a rebuild of the image.
Docker Pull Command