When cloning a git repository, one needs to clone the entire thing: all files, all history. However, in principle, it does not necessarily need to work this way, for example, when checking out to a client with limited storage or bandwidth, e.g., mobile.
The technical aspects of placeholders for missing files and directories, virtually merging many revisions into one, and fetching the missing information on demand from the parent repository seem complicated (but doable). Also complicated are the UI aspects of specifying what things you want and don't want, and what to do if you end up needing something you don't currently have.
Omitting history would be particularly useful for a large file which got checked in at some point in the past but which has since been deleted, and you are not interested in it. This scenario could occur when using a version control system as a backup system.
No comments :
Post a Comment