Skip to content

Latest commit

 

History

History
39 lines (34 loc) · 1.75 KB

README.md

File metadata and controls

39 lines (34 loc) · 1.75 KB
Raw

git-store: Using git as a KV store for file system

Motivation

When I do some computational experiments with different parameters, I want to records my parameters settings and result log file. Git itself can do that, but in this scenario, concepts like commit, tag, branch are not very useful. And most importantly, I don’t want to see all these files clutter my working trees. I just need a way to reference these files.

Git stores content of files, directories (trees) as objects in “.git””,directory, and these objects are not necessarily associated with commits. By using some lower-level plumbing commands we can make git a KV store for file system.

When using this utility, it’s important to keep the repoistroy clean,. You init a git repo and never commit things into it. Because when we write tree into git, we have to utilize git staging area (index). Currently we assume this staging area has nothing.

Commands

  • init
    • run git init and git config gc.pruneExpire never to prevents some git commands automatically prune unrechable objects.
  • write
    • write a content into “.git”, could be a file or a directory, or multiple files & directories.
    • options:
      • --tree,-t: Write as a tree, even only a single file is given.
      • --delete,-d: Delete the given files & directories after writing.
  • read
    • With a SHA-1 hash of a tree, it will read the tree into current directory.
    • With a SHA-1 hash of a blob object (single file), it will print it’s content into stdout.

Credits

  • Files inside boost and libs come from boost library. I extract these files out so that users doesn’t need to compile boost library themselves.