Date:         Fri, 18 Feb 2000 07:14:18 +0000
Reply-To:     John Whittington <John.W@MEDISCIENCE.CO.UK>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         John Whittington <John.W@MEDISCIENCE.CO.UK>
Subject:      Re: re-order variables
Comments: To: Lary Jones <ljones@BINGHAMTON.EDU>
In-Reply-To:  <200002172151.VAA05940@vicar.netnames.net>
Content-Type: text/plain; charset="us-ascii"

At 16:50 17/02/00 -0500, Lary Jones wrote (in part):

>A. DATA step option             DATA Test2 (KEEP Mailname series);
>    This specifies what is included in the output of the data step.
>    Has no effect until an OUTPUT statement is encountered or is
>    implied by the end of the data step.

I agree.

>B. SET statement option SET Test (KEEP Mailname series);
>    This specifies which variables are brought into SAS (the PDV,
>    or data vector) for processing in the DATA step.  I believe that
>    this form *will* affect variable order.

I don't think so. The code:

data test ;
   a = 1 ; b = 2 ; c = 3 ; d = 4 ;
run ;
data try ;
   set test (keep = d c b) ;
run ;
proc print data = try ; run ;

... produces output:

OBS    B    C    D

 1     2    3    4

... with no change in variable order.


>C. An independent statement     KEEP mailname series;
>    This specifies which variables will be available in the statements
>    which follow the KEEP statement *after* all the variables have been
>    loaded in the PDV.

Again, if I understand what Lary is saying here, I disagree.  As far as I
am concerned, an 'independent KEEP statement' has the same effect as (A)
above - i.e. a KEEP dataset option on the output dataset.  A KEEP statement
certainly does not make ann 'non-KEEPed' variables unavialable to
subsequent statements in the DAT step, in the manner that Lary implied:

7551  data test ;
7552     a = 1 ; b = 2 ; c = 3 ; d = 4 ;
7553     keep a b ;
7554     e = c + d ;
7555     put e = ;
7556  run ;

E=7
NOTE: The data set WORK.TEST has 1 observations and 2 variables.
NOTE: The DATA statement used 0.16 seconds.

>Corrections and alternate viewpoints are welcome ;-)

See above!!  In essence, I don't believe that a KEEP statement can ever
influence variable ordering in the PDV.  Incidentally:

7597  data test ;
7598     a = 1 ; b = 2 ; c = 3 ; d = 4 ;
7599     keep a b c d e ;
7600  run ;

WARNING: The variable E in the DROP, KEEP, or RENAME list has never been
         referenced.
NOTE: The data set WORK.TEST has 1 observations and 4 variables.
NOTE: The DATA statement used 0.16 seconds.

... illustrating that KEEP does not create variables (only 4 variables in
the output dataset) - as others have pointed out, it couldn't, really,
since the compiler has no idea whther it is a numeric or character
variable, nor what its length is.

Kind Regards,


John

----------------------------------------------------------------
Dr John Whittington,       Voice:    +44 (0) 1296 730225
Mediscience Services       Fax:      +44 (0) 1296 738893
Twyford Manor, Twyford,    E-mail:   John.W@mediscience.co.uk
Buckingham  MK18 4EL, UK             mediscience@compuserve.com
----------------------------------------------------------------