# Job Dependencies With needs — GitHub Actions

Source: https://www.skillbyai.com/en/github-actions/j-needs

> Build a pipeline graph.

## Parallel by default, ordered by needs

Jobs run in parallel unless `needs:` says one depends on another; a job with `needs: [test, lint]` waits for both and is skipped if either fails (unless its `if` uses `always()` or similar status functions). This forms a directed graph: run fast checks in parallel, then build, then deploy. Mistakes in `needs` (typos, cycles) are caught by linting before a run is attempted, as the cron example in the previous section showed.

## A pipeline graph

Not linted or run here; check the GitHub Actions documentation for current syntax.

```yaml
jobs:
  lint:      { runs-on: ubuntu-latest, steps: [ { run: npm run lint } ] }
  test:      { runs-on: ubuntu-latest, steps: [ { run: npm test } ] }
  build:
    needs: [lint, test]               # waits for both
    runs-on: ubuntu-latest
    steps: [ { run: npm run build } ]
  notify:
    needs: build
    if: ${{ always() }}                # runs even if build failed
    runs-on: ubuntu-latest
    steps: [ { run: echo "build result ${{ needs.build.result }}" } ]
```

## Fail fast with cheap jobs first

Put quick checks (lint, type check) in early parallel jobs so obvious problems fail within a minute.

**Quiz:** What happens to a job whose needs dependency failed?

- [ ] It retries forever
- [ ] It runs anyway
- [x] It is skipped, unless its if condition uses a status function such as always()
- [ ] It deletes the workflow

*Answer:* It is skipped, unless its if condition uses a status function such as always(). Dependencies gate execution.