This regression has two parts: - etr gets window resize events it didn't before (this is cause by some change outside etr) - etr handling of the resize events seems buggy (previously latent bug activated by the changes outside etr) I'm testing a fix to upstream. -ML