Date:         Thu, 2 Jan 1997 17:48:32 EDT
Reply-To:     hermans1@WESTATPO.WESTAT.COM
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Sig Hermansen <hermans1@WESTATPO.WESTAT.COM>
Subject:      Help with Macro ... sort order of key variables

     You may have to rethink the design of the program.  Even if you
     correct all of the syntax and runtime errors, it may prove too
     complicated to explain how and why it works.  If you define each major
     step in a Macro and focus on the contents of the datasets or
     macrovariables read and produced by each step, you will reduce the
     main program to one page.  You can then examine each step to make sure
     that the incoming and outgoing variables have the correct definition
     and that rows (obs, records) have the correct order for merges or
     summaries.
 
     So far as I can tell, the "not properly sorted" runtime error occurs
     because the samestep Macro orders the dataset plan differently than
     specified in the BY statement of the MERGE of plan, county, and state.
     The "variable ... is uninitialized" warnings arise because you named
     macrovariables in a LENGTH statement.  SAS creates columns in the
     output dataset for these supposed data step variables, but they remain
     missing.  SAS does not confuse macrovariable names with data step
     variable names (as I for have at times!).
 
     PROC SQL does not require that you order the data to do the
     equivalents of MERGES and summaries.  For that reason and others I
     almost always use SQL joins instead of datastep MERGES to link
     datasets on key values.
 
     --------------- Excerpt from original message -----------------------
 
     <Date:    Thu, 2 Jan 1997 14:36:13 EST
     <From:    Suhua Hu <sxh04@HEALTH.STATE.NY.US>
     <Subject: Need Help with My Macro Program
     ...........................................................
      <625   %macro samestep (outname=state, inname=elig, type1=0, type2=2,
      <626                         type3=3, byvar=);
      ..........................................................
      <634          proc sort;
      <635            by &byvar sex agecat;
      <636          run;
      <637        %mend samestep;
 
      ...............................................................
      <714    data final (drop=ctycode _type_);
      <715       merge plan county (rename=(qn=c7 yn=c8 qmm=c9 ymm=c10))
      <716       state (rename=(qn=c11 yn=c12 qmm=c13 ymm=c14));
      <717       by agecat sex;
      <718
      ..........................................
 
      <NOTE: VARIABLE C_ID IS UNINITIALIZED.
      <NOTE: VARIABLE C_NAME IS UNINITIALIZED.
      <NOTE: VARIABLE P_NAME IS UNINITIALIZED.
 
      <ERROR: BY VARIABLES ARE NOT PROPERLY SORTED ON DATA SET WORK.PLAN.
      <ERROR: BY VARIABLES ARE NOT PROPERLY SORTED ON DATA SET WORK.PLAN.
      <ERROR: BY VARIABLES ARE NOT PROPERLY SORTED ON DATA SET WORK.PLAN.