[PATCH 20/52] login: Add 64-bit time support to utmp/utmpx

[Florian Weimer]

* Joseph Myers:

> On Tue, 27 Apr 2021, Florian Weimer via Libc-alpha wrote:
>
>> The interface was fine when a system had a few fixed terminal lines that
>> weren't equivalent.  But I think it's mostly useless today.
>
> At a higher level, I think it's still useful to be able to list logged-in 
> user sessions (and e.g. how long they have been idle) on a multi-user 
> system, to help identify e.g. users to consult about a possible reboot.  
> In the absence of utmp, you need some other system for tracking such 
> information about sessions, and applications such as "w" and "who" need to 
> be taught to use such other API in place of utmp, before it can be 
> considered obsolete.

True, but even for that, it's been reported (I forgot where I'm afraid)
that the ut_id field is to short if the straightforward PTY encoding is
used.  Most users are probably interested in sessions (who is not logged
in), not a list of PTYs and users (which also includes sub-shells in
Emacs and screen).

Thanks,
Florian