For the last several Rawhide and F28 composes, the KDE default panel has appeared at the top of the screen, in a clearly messed-up form, like this: https://openqa.fedoraproject.org/tests/196071#step/desktop_notifications/46 The main panel has only a burger icon at the right end, and there's a weird translucent...mini-panel?...with just the word 'Default' or something on it, underneath it, at the top left hand corner. This makes using KDE out of the box pretty difficult, obviously. I can't yet tell you when this broke, as before this, the openQA tests were failing on something *else* (KDE failing to start, for a while, and before that, SDDM autologin not working on live images). So I can't pinpoint when exactly this started happening without some manual work. Proposing as a Beta blocker per criterion "No part of any release-blocking desktop's panel (or equivalent) configuration may crash on startup or be entirely non-functional."
Thanks, +1 blocker I'll be making an attempt at installing one of the recent f28 images locally to make debugging more fruitful. I'm really hoping it's something simple and not some gcc8-makes-something-to-wonky issue
Rex: you'll want to use the updated anaconda that actually runs on lives, then: https://bugzilla.redhat.com/show_bug.cgi?id=1551096 anaconda-28.22.2-1
Discussed at blocker bug meeting [1]: AcceptedBlocker (Beta) - clear violation of "No part of any release-blocking desktop's panel (or equivalent) configuration may crash on startup or be entirely non-functional", as several intended default elements of the panel, including key ones like the app launcher, are missing or invisible [1] https://meetbot-raw.fedoraproject.org/fedora-blocker-review/2018-03-05/
Looks like qt5-qtscript is getting miscompiled badly somehow, related to bug #1550245 It's used to evaluate javascript code for layout (and lots of other things)
qt5-qtscript: %changelog * Mon Mar 05 2018 Rex Dieter <rdieter> - 5.10.1-3 - support %%bootstrap, %%check: add autotests - build with -O1 to workaround serious autotest/code failures (f28+, #1551246)
I can confirm after installing qt5-qtscript-5.10.1-3 into f28 live session, stopping plasmashell, clear config, restart plasmashell results in a session working as expected: 1. alt-f2: kquitapp5 plasmashell 2. rm ~/.config/plasma-org* 3. alt-f2: plasmashell
This made it in before freeze, so setting ON_QA. When we get working composes (yup, composes are bust again, thanks systemd!), we can see if this is fixed.
Confirmed fixed in current F28 composes.