On Mon, 19 Feb 2018 12:10:02 +0100 Vincent Lefevre <vinc...@vinc17.net> wrote:
> On 2018-02-18 15:25:35 -0300, Lisandro Damián Nicanor Pérez Meyer wrote: > > wm4, Vincent: can you try setting QT_AUTO_SCREEN_SCALE_FACTOR=0 ? > > QT_AUTO_SCREEN_SCALE_FACTOR=0 works with -12, just like with -10. > > I suspect that different behaviors are due to the fact that among > HiDPI cases, screen sizes, DPI's, etc. are different. > > The patch consisted in truncating instead of rounding to nearest, > so that it might have solved the problem for users where the > real factor was between 1.5 and 2, thus changed from 2 to 1 > when converted to integer, 1 meaning no HiDPI (for which there > are no issues) if I understand correctly. > > I suppose that my real factor is between 2 and 2.5, so that truncating > gives the same result as rounding to nearest. Hence the same behavior. > > I don't know what broke for wm4. "With the patch applied, it scales > to 1, which is nonsense." is not very meaningful. This is what happens > for low DPI, for which no issues have been reported. > Maybe I wasn't clear enough, but I have a HiDPI screen, with a scale factor (as reported by X11) of approximately 1.9. With -12, it reduced the sizes of windows and GUI elements by half, and no combination of environment variables made it scale correctly. I suspect there's some other scaling factoring in somewhere, which somehow controls font sizes. Now I'm back to -10 (testing), and I won't upgrade it until the patch is reverted, so no new tests by me.