On Fri, 08 Sep 2023 at 11:57:35 -0400, Jeremy Bícha wrote:
> On Fri, Sep 8, 2023 at 11:48 AM Simon McVittie <s...@debian.org> wrote:
> > GLib 2.77.3 made the error behaviour of g_key_file_get_string()
> > more conventional, triggering this regression. GLib 2.78.0 reverts
> > that change (see https://gitlab.gnome.org/GNOME/glib/-/issues/3094,
> > https://gitlab.gnome.org/GNOME/glib/-/issues/3095) and that
> > version will be uploaded to Debian unstable soon; but the change
> > is likely to be reinstated in a future version of GLib (see
> > https://gitlab.gnome.org/GNOME/glib/-/issues/3098).
>
> Once the known broken apps (mc and notmuch) are fixed, I think we'd
> want to land that change in Unstable sooner similar to what I did with
> glib 2.76.4-4 (shortlived because of the Debian Freeze but it was
> there longer for Ubuntu 23.10). Optimistically, that could help
> identify other broken apps and get them fixed before this lands in
> "stable" Linux distros.
Yes, I think that makes sense: while we're outside freeze, we want
changes with an impact on dependencies to happen sooner rather than
later, so that the impact can be found and fixed. Cc'ing the cloned mc
and notmuch bug reports to make sure the maintainers of those packages
are aware that this is likely to happen rather sooner in Debian than
the GLib upstream version number would necessarily indicate.
smcv
