Set Statement, Retain Variables in PDV between observations

[PhilC]

Add an option /RETAIN to the SET statement which would command SAS not to clear the program data vector (PDV) prior to each dataset read. 

Example:

data Whole:
  set Left Right Center /Retain;
    BY v1;
run;

"Left" dataset:

v1   v4
A998 Q

"Right "dataset:

v1   v2
A998 06/01/18
B027 05/21/18
E001 03/31/18
F221 11/01/17
C379 10/07/17
G693 06/19/18

"Center" dataset:

v1   v3
A998 Z
A998 Y
A998 X
A998 W
B027 V
C379 U
D032 T

The result with the proposed /RETAIN option.  The bolded values are "retained" from the PDV unchanged from the prior observation:

v1   v4 v2       v3
A998 Q  .
A998 Q  06/01/18 
A998 Q  06/01/18 Z
A998 Q  06/01/18 Y
A998 Q  06/01/18 X
A998 Q  06/01/18 W
B027 Q  05/21/18 W
B027 Q  05/21/18 V
C379 Q  10/07/17 V
C379 Q  10/07/17 U
D032 Q  10/07/17 T
E001 Q  03/31/18 T
F221 Q  11/01/17 T
G693 Q  06/19/18 T

The result without the proposed /RETAIN option:

v1   v4 v2       v3
A998 Q  .
A998    06/01/18 
A998    .        Z
A998    .        Y
A998    .        X
A998    .        W
B027    05/21/18 
B027    .        V
C379    10/07/17 
C379    .        U
D032    .        T
E001    03/31/18 
F221    11/01/17 
G693    06/19/18 

[PhilC]

Another way to put it, all variables created by this SET statement using this proposed /RETAIN option would be treated as variables created within the data step and as if a RETAIN statement were applied to each variable.

[PhilC]

I would use this to perform date matching for data that needed to match data giving respect to the chronological ordering of timestamps.