Fix sorting and fallthrough for tasks on all desktops.

Sorry for the late reply, too.

SortDesktop needs to fall through to alphabetic sorting within a desktop.

I'm OK with this, but I'd like @ngraham to verify consistency on the indent issue.

Needs another revision to tackle an additonal problem, as per the bug reporter's feedback in the referenced bug#.

Does the PlasmaWindowManagement object need to be guarded in the same way?

• Open groups in popups
• Only when the Task Manager is full

I'm ready to push this to master - are you ok with realname attribution (Władysław Wokulski)?

Screenshot would indeed be nice, but it's a promising idea.

Sorry for the delay! Somehow I kept missing this. Patch looks fine, do you need help merging it or do you have write access?

What's all the unrelated code changes about mouse handling trying to achieve?

Would it be possible to also add the lineart versions as -symbolic? In my new app I rely on Kirigami's icon recoloring feature and it only works with monochrome icons, so I rely on -symbolic to request them.

LGTM aside from the minor naming nitpick.

LGTM, thanks!

The patch currently doesn't apply cleanly against master anymore, mind rebasing on the next update?

Could this maybe be input-games-symbolic (too)? Because I want a monochrome version always (my app recolors icons using the Kirigami feature, and that only works with monochrome icons).

Remote clients have no pid, but might be nicer to fix the model to return 0.

Fix typo.

Screenshot showing ellision:

Thank you! Just what I needed.

In D18848#407843, @rooty wrote:

In D18848#407838, @hein wrote:

This is a good change, but it's more worth finding out why it doesn't work already. Looking at the code, when KickoffItem is triggered it runs the switchToInitial() function in FullRepresentation.qml, which already switches tabs and is also what sets the state back to Normal. So setting the tab in the state shouldn't be necessary. Can you debug why setting currentTab in switchToInitial() apparently doesn't work?

I think it might be because of the switch from the state "Search" to the state "Normal", because this doesn't happen if the arrow keys are used to navigate to the right of Favorites, only when Search is invoked. There's a "currentTab: searchPage" (line 738) setting in the state "Search" that probably gets stuck. I didn't want to deal with changing "Search" because in my experience, Kickoff's search view can be pretty temperamental. This seemed like the safest course of action.

This is a good change, but it's more worth finding out why it doesn't work already. Looking at the code, when KickoffItem is triggered it runs the switchToInitial() function in FullRepresentation.qml, which already switches tabs and is also what sets the state back to Normal. So setting the tab in the state shouldn't be necessary. Can you debug why setting currentTab in switchToInitial() apparently doesn't work?

❤

Hmm ... I guess it's OK, but using libtaskmanager would have been nicer maybe (centralized the timeout behavior, and the monitoring).

I suggested this, so it would be weird to say no.

D18307 seems to have done the trick in more ways than one. I no longer need to blacklist the kscreen2 kded module and can successfully log in. This is written in Plasma/Wayland on nVidia drivers, congrats! There's still some wonkyness that may be from who knows where - the splash screen no longer renders - but so far, so good.

Sorry for the noise. Please review again (I'm off till tomorrow).

Revert an unintended change, fixes positionertest

Add missing include

Signal-slot hygiene

Reorder abort checks and unset m_deferApplyPositions if going for a reset

In D18598#401848, @davidedmundson wrote:

If the QML code calls setPositions({}) and when we apply m_positions.size() < 5
m_deferApplyPositions gets left on true forever. Is that ok?

If I blacklist the kscreen2 kded module, I can get the session to start (otherwise it crashes during the splash). However, all surfaces (e.g. the panel or krunner) don't react to mouse clicks and seem frozen-ish.

I tried to give this a quick spin on my Quadro P2000 Max-Q (ThinkPad P1), but KWin dies shortly into the session splash screen. Have yet to debug why.

How about we actually rename it "Desktop Menu".

In D18517#399690, @zzag wrote:

Okay, it's good now.

Connect to right signal

I think it needs to be made a bit more robust still. For one it's not called requestClose() without reason: Not all windows are closable / will be closed upon this request. It might also race with a different window closing going on in parallel. taskClosedWithMouseMiddleButton needs to be invalidated on exit, and the middle-click close and onItemRemoved probably need to compare persistent model indexes or window ids (easier) to check it's the same.

Don't queue the connections.

I like the idea. The Firefox implementation is smart enough not to do it when a tab is closed by means other than direct user mouse interaction though.