Thank you Tom, and you are right. It should be modified as needed.   Actually the original macro has more parameters being one the logical final statement (&LOGIC), which could be:   AND 1 OR 0   or whatever is required to not affect the whole statement logic.   The operator between pairs is obtained from this also (%scan(&LOGIC,1))   Daniel Santos @ www.cgd.pt ... View more

Found a convenient solution, still for the sake of diversity...   My own approach (which is a modified version from my personal macro functions library).   %macro join(ALIAS1,ALIAS2,VARLIST,VARNUM); %let varname=%scan(&VARLIST,&VARNUM,%str( )); %if %str(&varname) ne %str() %then &ALIAS1..&varname = &ALIAS2..&varname and %join(&ALIAS1,&ALIAS2,&VARLIST,%eval(&VARNUM+1)); %else 1 %mend join;   Used like this:   options mprint; proc sql noprint; create table mydata as select ... on %join(a,b,cat_order grp_order year,1); quit;   Will recursively build the join clause appending "and 1" at the end (doesn't affect the AND logic)   MPRINT(JOIN): a.cat_order = b.cat_order and MPRINT(JOIN): a.grp_order = b.grp_order and MPRINT(JOIN): a.year = b.year and MPRINT(JOIN): 1   ALIAS1=left alias,ALIAS2=right alias,VARLIST=columns list,VARNUM=first column to start     Daniel Santos @ www.cgd.pt ... View more

Check it yourself:   %let x=%substr("ABCD",2,1); %put &X; Daniel Santos @ www.cgd.pt ... View more

Hi.   Why not using PROC APPEND?   %macro import_loop; proc datasets lib=WORK nolist; delete M_ALL; quit; * delete first; %let outobs=max; %do i1=0 %to 24; %let i = %sysfunc( putn(&i1,z2.)); proc import out = testv1_&i datafile = '/TESTING/DVR/DVR_V1_PROD_01182017.xls' dbms = xls ; sheet ="M&i._ACCT_BAL_AM"; getnames = yes; run; proc sql outobs=&outobs; create table M&i._ACCT_BALANCE_AM as select "M_ACCT_BAL_AM_&i." as M_ACCT_BAL_AM_&i. ,a.* FROM testv1_&i a ; proc append base=M_ALL data=M&i._ACCT_BALANCE_AM force; run; * append; %end; %mend import_loop; %import_loop;     Daniel Santos @ www.cgd.pt   ... View more

Oh, I'm surely not a genius, but thanks for the compliment.   And no problem with that. If you want to set only CHR_PER_STR and have NUM_OF_STR calculated dynamically, instead of:   %let NUM_OF_STR=5; %let CHR_PER_STR=40;   use:   %let CHR_PER_STR=200; data _null_; set HAVE; call symputn('NUM_OF_STR',(lengthc(S)+&CHR_PER_STR-1)/&CHR_PER_STR); run;   This will set NUM_OF_STR according to S's size and the specified CHR_PER_STR.   Daniel Santos @ www.cgd.pt   ... View more

  When it comes to performance of merging data it cannot be reduced just to the sequence of joins.   If that's just what you are after, then the answer is pretty much: set the fullstimer option, run both several time and see what's best.   If you really want to go deep in this, then why not consolidate all the data inside the database?   Data is stored and retrieved in a much more efficient way inside the Database.   If for some reason that's not possible, then please describe: Each join type and keys, size (obs, record lenght) of each dataset.   Daniel Santos @ www.cgd.pt ... View more

Hi.   The following may work for you.   %let NUM_OF_STR=5; %let CHR_PER_STR=40; data _null_; array V {*} $ &CHR_PER_STR V1-V&NUM_OF_STR; S='the demonetisation exercise has been called a watershed for a country saddled with counterfeiters pushing'; _I=1; _J=1; do until (scan(S,_I,' ') eq ''); _W=scan(S,_I,' '); if lengthn(V[_J])+lengthn(_W) > lengthc(V[_J]) then _J+1; V[_J]=strip(V[_J])!!' '!!_W; _I+1; end; run;   Nothing fancy, it will split the text word by word and ditribute it by the number of desired variables.   This will also remove consecutive blanks between words if any.   Daniel Santos @ www.cgd.pt ... View more