On Wed, Feb 04, 2026 at 04:43:25AM +0100, Vincent Lefevre wrote: > On 2026-02-03 20:32:56 -0500, Thomas Dickey wrote: > > On Fri, Jan 16, 2026 at 12:19:04AM +0100, Vincent Lefevre wrote: > > > Package: xterm > > > Version: 406-1 > > > Severity: normal > > > > perhaps wishlist. > > No, as since the removal of utmp, this is now a required feature, as > said by Chris Hofstaedtler.
I forget who's Chris - some generally acknowledged authority on software
requirements? If so, you should mention the allusion by their role.
(wishlist, none the less - we've been here before too many times)
> > In a quick check, "xterm -ls" does run a "login" shell, but systemd
> > doesn't know about that (something to investigate).
>
> This should be independent from whether the shell is a login shell,
> anyway ("logind" is probably misnamed as the intent is to also
> handle the terminal sessions).
I don't see another route to getting the data into systemd's cache.
(its documentation is poor, but you may have something specific in mind)
I didn't notice this initially, but reviewing the (xterm) source see
that a failure to write utmp might interfere with running login.
That's something that I could improve, since fixing systemd is out of scope.
> > strace reminds me that xterm isn't able to write to utmp since that's
> > been removed. If it hadn't been removed, (again looking at strace),
> > "who" and "w" would see that information in addition to what it gets
> > via systemd's intentionally undocumented sessions data.
>
> "who" is unfortunately underspecified. "w -t" sees all the
> terminal sessions (but it does not use systemd for that, and
> in theory, the -t option should not be needed).
"who" doesn't set the data. It reads it. This bug report deals with
writing the data.
--
Thomas E. Dickey <[email protected]>
https://invisible-island.net
signature.asc
Description: PGP signature
