I was thinking over the past months about the intended direction of the project, as understood from earlier comments by the author, toward a versioning paradigm similar to git's. I came to wonder simply whether it had been considered whether support for a purely git backend would be feasible and desirable. In such a case, the system would rely on the low-level "plumbing" git operations, which allow the direct manipulation of objects in a tree. The advantages would be leveraging existing features of git, such as cloning and tree traversal, as well as the wide and free availability of git instances on servers. Users could synchronize against their own repositories rather than instances of some particular database.
I was thinking over the past months about the intended direction of the project, as understood from earlier comments by the author, toward a versioning paradigm similar to git's. I came to wonder simply whether it had been considered whether support for a purely git backend would be feasible and desirable. In such a case, the system would rely on the low-level "plumbing" git operations, which allow the direct manipulation of objects in a tree. The advantages would be leveraging existing features of git, such as cloning and tree traversal, as well as the wide and free availability of git instances on servers. Users could synchronize against their own repositories rather than instances of some particular database.