GiellaLT Documentation

GiellaLT provides rule-based language technology aimed at minority and indigenous languages

View GiellaLT on GitHub

Page Content

Git-svn history cleanup

Cf this project.

If you do this on a mac, you need macOS 10.13 High Sierra or older, later versions do not come with git svn, and it is a hassle to install your own version. Running an old macOS in a VM is easy and painless.

Step 1 - identify relevant svn history points

Check out at a suitable point in the past, typically just before the last known commit of the relevant directory in the git history.

Use svn log to find the earliest point of that dir tree as it was then.

Step 2 - Clone using git svn

Use the revision numbers you got in the previous step to restrict the cloning to the relevant history:

git svn clone -r86176:120958 --authors-file=authors.txt

REPEAT for each major change in dir structure, creating one local git repo each time.

Step 3 - stitch the histories together

Follow these instructions.

With the example above, the following steps were taken:

git fetch ../urj-Cyrl-eldre eldre:eldre
echo $(git rev-list gamal | tail -n 1) $(git rev-parse eldre) > .git/info/grafts
git filter-branch $(git rev-parse eldre)..HEAD


Step 4 - clean up usernames and emails

See this issue comment and the next one.

Step 5 - force push

When everything is collected and cleaned, force push, and inform relevant developers they need to reclone.