No, what I'm saying is: if that's the outcome of switching on windowed mode on a 1920x1080 resolution with (I assume) 4x zoom, and you're not happy with that window size, couldn't you (under the old algorithm without the "-1") get the window size you want by setting the zoom to 3x?
The main issue with this is that the result I'm speaking of would happen at the
default zoom for this resolution / DPI. I think it's a bit iffy that the default settings at the most common resolution nowdays, when changing only fullscreen -> windowed, produce awkward results.
However - perhaps a hybrid approach can achieve a nice result for both of us; by keeping the existing algorithm (without the -1) but also factoring in a minimum border region size - ideally, one that would kick in for 5x @ 1920x1080, but not at 3x @ 1366x768. It could then be that instead of a static -1, the subtracted value is "whatever it needs to be to get that minimum border region" - which would be chosen so that at 3x @ 1366x768, the subtracted value is zero. I'll look into how feasible this is next time I'm working on NL. Would you mind dropping a screenshot of the default window size / position on your screen in a NL version you believe provides a desirable result, so I can use it for comparison?
I do think it's important to have nice results for 1366x768 too; this is still the
second most common resolution after all.
