Deprecate and remove the `groupmems` utility

[ikerexxe]

I propose that we deprecate the groupmems utility in an upcoming release, with the goal of removing it completely in a subsequent release (e.g., 2-3 releases later).

The utility's use cases are split between root and regular users, and it fails to provide significant value for either.

• For root users, the functionality of groupmems is largely redundant. Administrative tasks like managing group memberships can be accomplished with more common and powerful tools like usermod. Removing it would not cause a loss of functionality for administrators.

• For regular users, the utility is designed to manage their own primary group membership. However, this requires the binary to be setuid, which presents a security risk and is not the default behavior in major distributions. This leaves the tool non-functional for its primary non-root use case.

The intended functionality for regular users is already broken or disabled in several key distributions:

• Fedora: the man page explicitly states that groupmems is not setuid, rendering it unusable for its intended non-root purpose.

In this operating system the groupmems executable is not setuid and regular users cannot
use it to manipulate the membership of their own group.

• Debian: the situation is confusing. The binary is not configured in a way that allows regular users to manage their groups, a situation highlighted in this bug report: https://groups.google.com/g/linux.debian.bugs.dist/c/hDArHoYe9Dk

• openSUSE: as noted in issue #999, openSUSE has already removed the groupmems binary, setting a precedent for its removal.

Proposed plan

• Next release: mark groupmems as deprecated in the release notes.

• Future release (2-3 releases later): remove the groupmems binary and its associated man page entirely.

Given that the utility is redundant for root and effectively broken for regular users across major distributions, its continued maintenance adds complexity for little to no benefit.

I'm opening this issue to gather feedback on this proposal. Are there any objections or use cases I might have missed?

[alejandro-colomar]

* Debian: the situation is confusing. The binary is not configured in a way that allows regular users to manage their groups, a situation highlighted in this bug report: https://groups.google.com/g/linux.debian.bugs.dist/c/hDArHoYe9Dk

Do we have groupmems(8) in Debian? I can't find it:

alx@debian:~$ apt-file find bin/groupmems
alx@debian:~$  

Maybe it has been removed already here? (I'm running Debian Sid.)

@zeha

[zeha]

* Debian: the situation is confusing. The binary is not configured in a way that allows regular users to manage their groups, a situation highlighted in this bug report: https://groups.google.com/g/linux.debian.bugs.dist/c/hDArHoYe9Dk

Maybe it has been removed already here? (I'm running Debian Sid.)

Indeed I removed it for trixie. Per my understanding our previous situation was useless to non-root, and root does not need the tool.

[alejandro-colomar]

Maybe we can remove it at once, without deprecation? Users can just keep running a previous version of shadow if they need it for some time. Considering that two of the three Big Distros have already removed it, we're not changing much.

[ikerexxe]

Indeed I removed it for trixie. Per my understanding our previous situation was useless to non-root, and root does not need the tool.

That sounds perfect, and it makes perfect sense.

Maybe we can remove it at once, without deprecation? Users can just keep running a previous version of shadow if they need it for some time. Considering that two of the three Big Distros have already removed it, we're not changing much.

Let's take it step by step, as other distributions may be affected and this gives them time to prepare their documentation so that users stop using this binary.

[alejandro-colomar]

Okay; that sounds reasonable.