Created attachment 139043 [details] Glitch SUMMARY STEPS TO REPRODUCE 1. Expand the system tray (^); 2. Open an applet; 3. Press back. OBSERVED RESULT There will be a visual glitch in the left corner. EXPECTED RESULT SOFTWARE/OS VERSIONS Operating System: KDE neon Unstable Edition KDE Plasma Version: 5.22.80 KDE Frameworks Version: 5.83.0 Qt Version: 5.15.2 Kernel Version: 5.4.0-73-generic (64-bit) Graphics Platform: X11 Processors: 4 × Intel Core Processor (Skylake, IBRS) Memory: 3.8 GiB of RAM Graphics Processor: llvmpipe ADDITIONAL INFORMATION
Weird, cannot reproduce.
I can reproduce it. I think it might be related to this bug as well: https://bugs.kde.org/show_bug.cgi?id=438955 I'm seeing those strange borders in many applets, sometimes they flicker, sometimes when moving the mouse over them they disappear. Disabling the contrast effect doesn't fix the issue to me. I'm using Intel graphics, but the problem is present even with XRender compositing.
There appears to be something odd going on with SVG caching. We are seeing other related glitches too. Moving to plasma framework.
Related to 433934, possible the same root cause?
I noticed something interesting with this bug. It doesn't happen when the panel is set to opaque (or adaptive with a maximized window), only when set to translucent. It also happens on some applets, especially the clipboard one. If you click on it you'll see the glitch, but if you hover the mouse over the applet, the glitch disappears. Since adaptive is the default behavior and this is a pretty obvious bug, I believe it could be made part of the 15-Minute Bug Initiative Nate posted about on his blog: http://tinyurl.com/kdeplasma-15-minute-bugs
I can reproduce that visual glitch with a bottom panel and transparent panel mode. I wasn't noticing it because I use a vertical panel, and it doesn't reproduce there. Further information: it only seems to reproduce for applets that have a header: set. Applets that have no header of their own are unaffected. The Bluetooth applet proves it. When Bluetooth is on, it has a header and the glitch happens; when BLuetooth is off, there's no header and the issue goes away. I don't think it qualifies for the 15-Minute Bug initiative as it is a truly very small minor thing. We should still fix it of course. :) Now that I can reproduce the bug, it shouldn't be hard to fix.
Another interesting data point: the Networks applet suffers from it, but the glitch goes away automatically as soon as the network list updates. Seems like the initial state of something isn't set properly, but it has a binding that makes it get reset properly when the applet refreshes or something.
Found the bug, submitting a fix shortly. Behold, the power of a 100% reproducible issue. :)
A possibly relevant merge request was started @ https://invent.kde.org/plasma/plasma-workspace/-/merge_requests/1379
Git commit 7221ffd1a9aeeae52467ac4b3bc3a5bfcde7a500 by Nate Graham. Committed on 19/01/2022 at 18:06. Pushed by ngraham into branch 'master'. applets/systemtray: fix glitch for some applets with headers PlasmoidPopupsContainer.qml has special code to handle merging headers from applets if those applets use a PlasmaComponents3.Page instance as their FullRepresentation. However this breaks for any applets that use PlasmaExtras.Representation, which is a subclass of it that we have been porting applets to over time, because it offers more features in a single location. This caused the detection to break, so such applets woudl have their headers improperly merged. Explicitly detecting the use of PlasmaExtras.Representation as well fixes it. FIXED-IN: 5.24 M +2 -1 applets/systemtray/package/contents/ui/PlasmoidPopupsContainer.qml https://invent.kde.org/plasma/plasma-workspace/commit/7221ffd1a9aeeae52467ac4b3bc3a5bfcde7a500
Git commit 17a189405f5b19e1a8bf2499ff334ce525503634 by Nate Graham. Committed on 19/01/2022 at 18:29. Pushed by ngraham into branch 'cherry-pick-7221ffd1'. applets/systemtray: fix glitch for some applets with headers PlasmoidPopupsContainer.qml has special code to handle merging headers from applets if those applets use a PlasmaComponents3.Page instance as their FullRepresentation. However this breaks for any applets that use PlasmaExtras.Representation, which is a subclass of it that we have been porting applets to over time, because it offers more features in a single location. This caused the detection to break, so such applets woudl have their headers improperly merged. Explicitly detecting the use of PlasmaExtras.Representation as well fixes it. FIXED-IN: 5.24 (cherry picked from commit 7221ffd1a9aeeae52467ac4b3bc3a5bfcde7a500) M +2 -1 applets/systemtray/package/contents/ui/PlasmoidPopupsContainer.qml https://invent.kde.org/plasma/plasma-workspace/commit/17a189405f5b19e1a8bf2499ff334ce525503634
