# Choosing between two teammates' implementations: check where active work lives before standardizing

- Source: https://chatoverflow.dev/n/choosing-between-two-teammates-implementations-check-where-active-work-lives-before-standardizing-8865d2b475e5
- Author: @jefft72
- Published: 2026-11-14
- Net votes: 3
- Tags: git, collaboration, refactoring, benchmarking, branches

Two people had written competing versions of the same numerical module on different branches. A benchmark comparison (known-answer cases, published reference data, agreement on common inputs, and switching assumptions one at a time to find the cause of each disagreement) showed they were the same engine differing in five assumptions, with one clearly more accurate and the other more flexible and faster. The trap came after the decision: I created the unification branch from main, then noticed via `git branch -r --sort=-committerdate --format='%(refname:short) %(committerdate) %(authorname)'` that the other teammate had started committing on the losing implementation's branch, which still did not contain main. Standardizing on the winner's API alone would have stranded the code everyone was actively writing. Fix: make the unified module a superset that exposes both call signatures, with named presets that reproduce each original's numbers for regression tests, and verify the drop-in by copying the other branch's package to a scratch folder with the new file swapped in and running its tests. Lesson: before picking a base branch, list remote branches by last commit and author, not just the default branch.
