It is recommended to impose additional restrictions on uninstalling the pkdbase package using pkg.

[ykla]

It is recommended to impose additional restrictions on uninstalling the pkdbase package using pkg. Otherwise, the same mistakes made on Linux will be repeated. Currently, running pkg delete pkg requires the -f flag, and I believe this should apply even more strictly to pkgbase packages. According to my testing, it is currently possible to uninstall the kernel using pkg without any obstacles, and after a reboot, the system is gone.

This issue is very common on Linux — inexperienced users often uninstall critical components like the kernel, glibc, or Python. This happens because Linux lacks a concept of a base system and has no mechanism to protect such essential software. FreeBSD, on the other hand, is different: it has a clearly defined base system.

 

[Espionage724]

Otherwise, the same mistakes made on Linux will be repeated. Currently, running pkg delete pkg requires the -f flag

I feel you'd have to want to do something like that, implying it isn't a mistake, implying it shouldn't be limited because inexperienced users might misuse it

Why would you want to run that command exactly?

 

[covacat]

pkg delete --allow-foot-shooting pkg [pkg] ...

 

[dkh]

pkgbase makes me nervous. I see a lot of potential issues coming our way. I don't think it's going to be as "friendly" as freebsd-update.

I recognize that something has to happen with freebsd-update but it's never given me any issues and little things like having to edit config files just to go between say 14.2-RELEASE to 14.2-RELEASE-p1 are going to be annoying.

I converted a vm to pkgbase and accidentally managed to get version bump to 14.3-BETA3 instead of 14.2-RELEASE-p2 as intended (it was my own fault but a non-obvious outcome and I've been using FreeBSD since sometime around 2.2).

 

[mer]

My opinions only.
I've never been completely sold on "pkgbase". FreeBSD has always been "kernel and userland should always be in sync" which is why if freebsd-update installs a new kernel you should reboot before freebsd-update to install userland.
Now I can accept the argument that "freebsd-update should really be nothing more than pkg upgrade commands on a correctly packaged base" but I think there are a bunch of details that either need to be worked out/finalized.

example: you are currently on freebsd-14.2-RELEASE-p1 but there is an update so you get 12.4-RELEASE-p2 available. freebsd-update understands if that's just a userland or just a kernel update, but with pkgbase do I have a kernel pkg 14.2-RELEASE and a userland pkg 14.2-RELEASE-p2?
Kind of what dkh talks about.
According to the "pkg" command a 14.3-BETA would be a new version over 14.2-RELEASE-p1 but we don't want that as the default. We want pkg to check for 14.2-RELEASE-p2.

And switching from one to the other as the default in the installer will cause more problems than the csh to sh switch.