Delete observations when all variabls are missing

It is a common question.

With the new function CMISS in 9.2, we can perform the task in ONLY one DATA step.
Please see the sample code at below:

data raw;
  output;  * obs with missing values;
  a=1; b='2'; c=.;
  output;
run;

data output;
  set raw;
  array aa {*} $ _character_;
  array nn {*} _numeric_;

  if cmiss(of _all_) = (dim(aa)+dim(nn)) then delete;
run;