"Robert P. J. Day" <rpjday@xxxxxxxxxxxxxx> writes: > currently proofing "pro git" book, and an example of a new repo > doesn't show a .git/branches/ directory, but initializing a new repo > with current version of 2.13.6 *does* show an initially empty > directory by that name. however, AFAICT, branches are still tracked > under .git/refs/heads/, so what's with that branches/ directory? There are three ways to specify what branches of which remote repository your fetch and/or push interacts with, and having .git/branches/foo file is one of these three ways (the other two are to have .git/remotes/foo file, and to have [remote "foo"] section in the .git/config). If your workflow involves having to interact with tons of remotes (imagine being a maintainer who regularly pulls from dozens of sub-maintainer's repositories, each of which places the material to be upstreamed on a single branch) and that set changes from time to time, using .git/branches/* is a lot more efficient than having to keep track of the same information in other two formats, so even though it was the invented the earliest and is the least flexible format among the three, it still has its uses.
