Date:         Fri, 15 Jan 1999 10:28:06 -0500
Reply-To:     "John T. Jones" <jonesj@PHARMARESEARCH.COM>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         "John T. Jones" <jonesj@PHARMARESEARCH.COM>
Subject:      Re: Reordering variables in a new data set
Content-Type: text/plain; charset="us-ascii"

Thank you John,
 
 
I have had problems using this method before though.
Do you know what that might be?  I wanted to be careful
when I suggested the retain staement.
 
jj
 
 
 
At 03:07 PM 1/15/99 +0000, you wrote:
>At 09:15 15/01/99 -0500, John T. Jones wrote:
>
>>2.  data new;
>>     retain var1 var2;
>>     set old;
>>    run;
>>    Caution: retain will hold last not missing
>>             value over iterations of missing.
>>             If have no missing then this is simplest.
>
>John, the 'caution' is unnecessary - and, in fact, not true.  The previous
>value of RETAINed variables is only RETAINed until a new value (*even*
>'missing') is assigned - e.g. by a SET statement.  Despite all the myths and
>concerns, there are NO situations in which the code you have written will
>not work as intended.  The only thing you could do wrong would be to include
>a variable in the RETAIN statement that did not exist in the dataset being
>SET - but that would merely result in that 'new' variable being added to the
>dataset with all values set to missing.
>
>Regards,
>
>John
>
>----------------------------------------------------------------
>Dr John Whittington,        Voice:    +44 (0) 1296 730225
>Mediscience Services        Fax:      +44 (0) 1296 738893
>Twyford Manor, Twyford,     E-mail:   medisci@powernet.com
>Buckingham  MK18 4EL, UK              mediscience@compuserve.com
>----------------------------------------------------------------
>