Date:         Wed, 14 Feb 1996 18:15:05 -0800
Reply-To:     Karsten Self <karsten@NEWAGE1.STANFORD.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Karsten Self <karsten@NEWAGE1.STANFORD.EDU>
Subject:      Re: How to merge a data set?
Comments: To: John Whittington <johnw@MAG-NET.CO.UK>
In-Reply-To:  <199602150150.BAA10236@mag-net.co.uk>

I've been strenuously trying to avoid this debate, but...
 
Could it be that you are arguing the wrong side of the coin.
 
It isn't that variables from a given dataset are retained, it's that the
dataset vector (DSV) for a conditionally SET dataset is not *initialized*
for the data step loops in which it the dataset is not SET.
 
I posit the following:
 
 
----------------------------------------------------------------------------
data one;
    input a b c;
    cards;
1 11 21
2 12 22
3 13 23
4 14 24
5 15 25
6 16 26
7 17 27
8 18 28
9 19 29
;
run;
 
data two;
    input aa bb cc;
    cards;
101 111 121
102 112 122
103 113 123
;
 
data three;
    input x y z;
    cards;
201 202 203
;
run;
 
data four;
    input xx yy zz;
    datalines;
301 311 321
302 312 322
303 313 323
;
run;
 
data whoami;
    set one;
    if ((_n_ eq 1 or a = 8)) then
        set two;
 
    if "black" eq "white" then
        set three;
 
    if a eq 4 then
        set four;
 
    run;
 
proc print data= whoami;
run;
 
 
----------------------------------------------------------------------------
                              The SAS System                              3
                                         17:22 Wednesday, February 14, 1996
 
     OBS   A    B    C    AA    BB    CC   X   Y   Z    XX    YY    ZZ
 
      1    1   11   21   101   111   121   .   .   .     .     .     .
      2    2   12   22   101   111   121   .   .   .     .     .     .
      3    3   13   23   101   111   121   .   .   .     .     .     .
      4    4   14   24   101   111   121   .   .   .   301   311   321
      5    5   15   25   101   111   121   .   .   .   301   311   321
      6    6   16   26   101   111   121   .   .   .   301   311   321
      7    7   17   27   101   111   121   .   .   .   301   311   321
      8    8   18   28   102   112   122   .   .   .   301   311   321
      9    9   19   29   102   112   122   .   .   .   301   311   321
 
 
On Thu, 15 Feb 1996, John Whittington wrote:
 
> On Sun, 11 Feb 1996, rcw@TENET.EDU wrote:
>
> >In <199602101218.MAA17947@mag-net.co.uk>, John Whittington
> ><johnw@MAG-NET.CO.UK> writes:
 
> >You example works here because you have two different SET statements,  both of
> >which have their own Data Set Vector. It is the DSV which is reset on each
> >execution, if they are executed The second SET is not executed - consequently
> >it acts like a RETAIN.
 
> Richard, you have lost me here.  The number of SET statements has no bearing
> on whether or not any one of them 'acts like a RETAIN'.  All variables which
> appear in *any* SET statement in the DATA step are effectively RETAINed, as
> per the quote from TFM in your other posting.  That EVEN applies for a SET
> statement which is NEVER executed, for example:
>
> data three;
>      set one;
>      if 3=4 then set two;  /* never executed                          */
>      if _n_=1 then f=999;  /* f is one of the variables in datset two */
> run;
>
 
---------------------------------------------
Karsten M. Self -- Sr. SAS Programmer/Analyst
            Sierra Information Services, Inc.
  Contracting for NBER at Stanford University
 
  Karsten@newage1.Stanford.EDU
  KMSelf@ix.netcom.com
 
What part of gestalt don't you understand?