Okay, I sacrificed decibel-audio-player and volti (wasn't using them
long ago anyway).
But it still breaks ruby-gobject-introspection. Sid has a newer version,
which depends on some non-existent libgirepository-1.0-1-with-libffi8
though. So it's probably better to wait until the dependencies settle down.
On 23.05.2022 18:00, Simon McVittie wrote:
On Mon, 23 May 2022 at 16:59:55 +0300, Bohdan Horbeshko wrote:
libgirepository-1.0-1: 1.72.0-1+b1 breaks python-gi (<< 3.42.0-1+b1)
Yes, that is a true statement. If python-gi still existed in Debian,
then you would not be able to use new versions of libgirepository-1.0-1
together with old versions of python-gi, because old versions of
python(3)-gi were compiled against libffi7 (or older), but the new
libgirepository-1.0-1 exposes libffi8 in its ABI. The changes to struct
sizes between libffi7 and libffi8 would cause python-gi to crash on at
least some CPU architectures.
Breaks: python-gi (with no version constraint) would have the same
practical effect, except that if someone somehow brings back Python 2
support in pygobject, a non-versioned Breaks would require more
coordination.
If you have legacy code that requires python-gi, I would recommend
using it via a container or chroot (perhaps a Docker, Toolbx or schroot
environment with Debian 11 or older, or perhaps Flatpak or similar),
rather than on a host system that needs to run an up-to-date version
of Debian.
smcv
