publicimageltd / configbundle

Organize configuration files in bundles

Configbundle

DISCONTINUED

After several changes and revisions, I was close to release a first "workable" version. And then I discovered stow, which does exactly what I envisioned, and has a much better way of 'stowing away' (bundling) configuration files. They found a way to solve the conceptual problem of how to track the "bundled" files and their targets which needs no "backlinks". Instead, they just store the configuration files in a directory structure which mirrors the target directory tree. This is really much more robust than my approach and simply a very, very good idea. In particular, this approach eliminates all risk of losing data, since all information about the configuration files (their target location) is hard-coded into the directory structure of the configuration bundle. This is transparent, solid, and simply good,

So for the moment being, I call it a day and leave the repository as it is for now. I did some Python exercises, so that's fine for me, but don't use this script.

Project description

Organize configuration files in bundles.

CLI tool to bundle configuration files in a directory, replacing the original files with symlinks. Bundling the file in a separate directory allows the user to treat configuration files which are spread all over the file system as one meaningful unit, e.g. by keeping the bundle under version control or exporting it as an archive for backup.

State of development

See the introductory note above. If one would want to revive the script, there is currently still a bug in the way the "unbundling" works (see the comments in the code). Since this bug is caused by my original approach to store a backlink with every file, this bug is actually an expression of a bad conceptual approach and actually need not be.

About the repository

The idea of handling links in this way is heavily inspired by Pass, which is a shell script which delegates all the bookkeeping logic to the file system. I really liked that idea and wanted to implement something like this, but not as a shell script. I chose Python to have an opportunity to learn how to write a Python script with modern tools, including virtual environment, testing and all. Part of this is to install the script as an executable binary, without having to activate the development venv all the time.

So here's the setup:

• Develop the script in a venv managed by Poetry.
• Use Pytest for testing.
• Install the Python script locally with pipx