Configure git:

git config --global user.name "Your Name Comes Here"
git config --global user.email you@yourdomain.example.com
git config --global color.diff auto
git config --global color.status auto
git config --global color.branch auto
git config --global core.autocrlf false
git config --global core.savecrlf false 
git config --global diff.renames true
git config --global merge.renameLimit 999999
git config --global diff.renameLimit 999999
git config --global core.excludesfile ~/.gitignore_global

You can now edit the global ignore file:

~/.gitignore_global

# Ignore project files from PHPStorm
.idea/*.xml
.idea/.name
.idea/scopes
.idea/*.iml

Some nice settings:

git config --global alias.st status
git config --global alias.ci commit
git config --global alias.br branch
git config --global alias.co checkout
git config --global alias.df diff
git config --global alias.undolast 'reset --soft HEAD^'
git config --global alias.datetag '!git tag `date "+%Y_%m_%d_%H%M"`'
git config --global branch.autosetuprebase always
git config --global push.default current

Checkout the first version with git:

git clone idefix.fechner.net:masterthesis.git masterthesis

Have several copies on several places and sync it with the master server (like svn):

mkdir repo
cd repo
git init repo
git pull ssh://user@server/dir/
git remote add origin git@server:pluginname.git
git config branch.master.remote 'origin'
git config branch.master.merge 'refs/heads/master'
git push origin master:refs/heads/master

git push origin :branchname

We have migrated now a repo but it contains several directories which should be splitted into seperate repos. The structure looks like:

git-root/
         AutoInvite
         phpraider

This can be done with filter-branch. At first checkout the repo as often as you want to split it:

git clone git@server:auto_invite AutoInvite
git clone git@server:auto_invite AutoInvite_phpRaider

Now we remove everything from the repo which should not be in:

cd AutoInvite
git filter-branch -f --tree-filter 'rm -Rf phpraider' --prune-empty -- --all
git gc
cd ..
cd AutoInvite_phpRaider
git filter-branch -f --tree-filter 'rm -Rf AutoInvite' --prune-empty -- --all
git gc
cd ..

Now we rewrite the root directory:

cd AutoInvite
git filter-branch -f --subdirectory-filter AutoInvite -- --all
git gc
cd ..
cd AutoInvite_phpRaider
git filter-branch -f --subdirectory-filter phpraider -- --all
git gc
cd ..

Push the new repos to a gitolite server:

cd AutoInvite
git push --all git@localhost:AutoInvite
cd ..
cd AutoInvite_phpRaider
git push --all git@localhost:AutoInvite_phpRaider
cd ..

Remove everything from your local repository:

git for-each-ref --format="delete %(refname)" refs/original | git update-ref --stdin
git reflog expire --expire=now --all
git gc --prune=now

The problem here is that the first commit cannot be changed with the rebase command. So we have to create a new first commit befor we can modify the real first commit:

git symbolic-ref HEAD refs/heads/newroot
git rm --cached -r .
git clean -f -d
# touch .gitignore && git add .gitignore # if necessary
git commit --allow-empty -m 'Initial commit'
git rebase --onto newroot --root master
git branch -d newroot

Now we can change commits with rebase -i as usual.

git remote add repotomerge /path/to/repo
git fetch repotomerge 
git merge repotomerge/<branch to merge>
git gc

We have two git repositories called subfolder1.git and subfolder2.git, both repositories will contain several branches and we want to build up a complete new structure in the way:

base directory
 |- subfolder1 (from repository subfolder1)
 |- subfolder2 (from repository subfolder2)
 |- more folder (newly added)
- file1 (newly added)

The repositories are available via git clone from:

http://server/subfolder1.git
http://server/subfolder2.git

We use the following script to checkout the repository and rewrite the path and merge everything into one repoistory with complete history and all branches:

mergeGitRepos.sh

#!/bin/bash
# (c) 2013 Matthias Fechner
 
 
me=$(basename "$0")
 
TMP=$(mktemp -d /tmp/$me.XXXXXXX)
echo
echo "building new repo in $TMP"
echo
 
set -e
 
cd "$TMP"
mkdir new-repo
cd new-repo
        git init
        cd ..
 
x=0
while [ $# -gt 0 ]
do
        repo="$1"
        shift
        dirname=$(basename "$repo" | sed -e 's/\s/-/g')
        dirname=$(basename "$dirname" | sed -e 's/.git$//')
        echo "Clone $repo"
        git clone --bare "$repo" "$dirname"
 
        cd "$dirname"
                echo "Checkout all branches"
                for remote in `git branch -r | grep -v master `; do git checkout --track $remote ; done
                git checkout master
                echo "Remove reference to origin, so we cannot push by accident"
                git remote rm origin
                echo "Move all files and directories to new location"
                git filter-branch --index-filter 'git read-tree --empty; git read-tree --prefix="$dirname/" "$GIT_COMMIT"' -- --all
                cd ..
 
        cd new-repo
                git pull --no-commit ../$dirname
                [ $x -gt 0 ] && git commit -m "merge made by $me"
                cd ..
 
        x=$(( x + 1 ))
done

The script is named mergeGitRepos.sh and is executed this way:

mergeGitRepos.sh http://server/subfolder1.git http://server/subfolder2

After it is done. you can copy your missing folder (more folders) and your missing files and commit them as usal.

The script failed and do not what I wanted. New approach is to filter every repository to get the correct path there.

At first we copy the repository on the server so we work on a test repository, that is important, because we will overwrite it several times to get everything merged.

cp -r repo.git repo_save.git

Now the real work:

git clone http://server/repo_save.git
cd repo_save
for remote in `git branch -r | grep -v master `; do git checkout --track $remote ; done
git checkout master
git remote rm origin
git filter-branch --index-filter 'git read-tree --empty; git read-tree --prefix="subfolder/" "$GIT_COMMIT"' -- --all
git push -f --all http://server/repo_save.git
git push --tags http://server/repo_save.git
cd ..
rm -rf repo_save

Redo this for every repository. Now all repositories should have there correct folder displayed for all branches.

Now we combine them to one repository:

mkdir new-repo
cd new-repo
git init
git remote add origin http://server/repo1_save.git
git fetch --all
git fetch --all -t
git pull
git checkout master
for remote in `git branch -r | grep -v master `; do git checkout --track $remote ; done
git checkout master
git remote rm origin


git remote add origin http://server/repo2_save.git
git fetch --all
git fetch --all -t
git merge origin/master
... merge all branches ....
git remote rm origin

git remote add origin http://server/new-repo.git
git push --all
git push --tags
cd ..
rm -rf new-repo

Now you have your new repository on the server in new-repo.git.

Clone it, test it thate everything like you expected it. If everything is fine, remove all backup and you are done.

You have a project and you want to make it public but you commited password to your repository? Lets say the passwords are in the file bin/config, so we can do the following:

git filter-branch --index-filter 'git rm -r --cached --ignore-unmatch bin/config' HEAD