behavior-preservation-checker

Behavior preservation verification works by executing both the old and new code versions against identical inputs, then comparing their outputs and side effects against a defined equivalence relation to detect any functional regressions caused by the refactoring.

The best way to check for regressions during automated code transformations is differential testing, where you execute both code versions and compare their I/O captures and side effects to ensure the structural changes do not break functionality.

Yes, you can use differential testing for pre-merge pull request reviews by running production-like inputs against both the original and modified code branches to confirm that their responses and database interactions remain identical before merging.

No, behavior verification requires the ability to execute both code versions and compare their outputs and side effects. Without an environment capable of running the old and new code against identical inputs, equivalence cannot be validated.

Available methods for side-effect capture and equivalence checking include differential testing, golden-master or snapshot testing, side-effect capture, and property-based equivalence, allowing you to compare outputs and database interactions between code versions.