I have a large git repository with binary files scattered over different branches. I want to switch to git-annex mainly for performance reasons, but I don't want to loose my history.

I tried to rewrite the (cloned) repository with git-filter-branch but failed miserably for several reasons:

  • --tree-filter performs its operations in a temporary directory (.git-rewrite/t/) so the symlinks point to the wrong destination (../../.git/annex/).
  • annex log files are stored in .git-annex/ instead of .git-rewrite/t/.git-annex/ so the filter operation misses them

Any suggestions how to proceed?

EDIT 3/2/2010 I finally got it working for my purposes. Hardest part was preserving the branches while injecting the new git annex setup base commit.

Clone repository

git clone original migrate
cd migrate
git checkout mybranch
git checkout master
git remote rm origin

Inject git annex setup base commit and repair branches

git symbolic-ref HEAD refs/heads/newroot
git rm --cached *
git clean -f -d
git annex init master
echo \*.rpm annex.backend=SHA1 >> .gitattributes
git commit -m "store rpms in git annex" .gitattributes
git cherry-pick $(git rev-list --reverse master | head -1)
git rebase --onto newroot newroot master
git rebase --onto master mybranch~1 mybranch
git branch -d newroot

Migrate repository

mkdir .temp
cp .git-annex/* .temp/
MYWORKDIR=$(pwd) git filter-branch \
 --tag-name-filter cat \
 --tree-filter '
    mkdir -p .git-annex;
    cp ${MYWORKDIR}/.temp/* .git-annex/;
    for rpm in $(git ls-files | grep "\.rpm$"); do
        git annex add $rpm;
        annexdest=$(readlink $rpm);
        if [ -e .git-annex/$(basename $annexdest).log ]; then
            echo "FOUND $(basename $annexdest).log";
            echo "COPY $(basename $annexdest).log";
            cp ${MYWORKDIR}/.git-annex/$(basename $annexdest).log .git-annex/;
            cp ${MYWORKDIR}/.git-annex/$(basename $annexdest).log ${MYWORKDIR}/.temp/;
        ln -sf ${annexdest#../../} $rpm;
    git reset HEAD .git-rewrite;
    ' -- $(git branch | cut -c 3-)
rm -rf .temp
git reset --hard


  • Find a way to repair branches automatically (detect branch points and run appropriate git rebase commands)

I'll be happy to try any suggestions to improve this migration script.

P.S. Is there a way to edit comments?