Zum Hauptinhalt wechseln
Logo farbenmeer
Blog / Veröffentlicht am

The Power of the XDG Base Directory Specification

Portrait von Max Strübing
Autorin

Max Strübing, Softwareentwicklung

I'm writing this article because I have the feeling that not many people know about this convention even if it exists for years and can lead to a very clear directory structure and file paths instead of a cluttered home directory with thousands of files.

The issue?

Everyone knows that feeling, you type a ls -la in your home directory, or something similar which lists also hidden files and you get back huge list of files where you have no idea what it is, where it came from and what it does. The problem is that most programs simply use your home directory as a data storage for any kind of data. That leads to exactly this problem, no one has an overview of their home directory anymore. When I run ls -la | wc -l I get 109 files and/or directories and I even clean it up regularly.

XDG Base Directory Specification to the rescue

The XDG Base Directory Specification is a widely used specification which specifies where your files should be located depending on their usage published by the freedesktop.org-organization.

But what is freedesktop.org?

freedesktop.org, formerly known as X Desktop Group (XDG), is an organization which has many open source projects including the X Window System, Wayland or systemd.

What does the specification say?

They published a specification which mainly consists of three important points:

  • Write user-specific data in $XDG_DATA_HOME

  • Write configuration files in $XDG_CONFIG_HOME

  • Write cache files in $XDG_CACHE_HOME

If these environment variables are not set, the specification says it should go in these directories:

  • $XDG_DATA_HOME should be $HOME/.local/share

  • $XDG_CONFIG_HOME should be $HOME/.config

  • $XDG_CACHE_HOME should be $HOME/.cache

They are also writing about other things like $XDG_RUNTIME_DIR which is used for communication and synchronization, $XDG_CONFIG_DIRS which is used to indicate where configuration files should be searched and $XDG_DATA_DIRS which is used to search for data files. But for me, the main points are the three from above. What you will do is you will read the environment variable, if it is present use this directory or use the fallback, create a directory with your program name in it, it should be relatively unique to not clash with programs of the same name. Save data there.

  • Read environment variable or use fallback

  • Create a directory with your program name in this directory

  • Place your caching/data/configuration files there.

As this is very common among Unix systems and also partly adapted in Windows (at least as far as I know) you really should consider using it. This has the advantage that if you search for configuration files for program x you will look either in your custom set $XDG_CONFIG_HOME/x or in $HOME/.config/x, because you know it will be there. Similarly, you will know where data files (sound files, datasets, save points, ...) or caching files will be located.

So, if you will ever write a tool which needs some configuration files, caching files or data files remember this specification, look it up again and implement a proper way to handling these cases rather than cluttering the $HOME of your users. If don't want to do it yourself you could also use a library for that, there is a library for every major programming language out there.

The full specification can be read here: https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html


Umfrage

Welches Thema interessiert dich am meisten?