The Key Points of Working Effectively with Legacy Code

5 days ago

Copy Link

Legacy Code is defined as code without tests, a concept from Michael Feathers' book 'Working Effectively with Legacy Code'.
The primary approach to working with legacy code is to first add tests to ensure behavior preservation before making changes.
A key challenge is the 'Legacy Code dilemma': needing to change code to add tests, but needing tests to safely change code.
Identify 'Seams' in the code—places where behavior can be altered without changing the source code—to break dependencies for testing.
Unit tests should be fast (<100ms per test) and not interact with infrastructure to be reliable.
Characterization tests capture the current behavior of code to create a safety net for refactoring.
Use 'Sprout' and 'Wrap' techniques to add new code to legacy systems with minimal risk when time is limited.
Scratch refactoring helps familiarize with opaque code by making temporary changes that are reverted after understanding is gained.
Avoid direct dependencies on library implementations to prevent lock-in and ease future upgrades.
Despite being written in Java/C++, the book's principles are language-agnostic and highly applicable to modern development.

Hasty Briefsbeta