Job Dependencies (DAG)

Jobs can DEPEND on others:

jobs:
  build:
    runs-on: ubuntu-latest
    steps: [...]

  test:
    needs: build
    runs-on: ubuntu-latest
    steps: [...]

  deploy:
    needs: [build, test]
    runs-on: ubuntu-latest
    steps: [...]

The pipeline is a directed acyclic graph (DAG). CI runner topologically sorts: build first, then test (parallel with anything else), then deploy (needs both).

Why?

• Reuse build artifacts: build once, test in N parallel matrix jobs.
• Skip downstream on failure: don't deploy if tests failed.
• Parallel independent paths: lint + build can run together if they don't depend on each other.

Failure semantics:

• A job fails → all needs: -dependent jobs are SKIPPED.
• Other independent jobs may continue (if fail-fast: false).
• Workflow status is "failure" if any job failed.

Modern systems also have:

• Conditional needs: if: success() vs if: failure() vs if: always().
• Concurrency groups: prevent two pipelines from running for same branch (concurrency: deploy-prod).