Martin Steigerwald said on Fri, 13 Jun 2025 10:07:02 +0200

>Hi!
>
>Did someone actually look what this new userdb stuff is about?
>
>https://systemd.io/USER_RECORD/

I like JSON. I like catsup too. But I wouldn't like catsup on dog
droppings, and I wouldn't like JSON on systemd.

>
>I did not read through all of it, but I noticed the length of that
>document.

And that says it all. Probably over 1000 words of that document are
used to describe what /etc/passwd and /etc/group already, and those
files can be described completely in a short paragraph.

>It is a bit like the Systemd network device naming.
>
>For RHEL they have a chapter in networking guide about it with 8
>sections. It think it was even more than 10 sections in earlier
>versions. A reason may be it it may be not supported to switch to the
>old style naming anymore so they may have dropped that from
>documentation.
>
>https://docs.redhat.com/en/documentation/red_hat_enterprise_linux/9/html/
>configuring_and_managing_networking/consistent-network-interface-device-
>naming_configuring-and-managing-networking
>
>But really: Are you even kidding me?
>
>I get that naming eth0 / eth1 / eth2 in (e)udev / mdev may give clash
>with kernel device naming as network drivers scan in parallel for
>network devices. But they overdid it. Quite so.
>
>There is such a plethora for a lot more simple approaches to network
>devices. The could have just gone "en1", "en2", "en3" and pin to MAC
>addresses on first appearance.

But noooo!

>So what do you think, did they overdo it with that userdb as well? Or
>do you think it is a good idea? Or somewhere in between?

It's a krappy idea, but the real krap is binding together all these
unrelated functionalities into one mess.

>
>I am worried about mixing all that complexity into low level software.

Me too, plus I don't like all the complexity in "userland" either. But
garbling the low levels is the worse problem.


SteveT

Steve Litt

http://444domains.com