The "Commit Cruncher" is the label GitClear's internal source code uses to refer to the underlying processing engine that classifies git commits as a set of 6 change operations, and records the provenance of each changed line.


The main context in which "Commit Cruncher" is relevant to discuss as a concept is relative to its diff processing predecessor, the Myers Diff Algorithm, described by GitClear's 2024 research comparing "lines to review" across 12,638 pull requests on GitClear vs. GitHub and in even greater depth in a well-crafted 2017 blog post by James Coglan.


Salient differences between the classic Myers vs Commit Cruncher diff algorithm are summarized below:


Dimension

Myers

Commit Cruncher

Date of origination

1986

2020

Implementation source

Research paper

Proprietary iterated source

Changes classified as

Add, delete

Add, delete, update, move, find/replace, copy/paste

Pull request processing

Compare contents of base commit to final commit

Follow changed lines through each commit they appear within

Used by

GitHub, Bitbucket, Gitlab, Azure Devops

GitClear

Median changed lines per PR

648 changed lines (27.8% more lines)

507 changed lines


The biggest differences between the result of the Myers and Commit Cruncher algorithms are observed within large pull requests:


open_in_new

Source: 30% Less is More: Exploring Strategies to Cut Pull Request Review Time