Date:         Thu, 17 Feb 2000 21:42:47 -0600
Reply-To:     shiling@math.wayne.edu
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Shiling Zhang <shiling@MATH.WAYNE.EDU>
Organization: Wayne State University
Subject:      Re: re-order variables
Comments: To: peter.crawford@DB.COM
Content-Type: text/plain; charset=us-ascii

Peter & _all_;

Retain statement is a saftest way to re-order a variable list if it is BEFORE a set
statement and all variables are from the data set. However, any attribute
statements(length, label, format, informat) can re-order a variable list too if they
are BEFORE a set statement. None of them initialize any values. But they may need
more information may have side effects. Unfortunately, the re-order function of
retain is never documented in SAS manual. I guess it is a by-product of retain
statement. Another difference is that any uninitialized variable in a retain
statement will not be created in PDV, and hence, an output data set. And also SAS
will issue a note. So retain is the best way in this context.

84   data t2;
85     set t1;
86     retain x y;
87   run;

NOTE: Variable Y is uninitialized.
NOTE: The data set WORK.T2 has 1 observations and 1 variables.
NOTE: The DATA statement used 0.16 seconds.


88   data t2;
89     length z 8;
90     set t1;
91     retain x y;
92   run;

NOTE: Variable Z is uninitialized.
NOTE: Variable Y is uninitialized.
NOTE: The data set WORK.T2 has 1 observations and 2 variables.
NOTE: The DATA statement used 0.16 seconds.
93
94   data t2;
95     format z best.;
96     set t1;
97     retain x y;
98   run;

NOTE: Variable Z is uninitialized.
NOTE: Variable Y is uninitialized.
NOTE: The data set WORK.T2 has 1 observations and 2 variables.
NOTE: The DATA statement used 0.22 seconds.

A Keep statement just marks variables in PDV like the other post pointed out, no
more no less. A Keep statement can be put anywhere in a data step and it will have
the same function.  When Keep statement refers to a variable that is not on PDV(not
internal variable), the variable  will never be created in a output dataset. SAS
will issue a warning message.

99   data t2;
100    format z best.;
101    set t1;
102    keep x y z;
103  run;

NOTE: Variable Z is uninitialized.
WARNING: The variable Y in the DROP, KEEP, or RENAME list has never been referenced.

NOTE: The data set WORK.T2 has 1 observations and 2 variables.
NOTE: The DATA statement used 0.16 seconds.


Peter Crawford wrote:

> perhaps it is because the RETAIN statement can assign initial values that the
> variables referred in this way can reserve their position/space in the data step
> data vector
> KEEP has no power to create or assign a value.
>
> Datum:         17.02.2000 20:39
> An:            SAS-L@listserv.uga.edu
>
> Antwort an:    taf2@is8.nyu.edu
>
> Betreff:       Re: re-order variables
> Nachrichtentext:
>
> To SAS-L:
>
> Could anyone explain why RETAIN changes the variable order in this
> situation, whereas KEEP does not?  In other words, the following code
> will change the variable order in the dataset:
>
> data test2;
> RETAIN mailname series;
> set test;
> run;
>
> but this code will not:
>
> data test2;
> KEEP mailname series;
> set test;
> run;
>
> Thanks!
>
>         --Tom
>         Tom Frenkel <taf2@is8.nyu.edu>
>         http://pages.nyu.edu/~taf2