> But you still need to be connected to work with parts of the code base you haven't used before. Which seems like a fair tradeoff to me. The existing "native

> But you still need to be connected to work with parts of the code base you haven't used before. Which seems like a fair tradeoff to me. The existing "native solution" for a very very large codebase would be to have it split into multiple, logical repositories. If you fetched one repo you needed to work on, but not all the dependencies or sibling repos, you still wouldn't be able to work on those other parts of the codebase until you connected.