On the server, set up the remote repositories:

1
2
3

mkdir project1.git
cd project1.git
git --bare init

Here’s a script to do them all in one shot, just modifiy the REPOS variable:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

#!/bin/sh
 
REPOS="project1 project2 project3 project4"
 
for repo in $REPOS
do
    repo_dir="$repo.git"
    mkdir -p $repo_dir
    echo "Creating git directory  $repo_dir"
    cd $repo_dir
    git --bare init
    cd ..
done
exit

Now, on the workstation:

1
2

sudo apt-get install git-core git-svn
sudo gem install nirvdrum-svn2git --source http://gems.github.com

Create the authors.txt in the following format:

1
2

dburger = David Burger <email@email.com>
jdoe = John Doe <jdoe@doe.com>

For one project, do the following:

1
2
3
4
5

mkdir project1
cd project1
svn2git  https://svn.myserver.org/repos/ses --authors ../authors.txt
git remote add origin hailstorm.myserver.org:/home/kenglish/repotest/ses.git
git push --all

The script:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

#!/bin/sh
 
REPOS="project1 project2 project3 project4"
 
for repo in $REPOS
do
    mkdir $repo
    cd $repo
 
    cmd="svn2git https://svn.myserver.org/repos/$repo --authors ../authors.txt"
    echo $cmd
    `$cmd`
    cmd="git remote add origin hailstorm.myserver.org:/home/kenglish/repotest/$repo.git"
    echo $cmd
    `$cmd`
    cmd="git push --all"
    echo $cmd
    `$cmd`
    cd ..
    echo "DONE EXPORTING $repo" 
done
exit

Don’t diss the shell script, I leave in the echoes in case i need to test stuff out…

Note: Another option for the authors file is place it in your home directory .svn2git/authors (e.g. /home/kenglish/.svn2git/authors). Svn2git will automatically detect it and use it.
Any questions? Comments?

My problem is that I have multiple git branch of a rails project and in each branch, I want to maintain a different value the branch setting my deploy.rb. For example:

In branch v1.0.0-stable

set :branch, "v1.0.0-stable"

In branch, I want v1.0.1-stable

set :branch, "v1.0.1-stable"

In the master branch, I want no value.

The problem is that I often merge from v1.0.1-stable into the master branch and this overwrites the deploy.rb. This can be frustrating because every time  I need to remember to remove that “set :branch” line from the deploy.rb. If I forget and deploy from the master, it  will deploy version “v1.0.1-stable” instead of master. This always results in my wondering  “Why didn’t the deploy work? Why aren’t my changes on the site?”

The only way that I could find to prevent this is to adding a line my gitattributes file.

In .git/info/attributes, I added:

deploy.rb merge="ours"

Ours is a merge strategy and this tells git always apply

      ours
           This resolves any number of heads, but the result of the merge is
           always the current branch head. It is meant to be used to
           supersede old development history of side branches.

Note:
If you Google git merge the first 5 pages of results are the man page for the git-merge command. Why does everyone feel the need to replicate the man page all over the internet. Seriously! If what I needed was in the man page, I wouldn’t be using Google!