Conclusion

We agree that we want to support Jujutsu as a client for working with Gerrit as outlined in the solution doc.

If Git decides to preseve the change-id header on rebase and cherry-pick (see here) we do not need to implement any of the mitigations to reach compatibility between Git users and Jujutsu users.

If Git decides to not add any support for the change-id header (as it is now) we still agree to the design. In this case we would implement some of the discussed mitigations.