Programming

Git is really good at two things you are doing:

• tracking separate files in related but disconnected repositories
• backing thing up intermittently when access is available

I don't know your exact scenario, but you've shared enough that I can confidently recommend:

• the files on each device that is removed and restored should in a separate repository
• you should at least consider using git init --bare somewhere in this recipe - you can put a remote on a removable drive, and use git pull and git push to sync to it

I haven't seen git update-index --skip-worktree mentioned yet. You can read about the motivation for this feature in the git scm docs.

I have used it in the past when a professor wanted us to clone repos for assignments that included some opinionated settings for VSCode that I didn't want to use. Skipping the work tree for that directory allowed me to change or delete the config files without git complaining every time I pushed or pulled or whatever, and the changes I made remained local.

You could set up a couple git aliases to "freeze" and "thaw" your config files on the second drive.

The odd config files on inconsistent drive should just be symlinks (I think you want hard links?) so that your repo can contain all your actual code and file tracking. If necessary, keep a script on hand that can be run when mounted to recreate broken links.

This is a very strange setup and goes against standard practice separation of software and hardware unless this is some embedded thing in which case you wouldn’t have a repo on it at all.

I think you need to give more details about why you have this crazy set-up. What are you trying to achieve?

You can git add the file but also add it to .gitignore. But I don't know if that really solves your problem because it isn't clear what your problem is.

If you commit the file on the secondary drive then it will be in the same repository as the rest of the code so you will end up with a large tree in that repo or you will need to move it over with the rest of the files. Why not make a symlink on the secondary drive pointing to the config in the repository? This way you can track it in git and have it located on the secondary.

Your project code structure should be independent of your hardware structure. Maybe include the config file to your local drive and setup some file sync to auto deploy those files to the mount drive? E.g. syncthing

You can just not git add the one file on the secondary drive when it is not connected. This schould not be that much overhead and (at least for me) it is much simpler than any other solution. Ps: Why do you want that one config file on the external drive and not on the main one?

Can you clarify if you want to track changes across all the files or just the ones that will always be available? And will you be automatically committing the state of the files at some regular interval or would you only be manually checking in changes?