Developer expresses frustration over large AI-generated pull requests and excessive comments

The Problem with Large Pull Requests

A developer has expressed significant frustration with the increasing size of pull requests (PRs), particularly those that are thousands of lines long. This issue is exacerbated by the use of AI, which can generate extensive code changes in a single submission. The core argument is that large PRs, regardless of their origin, make the code review process exponentially more difficult and time-consuming for human reviewers.

The Purpose of Small PRs

The developer clarifies that the rationale behind advocating for small PRs is not about creating individual, finished products, but rather about presenting work in digestible, reviewable, and understandable segments. This approach is intended to benefit the reviewer by reducing the cognitive load required to comprehend and evaluate the changes, thereby improving the quality and efficiency of the review process.

Critique of Excessive Code Comments

Beyond PR size, the developer also criticizes the practice of including overly detailed or redundant code comments. While acknowledging the value of documentation for functions (e.g., JSDoc, Rustdoc), the developer argues against comments that explain obvious variable names. The stance is that well-named variables should be self-explanatory, and if a variable requires extensive commenting, its name should be improved instead.

AI's Role in Code Review

Addressing the argument that AI can assist in understanding large codebases, the developer questions the utility of human review for AI-generated code if another AI is needed to interpret it. The suggestion is that if AI is used to create large changes, it should also be capable of breaking down those changes into smaller, more manageable units for human reviewers before submission.