A GitHub action, containerized and written in Go, to detect simple cycles or topological generations in directed multi-graphs. While this repo has utility, it is not novel and serves primarily pedagogical purposes.

This project was first implemented at the University of Texas at Austin as a group project with tcm5343, pabs159, and rory-tatum. A use case was identified that applied to our industry experience.

Below is an example GitHub Action job which utilizes the circular dependency detector. Simply write you graph input file to /testing/data and then pass that into the detector.

jobs:
  cdd:
    runs-on: ubuntu-latest
    steps:
      - name: checkout cdd
        uses: actions/checkout@v4
        with:
          repository: ${{ github.repository }}
          ref: ${{ github.GITHUB_REF_NAME }}  # this should be a working tag

      - name: install task
        uses: arduino/setup-task@v2
        with:
          repo-token: ${{ secrets.GITHUB_TOKEN }}
      
      - name: create graph input file
        run: |
          echo -e "1 2\n2 3\n3 1" > testing/data/some-input-graph.txt

      - name: circular dependency detector
        run: |
          task INPUT_FILE="testing/data/some-input-graph.txt" run

• Task shall be used in the CI/CD pipelines and for local development for orchestration. Run task for the list of tasks.
• Podman is used to manage containers in theTaskfile.
• golangci-lint is used for lint.

For local development, create a .env file at the root of the repository to modify your config. The only supported format for the input graph file environment variable (INPUT_FILE) is an adjacency list which follows the format used by NetworkX. An .env-example file exists at the root of the project.

user@machine:~/dev/circular-dependency-detector$ cat ./.env 
INPUT_FILE=testing/data/adj_list_no_cycle.txt  # no spaces for now, defaults to {i don't know yet}

Bob Dylan - Early Mornin' Rain