rgoussu.dev
rgoussu@goussu: ~/library/applicative-architecture
❯ ~/library/applicative-architecture cat refactoring-techniques.md

Refactoring techniques

# Behavior-preserving code transformations — the catalog, the smells that trigger them, and how to refactor safely.

Conceptsaved 2026-08-08 #refactoring#craft#design#legacy-code

Overview

Refactoring is changing a program's internal structure without changing its observable behavior, in small reversible steps, to make it cheaper to understand and modify. The canonical treatment is Martin Fowler's Refactoring (2nd ed., 2018): a catalog of named transformations triggered by named code smells, executed under a safety net of tests.

Key points

  • Small steps, always compiling, always green: the discipline is the sequence of tiny behavior-preserving moves, not the end state.
  • Smells → refactorings: long method → Extract Function; feature envy → Move Function; primitive obsession → Replace Primitive with Object; switch on type → Replace Conditional with Polymorphism; data clumps → Introduce Parameter Object…
  • The core catalog: Extract/Inline Function & Variable, Rename, Move, Change Function Declaration, Replace Temp with Query, Split Phase, Encapsulate Variable/Collection.
  • Tests first: no safety net, no refactoring — with legacy code, start with characterization tests and seams (Feathers, Working Effectively with Legacy Code).
  • Opportunistic, not scheduled: prefer the boy-scout rule and preparatory refactoring ("make the change easy, then make the easy change") over big-bang cleanup projects.
  • Know when to stop: refactoring serves the next change; polishing code nobody touches is waste.
  • To explore: strangler-fig pattern for large-scale rewrites, IDE-assisted vs. manual refactoring, refactoring at the architecture level.

Practice

  • Tennis Refactoring Kata (source) — warm-up: an hour of Extract/Rename/Inline on working-but-awful code, tests already green.
  • Parrot Refactoring Kata (source) — the canonical Replace Conditional with Polymorphism drill, one smell one refactoring.
  • Theatrical Players Refactoring Kata (source) — the opening example of Fowler's Refactoring 2nd ed.; replay his move sequence, then diverge and compare.
  • Samman kata descriptions & learning hours (source) — Emily Bache's structured material for deliberate, repeatable refactoring practice sessions.
  • Gilded Rose — legacy rescue (exercise) — the full discipline end to end: characterization tests first, then small steps to a design that welcomes the next feature.

Related

  • TDD — refactor is the third step of the cycle.
  • Clean code — the target state refactoring moves toward.
  • Hexagonal architecture — a common destination when refactoring tangled applications.