Option to scale data before calculating covariance / PCA?

[Teebusch]

Should there be an option to scale the variables for which the covariance is calculated?
It may not be desired to scale the all variables that are used for the imputations, especially since not all of them may be numeric.