> Did anyone mention normalization-preserving/normalization-insensitive behavior?

You did, 11 times so far this thread. I counted.