I have over 200 projects on my workstation, every one is a git repository. Most of them are superseded or otherwise abandoned. I still keep the code in case I ever need anything from them.
Currently, the active projects are non-bare git repositories in ~/Projects, the inactive projects are .tar.gz archives that simply contain the whole non-bare git repository within the same folder. So I got the following:
~/Projects/some-active/.git~/Projects/some-inactive.tar.gz
The advantage of this is that backups are really fast, inactive projects are just a single file which has to be checked/copied. Whenever I want to look into the projects, I have to extract the folder, look into it and compress it some time later when I think the project is inactive again.
Is there a better way to this?
Best Answer
First suggestion: Just keep them the same way as active repositories, except make sure all objects are in a single packfile (
git gcwill do that). rsync mostly just checks modification times; it doesn't read the whole file if everything else matches; it shouldn't be slow. (My backup consists of rsyncing over 1M files, many of which are Git repositories, and it's still acceptably fast...)Second suggestion: Keep them as bare repositories. Whenever you want to look at the files, use
tig's "tree browser" mode, or clone to /tmp (e.g.git clone -s ~/Projects/foo.git /tmp/foo). This way, each repo will have only about 5 files (packfile, pack index, config, packed-refs).