Created attachment 82260 [details] Test case Example attached, we do not update a large enough area for the previous position. My bet is on a round() of the scaled coordinates instead of taking the biggest rect, but I'll see on debug :)
Cool effect!
I think it’s just that the focus ring width is unaffected by the 2D transform, but the invalidation math assumes that it is.
(In reply to comment #2) > I think it’s just that the focus ring width is unaffected by the 2D transform, but the invalidation math assumes that it is. Yep, you are right. Is that the intended behavior to not transform the focus ring? It seems a bit weird. Should I look for a solution or do you have the intention to change this behavior? :)
(In reply to comment #3) > (In reply to comment #2) > > I think it’s just that the focus ring width is unaffected by the 2D transform, but the invalidation math assumes that it is. > > Yep, you are right. > Is that the intended behavior to not transform the focus ring? It seems a bit weird. I don’t think this was intentional. > Should I look for a solution or do you have the intention to change this behavior? :) I think a good solution would be to have the focus ring respect the transform. I am not sure how easy or even feasible this is.
Reset the assignee. I wished I would have time to fix this but I looks like I won't in the near future.