* Michael Stone <mst...@debian.org> [250403 23:42]:
On Thu, Apr 03, 2025 at 07:52:12PM +0200, Marco d'Itri wrote:
On Apr 03, Michael Stone <mst...@debian.org> wrote:
The issue isn't making a change, the issue is what change is the
right thing to do. IMO, dropping utmp without any kind of a
transition or deprecation period is the wrong thing to do. Hence
this thread.
I think it's a bit late now to disagree with the plan implemented
last year by multiple maintainers.
Except, of course, for the primary consumer of utmp...
I'm the one who gets the complaints that who isn't working right, and
there isn't a solution to that problem, since the systemd facility
doesn't provide the same information.
From what I got from the coreutils bug is that coreutils upstream
has the relevant code and it's supposed to work.
Maybe it needs some additional work to fully function
with current systemd versions. IIRC procps also only recently added
some new code to deal with new systemd behaviours.
But the thing that needs looking at is why who in Debian behaves
like it does, and if it doesn't work right, fix that in who or
wherever else it needs fixing in the stack.
Reintroducing utmp will just hide the problem, and we'll be in the
same situation when releasing forky, or forky+1, ...
Chris
PS: I never understood why there's both w and who, and why they are
different implementations.
