Use with GitHub Actions

GitHub Actions is currently en vogue and popular due to fairly unlimited build resources, along with the stunning own-goal scored by the previous incumbent. Usage with run.sh is easy. Just commit any file ending in .yml or .yaml in a directory .github/workflows within your GitHub repo---and you are all set. No additional registration is needed.

Below is an (adapted) example .github/workflows/ci.yaml from the RcppInt64 repository running a simple matrix build over Linux and macOS---and over two containers. The containers offer, respectively, use of R-devel and a possibly slightly faster r2u setup (as times for this can vary at GitHub). Note that the matrix build for any one of these can be turned off by placing a simple comment # character in front as can be seen in many example YAML files. Also note that this version now use the GitHub Action for r-ci (instead of the standard curl use shown in the other examples) which also includes the bootstrap step. Lastly, one can uncomment the dumping of logs in case of failures, and enable coverage as some of my repos do. (Note that you need to set a repository secret under the name `CODECOV_TOKEN` with the supplied token string.) For convenience this ci.yaml can also be downloaded. 


# Run CI for R using https://eddelbuettel.github.io/r-ci/

name: ci

on:
  push:
  pull_request:

env:
  _R_CHECK_FORCE_SUGGESTS_: "false"

jobs:
  ci:
    strategy:
      matrix:
        include:
          - { name: container, os: ubuntu-latest, container: rocker/r2u4ci }
          - { name: r-devel,   os: ubuntu-latest, container: rocker/drd }
          - { name: macos,     os: macos-latest }
          - { name: ubuntu,    os: ubuntu-latest }

    runs-on: ${{ matrix.os }}
    container: ${{ matrix.container }}

    steps:
      - name: Checkout
        uses: actions/checkout@v6

      - name: Setup
        uses: eddelbuettel/github-actions/r-ci@master

      - name: Dependencies
        run: ./run.sh install_all

      - name: Test
        run: ./run.sh run_tests

      #- name: Logs
      #  run: ./run.sh dump_logs
      #  if: failure()
        
      #- name: Coverage
      #  if: ${{ matrix.os == 'ubuntu-latest' }}
      #  run: ./run.sh coverage
      #  env:
      #    CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}

Use with Travis CI

Travis CI continues to be useful for those with remaining free credits, or a paid plan, so very basic use by Open Source project on Linux should be unaffected.

The following code examples is an actual .travis.yml file, also from the dang repository used above. It runs a simple matrix build over macOS and Linux. 


# Run CI for R using https://eddelbuettel.github.io/r-ci/

language: c
sudo: required
dist: focal

jobs:
  include:
    - name: linux
      os: linux
    - name: macOS
      os: osx

## Two core settings: BSPM for binary installation where possible, and no
## installation of 'Suggests:' to keep things lighter
env:
  global:
    - _R_CHECK_FORCE_SUGGESTS_="false"
      USE_BSPM="true"
    
before_install:
  - curl -OLs https://eddelbuettel.github.io/r-ci/run.sh
  - chmod 0755 run.sh
  - ./run.sh bootstrap

install:
  - ./run.sh install_all

script:
  - ./run.sh run_tests

after_failure:
  - ./run.sh dump_logs
 
#after_success:
#  - ./run.sh coverage

notifications:
  email:
    on_success: change
    on_failure: change

Use with Azure Pipelines

Azure Pipelines are an alternative worth considering as they also offer fairly generous usage credits. The example below is once again from the dang repository and is provided in its file .ci/ci.yaml (but in its git repo on Azure; these pipelines can also access other repos for example at GitHub). As in the other examples, a matrix over Linux and macOS is setup


# Evolved from starter pipeline via r-azure-pipelines repo

trigger:
- master

variables:
- name: R_LIBS_USER
  value: '$(Agent.BuildDirectory)/R/library'
- name: CRAN
  value: 'https://cloud.r-project.org'
- name: _R_CHECK_FORCE_SUGGESTS_
  value: false
- name: USE_BSPM
  value: true

strategy:
  matrix:
    linux:
      imageName: "ubuntu-latest"
    macos:
      imageName: "macos-latest"
  maxParallel: 2
  
steps: 
  - bash: curl -OLs https://eddelbuettel.github.io/r-ci/run.sh
    displayName: download
  - bash: chmod 0755 run.sh
    displayName: mode
  - bash: ./run.sh bootstrap
    displayName: bootstrap
  - bash: ./run.sh install_deps
    displayName: deps
  - bash: ./run.sh run_tests
    displayName: tests

Use with CircleCI

CircleCI offers another alternative with a very polish browser interface and generous credits. The example below is again derived from the dang repository. It uses the r2u4ci container, fetches `run.sh` and uses it. The file has to be stored as `.circleci/config.yaml` in a repository registered at CircleCI. 


version: 2.1

jobs:
  ci:
    docker:
      - image: rocker/r2u4ci

    steps:
      - checkout
      - run:
          name: Download
          command: curl -OLs https://eddelbuettel.github.io/r-ci/run.sh
      - run:
          name: Mode
          command: chmod 0755 run.sh
      - run:
          name: Bootstrap
          command: ./run.sh bootstrap
      - run:
          name: Dependencies
          command: ./run.sh install_deps
      - run:
          name: Test
          # set _R_CHECK_FORCE_SUGGESTS_="false" as needed
          command: ./run.sh run_tests
         
workflows:
  ci-workflow: 
    jobs:
      - ci

Use with Docker

A key advantage of the simple run.sh script is its versatility: just drop it into a Docker session to test a local repo!

The following short video, part of post #32: Portable Continuous Integration using r-ci in the r^4 series of r^4 posts, demonstrates Docker use for CI with the dang repository used above.

The commands used in the video are, essentially, just these: 

# launch docker using 'r-bspm:20.04' from rocker
docker run --rm -ti -v ${PWD}:/work -w /work rocker/r-bspm:20.04

# fetch the script
wget https://eddelbuettel.github.io/r-ci/run.sh && chmod 0755 run.sh

# bootstrap
./run.sh bootstrap

# install just depends
./run.sh install_deps

# test (with just depends)
export _R_CHECK_FORCE_SUGGESTS_="false"; ./run.sh run_tests

# alternate: install all
./run.sh install_all

# test (with all)
export _R_CHECK_FORCE_SUGGESTS_="true"; ./run.sh run_tests