Single Responsibility Principle

Introduction

SRP — a module should have one reason to change: one actor or axis of change drives modifications, not "one method only" literalism.

Business problem

Business pressure: Uber's core platform must ship weekly while keeping incident rates below SLO. Without Single Responsibility Principle, teams accumulate coupling — every feature touches the same modules, regressions correlate, and senior hiring loops fail because candidates cannot articulate refactor trade-offs.

• Velocity: Median PR size grows; cross-team coordination dominates delivery time.
• Quality: Production incidents cluster around modules with multiple reasons to change.
• Talent: Staff promotion cases require evidence of principled architecture decisions — not buzzwords.

Architecture motivation

Why architects care: Single Responsibility Principle addresses one reason to change per module. The naive alternative (God classes, concrete dependencies, duplicated business rules) works until team size and traffic make every change expensive and risky.

• Force: Requirements change along different axes — tax law, UI, persistence, notifications.
• Constraint: Cannot pause feature development for a multi-year rewrite.
• Outcome: Clear boundaries, faster tests, principled code reviews, and ADRs that survive turnover.

Real production story

InvoiceService generated PDFs, sent email, and calculated tax. Tax law changed — developers feared touching email code. Splitting into InvoiceCalculator, PdfRenderer, and Mailer cut change risk and test surface per class.

Enterprise case study

Uber — Single Responsibility Principle in production: Platform team inherited an 80k-line module where every feature violated SRP. Incremental extractions behind interfaces, package acyclicity in CI, and principle-based review rubrics reduced change-failure rate 45% without a big-bang rewrite.

• Before: Correlated regressions; reviews debated formatting.
• Decision: Apply smallest refactor that improves testability and names the principle.
• After: Faster unit tests, smaller PRs, staff-ready interview narratives.

Refactoring walkthrough

SRP extraction sequence for InvoiceService:

// Step 1: Characterization tests on InvoiceService.public API
// Step 2: Extract InvoiceCalculator (tax rules only)
// Step 3: Extract PdfRenderer (layout only)
// Step 4: Extract Mailer (delivery only)
// Step 5: InvoiceService orchestrates — thin facade
// Each class: one stakeholder / one change driver

Failure scenario

What breaks when Single Responsibility Principle is ignored or misapplied:

• Big-bang rewrite trap: Team freezes features for 9 months; business ships competitor features first.
• Golden hammer: Strategy pattern for every if/else — YAGNI and readability suffer.
• False DRY: Forced abstraction across unrelated domains — wrong coupling worse than duplication.

Scalability analysis

Scale dimensions: Single Responsibility Principle decisions compound across team count, codebase size, and deployment frequency — not just lines of code.

• Team scale: Conway's Law — module boundaries should align with ownership.
• Code scale: Package cycles block parallel development; enforce acyclic graphs in CI.
• Change scale: Feature flags + incremental extraction beat monolithic refactors.

Security considerations

Security is structural: Single Responsibility Principle affects blast radius, secret handling, and auditability.

• Boundary leaks: God classes mix auth logic with export code — privilege escalation paths hide in tangled methods.
• Test doubles: DIP enables security unit tests without hitting real payment APIs in CI.
• Supply chain: Stable abstractions reduce direct imports of vendor SDKs across the codebase.

Architecture review questions

• What is the single reason to change for the module under review (Single Responsibility Principle)?
• Which principle is violated — and what is the smallest safe refactor?
• Are tests sufficient to characterize behavior before extraction?
• What trade-off are we accepting (indirection vs testability)?
• Does this change align with package dependency rules and team ownership?
• Is there an ADR if this is an architectural boundary shift?

Staff engineer notes

• Principle-based reviews convert subjective taste into teachable decisions — your job is to make the trade-off legible.
• At Uber scale, Single Responsibility Principle fails at boundaries: interfaces, test fakes, and package imports — not diagram aesthetics.
• Good architecture is boring on the happy path and explicit on the failure path — if runbooks are fiction, the design is not production-ready.

Interview questions

Hands-on refactoring exercise

Refactoring exercise: Take a 200-line God class from your codebase (or the capstone module). Map smells to Single Responsibility Principle, write characterization tests, and perform one extraction that improves testability without changing public behavior.

• Deliverable 1: Smell inventory mapped to principles.
• Deliverable 2: Before/after dependency diagram.
• Deliverable 3: One merged PR with tests green.

In-depth explanation

Core idea: Single Responsibility Principle guides how you structure code so that one reason to change per module. It is a force balancer — not a rule to maximize at all costs.

• Definition: Single Responsibility Principle — durable guideline for structuring maintainable software.
• Violation signals: God classes, shotgun surgery, feature envy, divergent change.
• Refactor moves: Extract class, introduce interface, move method, replace conditional with polymorphism.
• When to relax: Early prototype, throwaway spike, or truly trivial script — optimize for learning speed.

Visual explanation

Three diagrams: principle architecture, refactor workflow, and common violation patterns:

Structural view

SRP — structural view:

God class (many reasons to change)
   ↓ split by axis of change
InvoiceCalculator → tax rules only
PdfRenderer → layout only
Mailer → delivery only

Execution workflow

Informative example

SRP smell checklist:

SRP smells:
- Class name has "And" or "Manager" doing unrelated work
- Tests need 10 mocks for one method
- Bug fix in feature A breaks feature B in same file
Fix: extract by reason to change, not by line count

Real-world use

Uncle Bob's SOLID and Clean Architecture popularized dependency direction for enterprise systems. The Pragmatic Programmer's DRY, orthogonality, and tracer bullets shape daily craft. Martin Fowler's refactoring catalog turns principles into safe, incremental steps. Package design rules from Robert C. Martin and Rebecca Wirfs-Brock keep monorepos navigable. These ideas appear in Uber's code review culture, Google readability standards, and Stripe's API design norms.

Trade-offs

• Pro: clearer change isolation, faster principled reviews, and interview-ready narratives.
• Con: more types and indirection — hurts readability if over-applied.
• Con: premature abstraction violates YAGNI — balance with KISS.

Best practices

• Map each review comment to a principle, not personal taste.
• Characterization tests before extracting legacy God classes.
• Enforce acyclic package dependencies in CI where possible.

Anti-patterns

• SOLID buzzword review — no concrete smell or proposed fix.
• Golden hammer — Strategy pattern for every if/else.
• Big-bang rewrite instead of incremental principle-driven extractions.

Common mistakes

• SRP taken as "one public method" — misses axis of change.
• DRY across unrelated domains — wrong abstraction worse than duplication.

Summary

Single Responsibility Principle at staff level means naming smells, proposing minimal tested refactors, and documenting trade-offs — the skill Uber senior engineers demonstrate in every architecture review.

Key takeaways

• Single Responsibility Principle is a force balancer — apply with evidence, not dogma.
• Production success depends on incremental refactors, tests, and explicit trade-offs.
• Use this vocabulary in reviews and staff interviews with concrete smells and fixes.