fvogt added a comment.
In D12405#257444 <https://phabricator.kde.org/D12405#257444>, @davidedmundson wrote: > The good: > > In one setup I was pleasantly surpised at how well it worked. Fonts did behave as expected. > Specatcle was able to capture all screens and individual rects correctly. > > The less good: > > When I switched round the primary monitor all through the GUI and rebooting I had a right mess. The @1x screen fonts were double the size, the @2x screen were 4x the size. > In this setup spectacle had an issue with "current screen" when it was on the non primary. I assume this problem: https://bugreports.qt.io/browse/QTBUG-64992 . That's possible - but unlikely: Except for the logical DPI difference the setups are identical. What happens with Xft.dpi: 96 and non-primary as 2x? > Possibly that's just a oneline fix wrt correct font DPI saving. Yup, that's the WIP part of this diff. Not sure how to do that though, as the code for this is in the frontend and not the (xrandr) backend. So if you unplug the primary monitor, newly started applications will behave strangely. But then again, this bug is in the current code as well: If you plug in a new monitor, the scale is 1 (never actually tried that). > Does this change make sense whilst Plasma is not using Qt scaling? It means that plasma uses the scale of the primary monitor (actually the font DPI, but that should be the same). So neither better nor worse than the current state. With Qt scaling, Plasma behaves as expected: Panels and menus scale depending on the monitor they're on. In exchange, some parts (like window tooltips) appear slightly misplaced. INLINE COMMENTS > davidedmundson wrote in outputconfig.cpp:143 > QPushButton* Indeed, how did that build? Maybe the diff actually failed to build but I never noticed. REPOSITORY R104 KScreen REVISION DETAIL https://phabricator.kde.org/D12405 To: fvogt, #plasma Cc: mart, hein, ngraham, graesslin, davidedmundson, plasma-devel, ragreen, Pitel, ZrenBot, lesliezhai, ali-mohamed, jensreuterberg, abetts, sebas, apol
