Fixed a new case of high CPU usage with scaling

[tsujan]

… by removing LXQtTaskButton::refreshIconGeometry, for which I found no use. The high CPU usage was a result of a recent change — or perhaps bug — in KWindowSystem.

Fixes #1926

EDIT: As @palinek explained later, LXQtTaskButton::refreshIconGeometry is for WM minimizing animations, like KWin's "magic lamp".

[palinek]

by removing LXQtTaskButton::refreshIconGeometry, for which I found no use.

IIRC this was for the minimize/maximize animation (to allow WM do fancy animation beginning/ending in the taskbar button area).

[tsujan]

IIRC this was for the minimize/maximize animation (to allow WM do fancy animation beginning/ending in the taskbar button area).

But we know that WM animations are independent of any task-bar. Also, a WM determines where its title-bar icons are — in case this is about them.

It's strange to me that such a code doesn't have a single comment.

[palinek]

But we know that WM animations are independent of any task-bar.

Then WM can't do the minimizing effect "shrink into the taskbar icon".

WM determines where its title-bar icons are — in case this is about them.

This is not about determining, where the window icon is. But this is about information, where the taskbar icon is.

[tsujan]

Then WM can't do the minimizing effect "shrink into the taskbar icon".

Oh, thanks! Now I get it.

Without that code, KWin does it here when a window is minimized by clicking its task button (and that's what I do out of habit); otherwise, when the minimize title-bar button is clicked, the "magic lamp" animation shrinks the window into the cursor position. IMO, if we keep the code, we should add a short comment to it.

Anyway, the bug is nasty — it makes the panel unusable with scale factors. On the other hand, I like animations and don't want to ruin them ;) Leaving this PR open until a stable compromise is found... But it should be done before 1.4.0.

[tsujan]

Done in #1931