Paul Eggert <[email protected]> writes:
> On 2026-04-18 15:40, Arsen Arsenović wrote:
>> Paul Eggert <[email protected]> writes:
>>
>>> On 2026-04-18 11:29, Collin Funk wrote:
>>>
>>>> The issue I have with this solution is that it requires the 'who'
>>>> program to maintain an ever expanding list of user names used by desktop
>>>> managers.
>>>
>>> Maybe filter them out if their login shell ends in "/nologin"?
>>>
>>>> Also, I guess this is mostly theoretical, but you would want to show
>>>> 'gdm-greeter' if a user actually named themselves that.
>>>
>>> The "/nologin" hack should work in that weird case.
>> Why not check for class == user?
>
> Oh, I never knew about sd_session_get_class. Yes, it sounds like
> Gnulib's readutmp.c should use this function to filter out bogus
> entries.
>
>
>
>> There's clearly already attempts to do this in the code, so it's worth
>> debugging what goes wrong there. Sorry if I'm missing something
>> obvious.
>
> Currently the Gnulib code checks only for class being "manager". I
> wonder why it does that? Anyway, it appears that 'who' should output
> only "user" entries.
Looking at this snippet of code:
if (seat != NULL)
a = add_utmp (a, options,
user, strlen (user),
session, strlen (session),
seat, strlen (seat),
host, strlen (host),
leader_pid /* the best we have */,
ctype, start_ts, leader_pid, 0, 0);
if (tty != NULL)
a = add_utmp (a, options,
user, strlen (user),
session, strlen (session),
tty, strlen (tty),
host, strlen (host),
leader_pid /* the best we have */,
ctype, start_ts, leader_pid, 0, 0);
I'm not very familiar with the concept of seats. Could we just check the
entry following each seat to see if it has a user entry associated with
it? If it does, skip printing the seat? Something like this:
for each utmp_entry:
if utmp_entry[0] == "seat" and utmp_entry[1] == "user":
# Don't print the seat if we have a user.
continue
print (utmp_entry[0])
Collin
