Bug 1226103

Confirmed this is still an issue with the KDE frameworks update from this week, also checked against neon testing and unstable, neither have the bug.

I said earlier that this did not appear to be specific to the Task Manger, my apologies, this is false. It effects both running apps and also launcher icons, but does not appear to effect other areas of the panel such as the notifications.

Works here, using Plasma Wayland on TW 20240611.

(In reply to Fabian Vogt from comment #2)
> Works here, using Plasma Wayland on TW 20240611.

Thanks for looking at this!

Hmm, odd. I just now tested it and it failed a VM running openSUSE-Tumbleweed-KDE-Live-x86_64-Snapshot20240605-Media.iso.

I neglected to mention that I also tested the Krypton live iso (downloaded about an hour ago) and that also had the same behaviour.

I'll try 0611 and see if anything changed. Do you have any idea why this might happen in a stock VM but you can't replicate it?

(In reply to pallas wept from comment #3)
> (In reply to Fabian Vogt from comment #2)
> > Works here, using Plasma Wayland on TW 20240611.
> I'll try 0611 and see if anything changed.

Downloaded and booted a VM with https://mirror.2degrees.nz/opensuse/tumbleweed/iso/openSUSE-Tumbleweed-KDE-Live-x86_64-Snapshot20240612-Media.iso and the issue persists as above.

(In reply to pallas wept from comment #3)
> (In reply to Fabian Vogt from comment #2)
> > Works here, using Plasma Wayland on TW 20240611.
> 
> Thanks for looking at this!
> 
> Hmm, odd. I just now tested it and it failed a VM running
> openSUSE-Tumbleweed-KDE-Live-x86_64-Snapshot20240605-Media.iso.
> 
> I neglected to mention that I also tested the Krypton live iso (downloaded
> about an hour ago) and that also had the same behaviour.
> 
> I'll try 0611 and see if anything changed. Do you have any idea why this
> might happen in a stock VM but you can't replicate it?

Not really. Maybe it's because the VM uses an absolute input device (tablet) compared to a relative input device (mouse).

(In reply to Fabian Vogt from comment #5)
> Not really. Maybe it's because the VM uses an absolute input device (tablet)
> compared to a relative input device (mouse).

This issue originally occurred on my baremetal machine and forced me to roll back to 0531. I just delivered the VM as a test case because it eliminates user error / machine-specific stuff. The same exact VM boots neon (selected from boot menu at startup, otherwise identical) and doesn't exhibit this behaviour. 

Is there anything I can do here?

(In reply to pallas wept from comment #6)
> (In reply to Fabian Vogt from comment #5)
> > Not really. Maybe it's because the VM uses an absolute input device (tablet)
> > compared to a relative input device (mouse).
> 
> This issue originally occurred on my baremetal machine and forced me to roll
> back to 0531. I just delivered the VM as a test case because it eliminates
> user error / machine-specific stuff. The same exact VM boots neon (selected
> from boot menu at startup, otherwise identical) and doesn't exhibit this
> behaviour. 
> 
> Is there anything I can do here?

You could check upstream (KDE, Qt) for similar bug reports. Maybe it's a regression in Qt or something...

(In reply to Fabian Vogt from comment #7)
> (In reply to pallas wept from comment #6)
> > (In reply to Fabian Vogt from comment #5)
> > > Not really. Maybe it's because the VM uses an absolute input device (tablet)
> > > compared to a relative input device (mouse).
> > 
> > This issue originally occurred on my baremetal machine and forced me to roll
> > back to 0531. I just delivered the VM as a test case because it eliminates
> > user error / machine-specific stuff. The same exact VM boots neon (selected
> > from boot menu at startup, otherwise identical) and doesn't exhibit this
> > behaviour. 
> > 
> > Is there anything I can do here?
> 
> You could check upstream (KDE, Qt) for similar bug reports. Maybe it's a
> regression in Qt or something...

Done, none existed (fortunately it's a recent regression so not many to filter through). I don't seem to be able to replicate it beyond opensuse, though, which is why I logged it here (I normally log upstream if I can replicate it for them)

I've now had the opportunity to test the latest snapshot across 4 separate baremetal machines, all had this same bug, and while TW's libs are a newer version and those libs may be the problem, I can't replicate it anywhere but OpenSUSE, so I can't replicate it to push it upstream. 

I'm puzzled how you managed not to replicate it since I can't avoid it. Intel or AMD CPU, nvidia or intel GPU, laptop or desktop, virtual or metal, trackpad or mouse, all do this. Maybe you tested on your running installation and have some magic setting which works around this bug and does not exist on the snapshot? Maybe you tried X11 by accident? (I've done that)

Is there anything we can do to sort this out? I'm trapped on 0524 by this (and a kernel bug, but that's OT). If you're able to figure out why you're not effected but everything else is, I'd love to know that workaround! :)

Great news, others have repro'd this outside of opensuse: https://old.reddit.com/r/kde/comments/1dvoadm/weird_issue_while_moving_icons_in_task_manager/

Apparently it's a widely known bug and I'm the only one bothering to report it...

Closing this in lieu of raising the bug upstream.

See: https://bugs.kde.org/show_bug.cgi?id=489809