http://bugzilla.suse.com/show_bug.cgi?id=1086789
http://bugzilla.suse.com/show_bug.cgi?id=1086789#c19
--- Comment #19 from Fabian Vogt
(In reply to Fabian Vogt from comment #16)
(In reply to Yifan Jiang from comment #15)
Thank you all for the analysis!
If I understood the background correctly, we are considering to restrict the scale factor number of gnome-tweak-tools (or, perhaps gsettings) to be a multiple of 0.25.
No, you just need to make sure to set Xft.dpi to an integer number.
This way, I think it will impact the whole GNOME stack which had already the capacity to support the Xft.dpi as a floating number.
Probably by accident. There's no reason to use non-integer values there, the precision is already good enough. For instance: 96 DPI * 1.8 scale = 172.8 DPI. That's 0.46 % off. What counts is that the scale factor can be retrieved from the integer DPI again, which works fine: round(172.8) / 96 = 172 / 96 = 1,7916... -> rounded 1.8 Plasma supports arbitrary fractional scaling as well and does not use any floating point DPI anywhere. Works absolutely fine. As the alternative is to restrict the scaling factor to multiples of 0.25, just rounding Xft.dpi without any functional loss seems like the best solution.
Is there a conclusion of QT's upstream bug not to support this?
I left a comment that using non-integer values violates the specification: https://bugreports.qt.io/browse/QTBUG-64738?focusedCommentId=457316#comment-... Also note that the bug report already says that MATE and Xfce behave correctly.
btw. I am looking into possibilities before chingkai's hand is free ...
-- You are receiving this mail because: You are on the CC list for the bug.