DESIGN.rst

Notes to self

pass-code assumes the pass hierarchy will be flat, so the --path=sub-folder argument to pass init is not supported now. Can probably be implemented by putting a .passcode file in the folder, checking whether the paths from arguments refer to a folder that exist and using the .passcode from those folder. Makes things quite complicated.

Since the original pass commands are used whenever possible and the file names passed to those are encrypted, git history is mostly undecipherable. It would be possible to translate encoded/decoded values in pass code git but I don't think that's a good idea.

When pass-code moves/removes/copies a folder, a git commit is issued for every file in that folder. When moving, I can just modify the filenames in the .passcode, but I feel that would make merge conflicts harder to resolve.

I haven't tested what happens when you have to merge two differing git histories, but I guess there will be a conflict every time you try to do so (at least on the .passcode file). Is a merge driver possible? Could I get by with doing union merges for .passcode and verifying referred encrypted files exist while reading it?

From {a, b} have branch x do cp a c, and have branch y do cp b c. Now the merge has two new encoded files, both of which refer to c and have different content. Only conflicting file is the .passcode, so merge driver can't remove one of the encoded ones. I could do union, and prompt to manually delete one of the conflicting files.

pass code init

I should make pass code init convert an existing password-store to pass-code automatically.

• Fail if pass init not run already.
• Create .passcode if doesn't exist.
• Encode current passwords if they exist.
• Check .passcode and skip encoded ones
• If git repo, make a new branch and do git filter-branch magic.
• Don't forget to change filenames in commit messages
• pass init --path=sub-folder is going to cause problems
• Maybe preserve folder structure with encoded:*/decoded in .passcode ?