This makes sense for configs that could ostensibly do this, but that's only a fraction of them. Regardless of what we do about those (if anything), the other ones at least shouldn't be laid out like this.
This makes sense for configs that could ostensibly do this, but that's only a fraction of them. Regardless of what we do about those (if anything), the other ones at least shouldn't be laid out like this.
The list of configuration files/directories is known. Why can't those who bother present them as they desire? Instead of forcing their will on others.
There reason for ~/.config is separation from data and cache. And the reason behind XDG Base Directory Specification is a generic way to run application with another config `XDG_DATA_HOME=foo bar`.
Who/what is the "we" here? Last I checked my homefolder was still chock full of dotfiles in-use.
Similarly, things also like to hide state in ~/.cache
If they didn’t then rm -rf .appname would restore it to default settings, and we can’t have that.
This is the sort of issue that leads to “this page intentionally blank” type use of resources.
“When everything is a hammer...” when all you have to work with is a file system it always looks messy.
I’m more interested in making sure the right values exist in memory for runtime use.
Source, config, and comments could all live in one file and be specially delimited for look up too.
It’s all arbitrary at that point. This is really focusing on the wrong issues in computing.
One dotfile has the maintainer's eccentric syntax highlighting preferences, another has critical build system settings needed to make any use of the code. Which ones do I really need to knand which are just conveniences for users of a particular IDE?
There are very few pieces of information where, to me, the natural place to look for it is "a dotfile at the root of the project." gitignore is an exception here.