Git bisect command command uses git to find in which commit a bug was introduced. After that, I am sure you will wonder where it's been you whole career.
The first thing to do is find one commit where the bug was not present and remember its hash (for ex. 1234). The next thing is find a commit where the bug was present and remember its hash (for ex. 5678).
And then we can start the search of the bug, type next commands in git repository:
git bisect start git bisect good 1234 git bisect bad 5678
In this case we are setting to git that good commit was the one with hash 1234 and application is broken in hash 5678.
After that, git moves HEAD to a commit just in the middle between good and bad commits, and checkouts that version.
Next step is trying the checkout version and see if the bug was present or not. If it is not present then we execute:
git bisect good
on the other side if the bug was already introduced we must execute:
git bisect bad.
At this point git moves the HEAD in the middle between current commit and top or bottom of commits (depending if bug was already present or not). And this starts a recursive process (in fact git bisect uses binary search) until bug is found.
Finally git will show us the commit with the error, so we can inspect what was introduced. Finally we must execute:
git bisect reset
to leave HEAD pointing to last commit.
You can find an screencast in Git Bisect in Action
Thanks to @lightguard for showing me that command.
We keep learning,