My starting point was a large private project, of which some parts I'd like to make public. I arranged for all the public parts to be located
within one subdirectory. I thought I'd make that available as a separate repo with commits pulled out using git subtree split. The initial split
gave me what I expected: a new disconnected branch with a few commits rooted on the subdirectory.
Then, following some possibly dubious advice, I removed the subdirectory and added it back with git subtree add. Later I performed a second split and that linked the initial disconnected branch with the commit caused
by git subtree add. I pushed the result to the subtree repo, and found that it now contains a complete history of the outer project. The image shows the structure I see within the subtree repo.
Is this expected? Is there a way to avoid it?
