twinrun

Twinrun is a powerful tool designed to help developers identify behavioral changes between the old and new versions of code when changes have been made. Unlike traditional methods which simply highlight the lines that have changed, Twinrun dives deeper to determine what the program actually does differently under identical inputs. This utility is essential in mitigating unintended side effects often introduced during code refactors or modifications.

Key Features

• Behavioral Output Comparison: Twinrun operates by running both the old and new versions of the code on the same set of inputs. It reports discrepancies in outputs, unveiling changes that were not intended by the developer.

• Oracle-Based Validation: Employing the old code as the "oracle," Twinrun eliminates the need for explicit specifications or assertions, allowing it to identify deviations in behavior intuitively. If the behavior of code unexpectedly changes, Twinrun will flag the exact call that indicates it.

Usage Example

To utilize Twinrun, simply invoke it from your command line, specifying the base and head revisions:

$ twinrun . --base main --head HEAD
twinrun main..HEAD

  DELTA  billing/cart.py :: Cart.total   +7 more calls
         Cart(0).total()
         base  return       int        0
         head  return       float      0.0

  DELTA  billing/cart.py :: Cart.add     +21 more calls
         Cart(0).add(1)
         base  return       NoneType   None
               after  self=[('items', [1]), ('rate', 0)]
         head  return       NoneType   None
               after  self=[('items', [2]), ('rate', 0)]

Twinrun keeps track of each call, assessing how many callables were checked and how many discrepancies it discovered.

Integration in CI/CD

Twinrun is designed to integrate seamlessly into Continuous Integration workflows. It can be set up to validate code submissions automatically in pull requests and ensures that regressions are caught before merging changes into the main branch, maintaining code quality.

Intended Changes and Acceptance

When deliberate changes to the code are made and accepted, Twinrun allows these findings to be recorded, establishing them as expected behavior for future references. This promotes transparency and clarity in the evolution of a codebase.

Conclusion

Twinrun stands out as an essential tool for developers who prioritize maintaining code integrity during modifications. It provides valuable insights into how changes can affect program logic and outputs, significantly reducing the risk of unintentional regressions.