automated tests look good, let's push this and see where we're at.
There were errors in every part of this, unfortunately. I've fixed desktop-backgrounds , and sent an upstream PR for f32-backgrounds. As it stands it doesn't work properly at all. Systems which still have f31-backgrounds installed will still show the F31 background (on KDE and GNOME). I'm not entirely surely what'll happen on systems without f31-backgrounds installed, but they won't show the F32 backgrounds.
I'm now gonna +1 the update I just edited, as I only changed one package and it works, I swear. :)
Been running it for a day, no new issues so far.
Working fine aside from https://bugzilla.redhat.com/show_bug.cgi?id=1809107#c10 .
This update is inter-dependent with this pam update, which is not allowed by policy. The new authselect and the pam it depends on should be in the same update, or this update should not have been submitted until after the pam update went stable.
This is why most of the openQA tests fail.
Seems to have an issue with styling of the buttons on the update page:
Seems to have an issue with styling of the buttons on the update page:
I've been running a previous scratch build version of the fix, but just rebooted with this and things still look good so far.
Note this includes a change which can potentially break existing configs. If you have a config that inherits from a Fedora 31, 32 or Rawhide config, or a RHEL 8 config, and adds to the yum.conf
section - something like this, from openQA, which does exactly this:
include('/etc/mock/fedora-31-x86_64.cfg')
config_opts['plugin_conf']['bind_mount_enable'] = True
config_opts['plugin_conf']['bind_mount_opts']['dirs'].append(('/opt/update_repo', '/opt/update_repo'))
config_opts['plugin_conf']['bind_mount_opts']['dirs'].append(('/dev/ttyS0', '/dev/ttyS0'))
config_opts['yum.conf'] += """
[advisory]
name=Advisory repo
baseurl=file:///opt/update_repo
enabled=1
metadata_expire=3600
gpgcheck=0
"""
it will suddenly stop working with this version of mock, with an error KeyError: 'yum.conf'
. This is because of this commit, which changes these configs from having a yum.conf
section to having a dnf.conf
section.
These updates do seem to be somewhat against the Fedora and EPEL update policies, which discourage shipping major releases with incompatible changes as updates for stable releases.
I believe this has a regression in libdnf which causes PackageKit crashes (as some folks also noted on the F31 update, but didn't -1 it so it went stable :<). The crash is https://bugzilla.redhat.com/show_bug.cgi?id=1795004 . This is frequently happening in the openQA Cockpit update test for F31 and Rawhide. I'll -1 this so it doesn't go stable for F30 at least until this is investigated.
openQA tests all look good now.
openQA failures are due to UI changes requiring needle updates. I'm working on that now (over very slow conference wifi :<)
openQA failures are due to UI changes requiring needle updates. I'm working on that now (over very slow conference wifi :<)
This wasn't pushed stable? It was obsoleted. For Rawhide I spotted some later fixes and backported those too...hopefully it should be OK.
Done.
https://pagure.io/389-ds-base/pull-request/50815 should fix the problem.
ah, no, it's just a straight-up error. src/lib389/lib389/instance/setup.py
added some uses of ensure_list_str
, but neglected to import it from lib389.utils
, where it's defined.
Changes look good to my eyeballs.