git-step.sh

#!/bin/bash

show_usage () {
  echo "Usage: `basename $0` [START [END]]"
  echo
  echo "Steps through the commit history from START to END,"
  echo "then returns to the branch or commit from before execution."
  echo
  echo "START defaults to the root commit (beginning of history)."
  echo "END defaults to current branch/commit."
}

initial_branch=$(git symbolic-ref --short -q HEAD)
initial_commit=$(git rev-parse HEAD)
reset_to=${initial_branch:-${initial_commit:-"master"}}

if [[ ( $1 == "--help" ) || $1 == "-h" ]]; then
  show_usage
  exit 0
fi

if [ $# -eq 0 ]; then
  end=$reset_to
elif [ $# -eq 1 ]; then
  start="^$1^"
  end=$reset_to
elif [ $# -eq 2 ]; then
  start="^$1^"
  end=$2
else
  show_usage
  exit 1
fi

for commit in $(git rev-list $end $start --reverse); do
  git checkout $commit
  read
done

git checkout $reset_to

Steps through a git repository's history commit by commit (e.g. for a live demo). Inspired by http://stackoverflow.com/questions/3296260/how-to-step-through-a-git-repository#answer-19307936.

Nice script. But it chokes on me after a branch gets merged into master in some cases (not all). To get it back on track:

1. Manually checkout the repo to the commit after the merge
2. Run git rev-parse <short_commit_hash_of_current_commit>
3. ./git-step.sh <long_commit_hash_from_step2>

UPDATE: Disregard this note. I started the git-step.sh script in a detached head state. It was only choking because it got to the commit where my initial detached head was.

omg thank you for this script, so useful.