Re: [PATCH 3/3] readutmp: load libsystemd with dlopen instead of linking
Bruno Haible via Gnulib discussion list <[email protected]> Mon, 20 Jul 2026 01:02:48 +0200
| Newsgroups | gmane.comp.lib.gnulib.bugs |
|---|---|
| Message-ID | <3596751.AOvM4ru3NT@cagnes> |
Hi Paul,
Paul Eggert wrote:
> > There's no static checking or complaining involved anywhere that I know of.
>
> Then I'm lost. I thought a major goal of this effort was to pacify
> Debian installation procedures for coreutils on limited installations
> that don't have libsystemd installed.
Sort of. I explained the goal of Luca's readutmp patch in this mail:
https://lists.gnu.org/archive/html/bug-gnulib/2026-07/msg00065.html
As you and I have agreed that this effort is based on an incorrect
understanding of the coreutils package and its dependencies (and
Pàdraig seems to agree [1]), we can consider this patch (now
'[PATCH v2 5/5] readutmp: load libsystemd with dlopen instead of linking'
in his latest patch set) as rejected.
What to do instead, in my opinion, is detailed in
https://lists.gnu.org/archive/html/bug-gnulib/2026-07/msg00098.html
I will submit two proposed meta-information files to coreutils:
- A DEPENDENCIES file, that states the dependencies (both for
the user who runs './configure && make && make install', as
well as for the distros).
- A PACKAGING file, that gives suggestions to distros, especially
regarding how to create minimal environments without libsystemd.
Bruno
[1] https://lists.gnu.org/archive/html/bug-gnulib/2026-07/msg00037.html