CI with TravisCI

Objectives

• Learn what Continuous Integration is
• Set up Travis CI on a GitHub repository
• Configure Travis CI for multiple test builds

Duration: 45 minutes

What is Continuous Integration?

• Changes are continually integrated into main branch of software
• This is enabled by running suite of tests for every change made to the code base
• Variety of GitHub-integrated services
  • Travis-CI: Linux, OSX
  • Appveyor: Windows, Linux (new)
  • CircleCI: Linux, OSX (docker containers)
• These services can run on every Pull Request–make sure the PR doesn’t break anything
• Pull Requests can be configured to require checks to pass before merge
• We’ll be setting up Travis-CI

First steps: Initial Travis config for a single Python version (3.6)

• Sign into https://travis-ci.com with GitHub
• Configure Travis to run for our repository
• Make a local branch
• Create .travis.yml

language: python
sudo: false

# Only run on master since we'll be using branches for our PR
branches:
  only:
    - master

# What Python version to use
python:
  - 3.6

# How do we install?
install:
  - pip install .

# How do we test?
script:
  - pytest

• Individual sections (install, script) become shell scripts
• Commit and push to remote repository; make Pull Request
• See what the PR looks like, the status checks
• Break Time

More modifications

• Run on Python 2.7, 3.5, 3.6
• Install dependencies (before_install)
• Use environment variable to install miniconda

matrix:
  include:
    - python: 3.6
      env: MINICONDA="y"

before_install:
  - |
    if [[ $MINICONDA == "y" ]]; then
      wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh -O miniconda.sh
      bash miniconda.sh -b -p $MINICONDA_DIR
      export PATH="$MINICONDA_DIR/bin:$PATH"
      hash -r
      conda config --set always_yes yes --set changeps1 no
      conda update -q conda
      conda info -a
      conda install numpy
    fi

More things possible

• https://docs.travis-ci.com
• Deploy build products to S3
• Push to docs to GitHub Pages
• Deploy built packages to PyPI
• Push code coverage results to another service
• Send notification to another webhook