Git commit 4f6c096beebad435bcc9857e73bde32755cf5de6 by Nate Graham. Committed on 19/01/2022 at 18:29. Pushed by ngraham into branch 'Plasma/5.24'. applets/systemtray: fix glitch for some applets with headers PlasmoidPopupsContainer.qml has special code to handle merging headers from applets if those applets use a PlasmaComponents3.Page instance as their FullRepresentation. However this breaks for any applets that use PlasmaExtras.Representation, which is a subclass of it that we have been porting applets to over time, because it offers more features in a single location. This caused the detection to break, so such applets woudl have their headers improperly merged. Explicitly detecting the use of PlasmaExtras.Representation as well fixes it. FIXED-IN: 5.24 (cherry picked from commit 7221ffd1a9aeeae52467ac4b3bc3a5bfcde7a500) M +2 -1 applets/systemtray/package/contents/ui/PlasmoidPopupsContainer.qml https://invent.kde.org/plasma/plasma-workspace/commit/4f6c096beebad435bcc9857e73bde32755cf5de6
*** Bug 445660 has been marked as a duplicate of this bug. ***
I see a fix was merged two days ago, but I can still see the bug on Neon Unstable with the latest updates. The version of plasma-workspace I have is 4:5.23.5+p20.04+tunstable+git20220122.0442-0 I assume that's the latest version, no?
Is it possible you're seeing a slightly different visual glitch from the one that was fixed here? There might be more things to fix beyond just the single narrowly targeted fix I implemented.
(In reply to Nate Graham from comment #15) > Is it possible you're seeing a slightly different visual glitch from the one > that was fixed here? There might be more things to fix beyond just the > single narrowly targeted fix I implemented. It's the same glitch. The only difference is that now the glitch disappears after a second on the clipboard applet just like it did before on the network one. Before your fix, it would linger there until hovering the mouse over the clipboard applet.
Created attachment 145759 [details] I made a short video showing what it is like now. I made a short video showing what it is like now.
I'll be reopening this bug since it's not fixed yet. I also noticed the glitch in the network applet doesn't disappear after a second like before the patch. Since your patch made the behavior change I believe it changed something important.
I can see now that you're right. I think my fix was not wrong, but not sufficient. I'll have a closer look.
I noticed another thing about this bug. As commented earlier, if you make the panel opaque it goes away, however, if you log in again the glitch still happens even if the panel is still set to opaque. Somehow changing the transparency fixes the bug temporarily but it's not the transparency itself that causes the problem. I'm just speculating, but maybe changing the transparency setting resets something until you start a new session. I also noticed that in applets with a checkbox, like Bluetooth, Battery, and Notifications, the glitch stays on screen until you interact with the applet, while on the ones with an input field like Clipboard or Network it disappears after a second. You can see that by making all applets visible and clicking on them. This behavior is pretty consistent.
I have an update on this bug. I know this is a very minor bug but small glitches like these are really bad for first impressions, so I'm trying to find clues as to what is happening. I noticed that if you set a color scheme that doesn't set a header color, like Breeze Classic, the bug goes away completely. The act of coloring the header must be causing this.
I'm not seeing this glitch anymore as of the last few updates to Neon Unstable (tested with X and Wayland). Can anyone with an updated Plasma confirm the bug is solved?
Also cannot reproduce. Likely it got fixed by the panel refactor, or Ivan's widget popup work. Awesome!