https://bugs.kde.org/show_bug.cgi?id=373232
--- Comment #214 from Ruth Ivimey-Cook <r...@ivimey.org> --- I did some more work trying to understand the issue yesterday and as part of that created some debug prints to show what is happening. Note that the debug includes a macro "delay()" which was intended to slow things down enough to watch what happens. Change the delay amount or remove the call to delay() as appropriate. I had previously presumed the bug was an 'off-by-one' error but the debug code indicates that updates do happen on the appropriate pixel areas. I no longer think this explains it. However the lines do appear more reliably when the drawTextFragments() call isn't followed by anything else. I think, though at present this is a guess, that the problem is a rounding error. If internally the scaling code is rounding down rather than up, a fractional half or third of a pixel could be reduced to 0, and that when then scaled up that results in whole rows or columns of pixels not written. I say this is a guess because I haven't yet found the evidence for it. What is also obvious from the resulting trace is that there is a lot of apparently unnecessary repainting going on. Partly this is because ::paintEvent is being called with a damage rectangle equal to the terminal widget size, for reasons I cannot yet explain. It would be good to eliminate the excessive repainting because it partly masks the actual issues, and partly because it is just making everything slower. -- You are receiving this mail because: You are watching all bug changes.
