Jenkinsfile · Lesson 1 of 10 · 3 min read
Declarative vs scripted pipelines
A Jenkinsfile defines a pipeline as code, stored and reviewed alongside the application. Interviews typically ask you to write or critique one on a whiteboard: build → test → scan → push → deploy to staging → approval → production, with parallel tests, credentials handled safely, and clean failure handling. This topic is a complete reference to declarative syntax plus the patterns that make pipelines fast, safe and maintainable.
Two syntaxes, one engine
Both run on the same Pipeline engine (CPS Groovy). They differ in structure:
// Declarative
pipeline {
agent { label 'linux' }
stages {
stage('Build') { steps { sh 'make build' } }
}
post { always { cleanWs() } }
}
// Scripted
node('linux') {
try {
stage('Build') { sh 'make build' }
} finally {
cleanWs()
}
}
| Declarative | Scripted | |
|---|---|---|
| Structure | Fixed, opinionated blocks (pipeline, agent, stages, post...) |
Free-form Groovy |
| Validation | Syntax checked before the run starts; clear errors | Errors appear at runtime, mid-build |
| Readability | Consistent across teams; UI tooling (Restart from Stage, Blue Ocean, directive generator) | Depends on the author |
| Flexibility | Arbitrary logic only inside script {} or library steps |
Unlimited |
| Built-ins | post, when, options, matrix, parallel, input directive |
You write try/catch/finally yourself |
Senior recommendation: declarative by default; keep script {} blocks small; move complex logic into shared-library steps. Scripted pipelines still appear inside shared libraries and in older estates — you should be able to read them.
Enlarge diagram
Swipe sideways to see all of it, or tap to open it full screen.
Where a Jenkinsfile comes from
- Pipeline script from SCM / Multibranch — Jenkins checks out the repo and reads
Jenkinsfile(path configurable). In declarative pipelines,checkout scmhappens automatically at the start of each stage's agent unless you setoptions { skipDefaultCheckout() }. - Inline script in the job config — fine for experiments, not for real pipelines (not versioned).