Date:         Fri, 28 Mar 1997 07:32:15 GMT
Reply-To:     David Michael Wright <david@CATS.UCSC.EDU>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         David Michael Wright <david@CATS.UCSC.EDU>
Organization: University of California, Santa Cruz
Subject:      Re: A BETTER SAS, part 1

In article <3332FD08.6D6B@sbphrd.com>,
Steven_Hanks-1  <Steven_Hanks-1@sbphrd.com> wrote:
|David Michael Wright wrote (And Steve Annotated)
|
|> SAS should elliminate all proceedures and replace them with functions.
|> Then you wouldn't have the bizarre macro language that is nearly
|> impossible to remember and learn, or a separte matrix language with
|> it's convoluted syntax and limted functionality (not to mention
|> expense)
|
|OK, I'll agree that macro language looks strange. But 'bizarre' and
|'impossible to remember'?....
 
A simple tour through the days' post brings up couple of different answers
on how to assign records to macros: - very simple to do in common
languages with arrays:
 
 
On Tue, 25 Mar 1997 07:03:13 +0000, Roland <RolandRB@nospam.netcomuk.co.uk>
wrote:
 
data _null_;
  set whatever;
  call execute('%top('||first||','||last||')');
run;
 
I tried something like this once, but got dizzy looking at the
pipe and apostrophes and it did not work.
 
 
Scott Came, Sr Programmer/Analyst wrote:
 
data obs;
   input first last;
   cards;
1 3
4 4
5 6
7 9
;
run;
 
%macro top (first, last);
   %* a trivial TOP macro--yours will do real stuff in here. *;
   %put RUNNING TOP:  first=&first last=&last;
%mend top;
 
%macro doit;
 
   options nonotes; %* optional--makes log look nicer *;
 
   %local ptr nobs i first last;
 
   %* read number of obs in dataset OBS into a macro var. *;
 
   proc contents data=obs
                 noprint out=conout(keep=nobs);
   run;
 
   data _null_;
      set conout;
      call symput("nobs",compress(nobs));
      stop;
   run;
 
  %* loop thru the observations in OBS, feeding FIRST and LAST into TOP
*;
 
   %do i = 1 %to &nobs;
 
      data _null_;
         set obs(firstobs=&i);
         call symput("first",compress(first));
         call symput("last",compress(last));
         stop;
      run;
 
      %top(&first,&last)
 
   %end;
 
   %* clean up *;
 
   proc datasets library=work nolist;
      delete nobs;
   run;
 
   options notes;  %* again, optional *;
 
%mend doit;  %doit
 
I rest my case!