2022-05-03 01:31:11 -04:00
|
|
|
---
|
|
|
|
title: "Which commit broke the build? Using Git Bisect"
|
|
|
|
date: 2022-05-03T01:15:55-04:00
|
|
|
|
draft: false
|
2022-11-12 11:22:54 -05:00
|
|
|
tags: ["Git"]
|
2022-05-03 01:31:11 -04:00
|
|
|
math: false
|
|
|
|
---
|
|
|
|
|
|
|
|
Lets imagine a scenario where in the latest merge a test
|
|
|
|
starts failing. Lets say these tests are saved
|
|
|
|
in `test.sh`. Instead of having to test each individual
|
|
|
|
commit in the merge, to see where the test fails, luckily
|
|
|
|
`git bisect` narrows it down in a more efficient way!
|
|
|
|
|
|
|
|
To use:
|
|
|
|
```bash
|
|
|
|
git bisect start [good] [bad]
|
|
|
|
git bisect run test.sh
|
|
|
|
```
|
|
|
|
Where `[good]` and `[bad]` are replaced with their respective
|
|
|
|
commit hashes.
|
|
|
|
|
|
|
|
Under the hood, Git will run a binary search between the good
|
|
|
|
and bad nodes in the commit tree.
|
|
|
|
|
|
|
|
As a reminder, don't forget to make `test.sh` an executable.
|
|
|
|
Starting in Git 2.36 it will provide a warning, but earlier
|
|
|
|
versions will perform the search anyways even with it all
|
|
|
|
failing.
|
|
|
|
|
|
|
|
|
|
|
|
Read more: https://git-scm.com/docs/git-bisect
|