Hi, Am in meeting all day, so don't have much time.  Call execute is a code generator.  Basically you need to think about what you want as the final code, then you piece the information together into a string and that string is then sent out as code *after* your datastep code finishes.  Therefore as Jaap has said, the code you generate if you just put it into a new file would not execute and give those errors.  You need to build the full code: data _null_;      call execute('data want; set have;'); /* send this line */      do I=1 to 10;           call execute('attrib avar'||strip(put(I,best.))||";"); /* Send a line with an attrib statement for each do loop */      end;      call execute('run;'); /* Finish the code */ run; This will generate the code: data want; set have; attrib avar1; attrib avar2; ... run; You can do a similar thing with macro loops, its basically different methods for the same thing.  Both can be useful. ... View more

I would just like to add to Jaap Karman's post that if you are using Oracle, I would assume that you have Toad, or SQL Developer or some SQL interface to Oracle installed.  What I would suggest is that you write the code there that you want to execute and test it.  Once it works then copy it over and paste it in between the brackets on the execute statement.  Then, assuming your connection information is correct that whole block should be sent to OC's SQL parser to run, much like when run through Toad, and hence should then return results.  TBH I have not seen that type of call either, generally speaking we would use SQL to extract data from a database, e.g. select * from table, and hence get some results back.  I have no idea what the function: st_pr_name, is actually doing, it could just be a comment for instance hence no results. ... View more

Sorry, could you provide some test data and required output data as I can't tell from the above what you are trying to do, do you mean something like: 700 1033935598 1033935589 TEL         01JAN2001 E Daytime Only     001 700 1033935598 1033935589 TEL         01JAN2001 E Anytime               002 Then that's straight forward (assuming sorted): data want;      set have;      by var1 var2 var3;      retain count;      if first.var1 then count=1;      else count=count+1;      newvar=put(count,z3.); run; ... View more

Wasn't entirely sure what the segments were here - how does an age fit into more than one segment?  Anywas, by datastep: data indat;      input age; datalines; 34 42 30 ; run; data _null_;   array segment{4,2} $200. (  "0",  "28",                               "29", "32",                               "33", "41",                               "42", "50");   call execute('data want; set indat;');   do i=1 to 4;     call execute(' attrib ageseg_'||segment{i,2}||' format=best.;                    if '||segment{i,1}||' <= age < '||segment{i,2}||' then ageseg_'||segment{i,2}||'=1; else ageseg_'||segment{i,2}||'=0; output;');   end; run; ... View more