Recommended Free Tools
A branch is where work lives; it is not necessarily the clearest unit for reviewing it. In “I Miss Reviewing Changes, Not Branches,” Mauro Damian Perez Garcia argues for reviewing one logical decision at a time—so reviewers can understand the intent, assess risk, and disagree before that disagreement blocks a larger bundle of work.
Why a branch can be the wrong unit of review
A pull request presents a set of differences, but that set may contain several distinct decisions. A rename, a refactor, a behavior change, and a new feature can all arrive together simply because they share a branch. Perez Garcia’s point is that this packaging can make it harder to see what deserves scrutiny and why.
As an Amazon Associate I earn from qualifying purchases.
In the author’s account, large reviews can obscure risky lines and raise the cost of disagreement: one objection may hold up a much larger amount of work. He also describes having shipped and approved bugs in large reviews. These are personal observations, not measured evidence that branch-based review causes a particular failure rate.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →The alternative is to make the logical change—the decision a reviewer needs to understand—the focus. A branch or pull request can still contain or host that work; the distinction is about how the work is presented and discussed.
#1 Best Overall
How stacked changes make a sequence easier to review
Related work does not always need to be reviewed as one indivisible package. Perez Garcia gives the example of a sequence that introduces an interface, implements it, then migrates callers. Presented as stacked changes, each step has a more specific question: is the interface right, does the implementation satisfy it, and are the callers being migrated appropriately?
The steps depend on one another, so this is not a claim that each change is independent. The benefit the author argues for is a clearer review sequence: design discussion can happen around the interface before implementation and migration add more material to the conversation. Stacked pull requests are useful where the team’s hosting platform supports them.
Habits that make changes easier to review
- Write commits that stand alone. Give each commit a coherent purpose so a reviewer can follow the reasoning rather than reconstruct it from a mixed diff.
- Separate mechanical edits from behavioral ones. Keep broad renames or other mechanical changes apart from changes that alter what the code does, making the behavior easier to inspect.
- Raise design questions early. When implementation depends on an interface decision, present the interface first so discussion can happen before dependent work accumulates.
- Use stacked pull requests when supported. Represent dependent steps separately while making their order and relationship clear.
- Name the risky part. Tell reviewers which decision or portion of the change needs their closest attention instead of leaving them to infer the review priority.
These are workflow practices, not a formula for a guaranteed review size or outcome. The author’s recommendation is deliberately modest: “None of this requires new software.”
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesWhat to weigh before changing a team’s workflow
Change-centered review is not automatically easier in every code-hosting setup. A stack may need rebasing as earlier changes evolve, and rewriting commits with a force-push can affect whether existing comments remain useful. The team’s platform and conventions therefore shape whether stacked review feels practical.
| Question | Branch-centered pull request | Change-centered or stacked review |
|---|---|---|
| What is presented to the reviewer? | The branch’s combined set of changes. | A logical decision, or a sequence of dependent decisions. |
| How are dependent changes represented? | They may appear together in one pull request. | They can be shown as ordered, related steps. |
| What happens when earlier work changes? | No general rebasing burden is established. | A stack may need rebasing as earlier changes evolve. |
| What happens to comments after rewritten commits? | No general outcome is established. | Comment context after a force-push is a tooling concern to check. |
| What platform effort is required? | Depends on the team’s existing workflow. | Stacking depends on whether the team’s hosting platform supports it and how it handles updates. |
The comparison is about trade-offs, not a universal ranking. A team can adopt the underlying habits—clear commits, separated mechanical and behavioral work, early design discussion, and explicit risk notes—even if stacked pull requests do not fit its tools.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Make disagreement cheaper, one decision at a time
Perez Garcia’s argument is ultimately about making a colleague’s judgment easier: show a change in a form that lets the reviewer understand one decision and challenge it without having to untangle unrelated work. As the author puts it, “The goal was never to have opinions about Git. It was to make it easy for a colleague to understand one decision at a time, and to disagree with it while disagreeing is still cheap.”
Read Mauro Damian Perez Garcia’s original article on DEV Community.
Quick Recap
Best Value
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
