Hacker News Reader

Comment by mcculley | Hacker News Reader

I have always wondered why the admin user is named "root". It makes sense for the "root" filesystem as filesystems are structured as a tree. But the set of users is flat. Even considering groups, there is no tree structure to group membership.

View on HN

People learning Linux find it confusing. They ask what root is and understand the concept of root of the file system. Then I say there is a user called root which confuses them. Then a group root only root is in it when the user root is created. But the two root words are not two groups one is the owner root (who is user root) it isn't the group.

I've been doing linux for awhile and that explanation is confusing to me, so I see where they're coming from.

I usually oversimplify: "root" is a keyword typically used to be all incompacing. The system root contains all files in the system. The user root is a super-admin that has all user rights in the system. The group root is a super-admin group the root user is assigned to and has all group rights.

"encompassing"

> The system root contains all files in the system.

Note that, strictly speaking, this is a misconception. The root file system is the one which is at the root of the hierarchy of mounted file systems.

Even the notion of something being (or not being) "in the system" can be misleading, because today's systems, including storage devices, can be distributed in various ways.

Or annoying, apparently. A good friend of mine’s father, having recently acquired a unix based POS system for his retail store, decided he’d prefer a nicer name than ‘root’. Apparently, ca. 1984 unix was rather particular about names (instead of uid) and didn’t take kindly to such. Much pain ensued.

Perhaps because you get write access to the root directory (and from there on down the whole tree), while normal users are kept within their respective home directories?

That means sense as user directories used to be kept in /usr.

The root user is the root of the permission tree. All power comes from and can be taken back by the root user. Everyone else has less power.

The first program was also traditionally executed under a root user. You can see this carry forward with init scripts running as root by default.

Root runs the root process. Processes form a tree.

There are no root "permissions", because permissions are not a tree structure. Users can't give permissions to other users, recursively.

They can in the temporal sense, the tree isn't tracked by software but it still originates that way. As the first account root creates the second and may decide to delegate other permissions. Given sufficient permission that account may too decide to create others and so on. If root rescinds permissions its not recursive which I think is what you are getting at.

Well, maybe not group membership, but certainly there is a 'root' to the tree of permissions and capabilities, and I think it's true that some of them derive from more basic capabilities.

If each capability (cross the namespace or object to which it is applied, for non-global capabilities) is a leaf, then the global capability itself is a branch from which all of those leaves have sprouted, and the set of all capabilities live in a tree...

There is Root at the base of the trunk uniting all of the branches.

Think about it like this, who owns the root of the file system? Someone has to own it so they named that user root. Though, I have also thought, well then why not have another user who owns all the administrative directories? I suppose it makes root redundant if all root does is own/manage the root of the file system. So root it is.

In Unix Version 7 (and it might also have been like this in previous versions), the home directory for each user was in /usr, so the home directory for a user named "david" would be /usr/david. The home directory for the "root" user was /.

In old Unix systems the home directory of UID 0 was effectively / so it made sense to call this user root.

Reply on news.ycombinator.com