On Wed Sep 28 09:17:47 2022 +0000, Zhiyi Zhang wrote:
Well, you never know what the window managers will do. Sometimes even some extra events will break things. Better keep it the same as the old one to avoid any possible regressions. Also, most of the time the virtual screen origin is not changed so it does matter performance-wise.
I understand but if things gets broken because we send a request that is supposedly doing nothing, I fear that it means things would be much more broken when windows actually need to be moved.
In any case, to keep that condition here we'd need to pass the information about whether the virtual rect changed origin. Maybe I can compute that from `X11DRV_resize_desktop`, from the current desktop position, but I think it makes things simpler if we don't have to.
Regarding regressions, this is already changing the logic quite a bit: previously I think only the windows from the process' changing display modes were moved. That doesn't seem very useful, so maybe it's better just not to move any window at all to keep the behavior closer to what it currently is?